Patent Application
20060168334
The present invention generally relates to network elements in computer networks. The invention relates more specifically to a method and apparatus for network element management of server failures.
The approaches described in this section could be pursued, but are not necessarily approaches that have been previously conceived or pursued. Therefore, unless otherwise indicated herein, the approaches described in this section are not prior art to the claims in this application and are not admitted to be prior art by inclusion in this section.
In a business-to-business environment, applications executing on computers commonly communicate with other applications that execute on other computers. For example, an application “A” executing on a computer “X” might send, to an application “B” executing on a computer “Y,” a message that indicates the substance of a purchase order.
Computer “X” might be remote from computer “Y.” In order for computer “X” to send the message to computer “Y,” computer “X” might send the message through a computer network such as a local area network (LAN), a wide-area network (WAN), or an inter-network such as the Internet. In order to transmit the message through such a network, computer “X” might use a suite of communication protocols. For example, computer “X” might use a network layer protocol such as Internet Protocol (IP) in conjunction with a transport layer protocol such as Transport Control Protocol (TCP) to transmit the message.
Assuming that the message is transmitted using TCP, the message is encapsulated into one or more data packets; separate portions of the same message may be sent in separate packets. Continuing the above example, computer “X” sends the data packets through the network toward computer “Y.” One or more network elements intermediate to computer “X” and computer “Y” may receive the packets, determine a next “hop” for the packets, and send the packets towards computer “Y.”
For example, a router “U” might receive the packets from computer “X” and determine, based on the packets being destined for computer “Y,” that the packets should be forwarded to another router “V” (the next “hop” on the route). Router “V” might receive the packets from router “U” and send the packets on to computer “Y.” At computer “Y,” the contents of the packets may be extracted and reassembled to form the original message, which may be provided to application “B.” Applications “A” and “B” may remain oblivious to the fact that the packets were routed through routers “U” and “V.” Indeed, separate packets may take different routes through the network.
A message may be transmitted using any of several application layer protocols in conjunction with the network layer and transport layer protocols discussed above. For example, application “A” may specify that computer “X” is to send a message using Hypertext Transfer Protocol (HTTP). Accordingly, computer “X” may add HTTP-specific headers to the front of the message before encapsulating the message into TCP packets as described above. If application “B” is configured to receive messages according to HTTP, then computer “Y” may use the HTTP-specific headers to handle the message.
In addition to all of the above, a message may be structured according to any of several message formats. A message format generally indicates the structure of a message. For example, if a purchase order comprises an address and a delivery date, the address and delivery date may be distinguished from each other within the message using message format-specific mechanisms. For example, application “A” may indicate the structure of a purchase order using Extensible Markup Language (XML). Using XML as the message format, the address might be enclosed within “<address>” and “</address>” tags, and the delivery date might be enclosed within “<delivery-date>” and “</delivery-date>” tags. If application “B” is configured to interpret messages in XML, then application “B” may use the tags in order to determine which part of the message contains the address and which part of the message contains the delivery date.
A web browser (“client”) might access content that is stored on remote server by sending a request to the remote server's Universal Resource Locator (URL) and receiving the content in response. Web sites associated with very popular URLs receive an extremely large volume of such requests from separate clients. In order to handle such a large volume of requests, these web sites sometimes make use of a proxy device that initially receives requests and distributes the requests, according to some scheme, among multiple servers.
One such scheme attempts to distribute requests relatively evenly among servers that are connected to the proxy device. A proxy device employing this scheme is commonly called a “load balancer.” When successful, a load balancer helps to ensure that no single server in a server “farm” becomes inundated with requests.
When a proxy device receives a request from a client, the proxy device determines to which server, of many servers, the request should be directed. For example, a request might be associated with a session that is associated with a particular server. In that case, the proxy device might need to send the request to the particular server with which the session is associated.
If the server to which the proxy device sent the request is not able to service the request, one of several scenarios may occur. In one scenario, the server might send no response whatsoever. Under this scenario, after a specified amount of time has passed since the client sent the request without receiving a corresponding response, the client may determine that a “timeout” event has occurred. The client may take a specified action that is associated with the timeout event, such as notifying a user that a response to the request could not be obtained.
In another scenario, the server might send an HTTP-specific response that indicates that the server is not able to service the request. For example, the server might send a “500” code in an HTTP header. The client may receive the HTTP-specific response and take a specified action that is associated with the HTTP-specific response, such as notifying a user that the request could not be serviced.
Under either scenario, the only recourse left to the client is to resend the request. However, when the client resends the request, the resending wastes both network bandwidth and the client's processing resources. Furthermore, although HTTP provides codes whereby a server can notify a client, in a protocol header, that the server is unable to service a request, sometimes clients and servers communicate using protocols other than HTTP. Some of these other protocols do not have such built-in notification mechanisms.
A less wasteful, more productive, and more widely applicable technique for managing server failure, or the inability of a server to service a request, is needed.
The present invention is illustrated by way of example, and not by way of limitation, in the figures of the accompanying drawings and in which like reference numerals refer to similar elements and in which:
FIGS. 3B-C depict a flow diagram that illustrates one embodiment of a method of balancing data traffic among multiple servers based on application layer message content;
A method and apparatus for network element-managed server failover using application layer messages is described. In the following description, for the purposes of explanation, numerous specific details are set forth in order to provide a thorough understanding of the present invention. It will be apparent, however, to one skilled in the art that the present invention may be practiced without these specific details. In other instances, well-known structures and devices are shown in block diagram form in order to avoid unnecessarily obscuring the present invention.
Embodiments are described herein according to the following outline:
1.0 General Overview
2.0 Structural and Functional Overview
3.0 Implementation Examples
4.0 Implementation Mechanisms—Hardware Overview
5.0 Extensions and Alternatives
1.0 General Overview
The needs identified in the foregoing Background, and other needs and objects that will become apparent for the following description, are achieved in the present invention, which comprises, in one aspect, a method for network element-managed server failover based on application layer messages. According to one aspect, a network element such as a router intercepts a request from a client and selects a first server toward which the network element then sends the request. In response to detecting that the first server is unable to service the request, the network element selects a second server toward which the network element then sends the request. Thus, the network element assumes the responsibility of ensuring that a currently capable server will handle the client's request. The client does not even need to be aware that the network element is doing so.
According to one aspect, the network element may detect a server's inability by inspecting the contents of an application layer message, sent from the server, for an indication from the server itself that the server is unable to service requests. Thus, a server may indicate its inability to service a request even when the server uses protocols that do not provide built-in mechanisms for indicating such inability.
In other aspects, the invention encompasses a computer apparatus and a computer-readable medium configured to carry out the foregoing steps.
2.0 Structural and Functional Overview
Client application 110 is coupled communicatively with network element 102. Server applications 112A-N are coupled communicatively to network element 106. Server applications 114A-N are coupled communicatively to network element 108. Client application 110 and server applications 112A-N and 114A-N may be separate processes executing on separate computers. According to one embodiment, server applications 112A-N and 114A-N are web servers. According to one embodiment, server applications 112A-N and 114A-N are database servers that communicate with databases.
Network elements 102 and 104 are coupled communicatively with a network 116. Network elements 104 and 106 are coupled communicatively with a network 118. Network elements 104 and 108 are coupled communicatively with a network 120. Each of networks 116, 118, and 120 is a computer network, such as, for example, a local area network (LAN), wide area network (WAN), or internetwork such as the Internet. Networks 116, 118, and 120 may contain additional network elements such as routers.
Client application 110 encapsulates application layer messages within data packets and addresses the data packets to virtual addresses, such as virtual IP addresses, each of which may be associated with multiple servers. For example, a first virtual IP address may be associated with server applications 112A-N, and a second virtual IP address may be associated with server applications 114A-N. Network elements that intercept data packets destined for the first virtual IP address route the data packets toward network element 106. Network elements that intercept data packets destined for the second virtual IP address route the data packets toward network element 108.
Network elements 106 and 108 intercept the data packets that contain the messages. Network elements 106 and 108 select, from among server applications 112A-N and server applications 114A-N, respectively, particular server applications toward which the data packets should be sent. The particular server applications may be selected based on a load-balancing algorithm, or based on session mappings, or based on relative priorities of the server applications.
For example, each of server applications 112A-N may be assigned a priority relative to each other of server applications 112A-N. If server application 112A has a higher priority than server application 112B, then network element 106 may attempt to send data packets to server application 112A before sending data packets to server application 112B; network element 106 might send data packets to server application 112B only if server application 112A is unable to service a request contained with the data packets.
In one embodiment, network elements 106 and 108 store intercepted requests and determine whether the server applications to which requests are sent in this manner are unable to service the requests. In response to determining that a particular server application is unable to service a request, network elements 106 and 108 select another server application and send the request to that other server application. In one embodiment, network elements 106 and 108 repeat this server application selection process until network elements 106 and 108 intercept, from a selected server, a response that does not indicate that the request could not be serviced. Network elements 106 and 108 send such “actual” responses to client application 110.
Because network elements 106 and 108 ensure that requests are distributed to server applications that are capable of responding to those requests, client application 110 does not need to be sent any responses that indicate that those requests could not be serviced. Thus, network bandwidth may be reserved for transporting “actual” responses, and client application 110 does not even need to be “aware” that network elements 106 and 108 are handling server failover. To client application 110, it does not appear that any of server applications 112A-N and 114A-N have failed.
In block 202, the inability of a first server to service a request is detected at a network element. For example, network element 106 may intercept a request from client application 110, store a copy of the request, select server application 112A from among server applications 112A-N, send the request to server application 112A, and detect that server application 112A is unable to service the request. Network element 106 may make this determination in any one of several ways.
For example, network element 106 might determine that a specified amount of time has passed since network element 106 sent the request to server application 112A, and that in that time, no corresponding response has been received on the TCP connection to server application 112A. For another example, network element 106 may determine that a specified amount of time has passed since network element 106 sent a TCP SYN packet to server application 112A, and that in that time no TCP SYN/ACK packet has been received from server application 112A. As a result of detecting such timeout events, network element 106 may determine that server application 112A has failed and is unable to service the request.
For another example, network element 106 might intercept, from server application 112A, an HTTP-specific message that indicates that server application 112A is unable to service the request. More specifically, server application 112A might send, toward client application 110, a message that has an HTTP header that indicates status code “500.” Network element 106 may intercept the data packets that contain this message, assemble and inspect the message, and determine from the code that server application 112A is unable to service the request.
For another example, network element 106 might intercept a message that indicates, in the body of the message itself, that server application 112A is unable to service the request. More specifically, server application 112A might send, toward client application 110, an XML document that contains a specified XML element that indicates that server application 112A cannot service the request. One or more protocol headers, such as an FTP header, a Simple Mail Transfer Protocol (SMTP) header, or some other protocol header, may precede the XML document within the collective payload portions of multiple TCP data packets. A protocol header may relate to a proprietary protocol. Network element 106 may intercept the packets that contain the XML document, assemble and inspect the XML document, and determine, from the presence of the specified XML element within the XML document, that server application 112A is unable to service the request.
In one embodiment, when a network element such as network element 106 determines that a server application such as server application 112A is unable to service requests, the network element generates an entry for the server application in a list of server applications that the network element will not select, at least for a specified period of time. The entry may be associated with a timestamp that indicates to the network element when the entry should be removed from the list.
In block 204, in response to detecting the inability of the first server to service the request, a second server is selected from among a plurality of servers. For example, in response to detecting that server application 112A is unable to service the request, network element 106 may select server application 112B from among server applications 112A-N. The selection may be based on a load-balancing algorithm, for example, or based on relative priorities assigned to server applications 112A-N.
In block 206, the request is sent toward the second server. Thus, the first server “fails over” to the second server. For example, using the copy of the request that network element 106 stored earlier, network element 106 may send the request to server application 112B. If network element 106 intercepts, from server application 112B, an actual response to the request, then network element 106 may send the response toward client application 110. Client application 110 does not need to be made aware that server application 112A was unable to service the request.
2.1 Maintaining Session State at Network Elements
According to one embodiment, network elements 102, 104, 106, and 108 inspect the contents of communications that those network elements intercept. The network elements determine, from the contents, both session-identifying information and session state information. Using the session-identifying information to identify the sessions, the network elements store the session state information and associate the session state information with the session to which the session state information pertains.
Because the network elements store session state information, server applications 112A-N and server applications 114A-N do not need to store session state information. For example, when network element 106 intercepts a request from client application 110, network element 106 may determine state information, such as the identities of items in a shopping cart, based on the request. Network element 106 may generate a new request for only those information items that network element 106 needs in order to generate a response to client 110, and send the new request to one of server applications 112A-N. The server application receiving the request may respond to network element 106 with the requested information items. Receiving the response, network element 106 may generate a new response using the state information stored at network element 106 and the information contained in the response from the server application. After generating the new response, network element 106 may send the new response toward client application 110.
Because all, or at least some, session state information may be stored at the network elements in addition to or instead of the server applications, the consequences of server application failure can be managed. When a server application fails, the session state information for session that the server application was handling is not lost, since the session state information is stored on a network element. Thus, when a server application fails, recovery may be achieved by the network element directing requests to a different server application. This may all occur without the knowledge of client application 110.
Furthermore, server applications do not need to synchronize session states among themselves. As a result, the overheard associated with inter-server-application communications that would be needed to accomplish such synchronization may be avoided.
As is discussed above, in one embodiment, server applications 112A-N and 114A-N are database servers. Client application 110 may communicate with these database servers using a proprietary database protocol. If a client is a database application using JDBC or ODBC, then the network elements may behave like JDBC drivers that interpret the database protocol. For example, client application 110 may send SQL statements to the database servers. According to one embodiment, network elements 102, 104, 106, and 108 inspect communications between client application 110 and the database server for database connection information. Database connection information may include parameters that are negotiated in order to establish a communications path between the client application and the database server. Obtaining such database connection information, the network elements store the database connection information locally as database connection state. If a database connection fails, or if a database server using such a connection fails, then the network element can reconstruct the database connection with the same or a different database server using the stored database connection state. All of this may be performed without the knowledge of client application 110. Because the database connection state information stored at the network element is used to reconstruct a database connection, the client application 110 does not need to intervene or have any special logic to handle the reconnection. The failure is being managed without any disruption to client application 110.
Among the information items which may be stored at a network element as part of the database connection state, and used to reconstruct a database connection, are the following: The URL for the connection, whether auto commit mode is “true” or “false,” whether read only mode is “true” or “false,” prepared statements that have been precompiled for optimization, and callable statements that have been precompiled for optimization.
In one embodiment, the network elements “virtualize” database connections to database servers. In other words, client application 110 might indicate a particular database object (e.q., in an SQL query) that the client application wants to access, but client application 110 might not indicate any particular database server to which the client application's request should be directed. Intercepting the request, network element 106, for example, may select a database server from among servers 112A-N, and direct the client application's request to the selected database server. If the database object is only accessible through a particular subset of database servers, then client application 110 may select one of the database servers within the particular subset.
When a database connection fails, network elements may reconstruct database connections immediately, in response to the failure. Alternatively, network elements may wait for subsequent requests from client applications that were using the failed database connections before reconstructing database connections for those client applications. Alternatively, each network element may maintain a “pool” of database connections to each database server, and allocate/modify a currently unused database connection in the pool whenever a database connection fails.
Whether the state information stored at the network server is session state information or database connection state information, the session identifier determined from messages sent from client application 110 may be determined from an express indication of the session identifier within the message, or by implication from other information associated with the messages. For example, network element 106 may look into the HTTP header of an HTTP request and inspect a cookie that expressly indicates a session identifier. For another example, network element 106 may look into the HTTP body of an HTTP request and locate a string that says “sessionid=123,” where “123” is the session identifier. For another example, network element 106 look into the content of a database transport protocol and locate the binary data that identifies the database connection the request is associated with. For yet another example, network element 106 may determine an IP address from which the message came, and determine that the IP address is associated with a particular session identifier. Such associations may be maintained within the network elements.
According to one embodiment, each network element comprises multiple “blades.” Each blade may store session state information. Two or more blades may synchronize session state with each other so that if one blade fails, the other blades have the session state that was maintained by the failed blade, and operations may continue without any interruption noticeable by client application 110. An administrator may specify which blades synchronize with each other.
Session state information may be obtained from a sequence of packets by appending the packets together at a network element, and inspecting the contents of the payload sections of the packets. Such inspection may involve, for example, inspecting the body of an HTTP message that is collectively contained in the payload sections. The session state information may be contained in the body. Techniques described herein may be applied to protocols other than HTTP, however; the techniques described herein should not be read as being limited to a particular protocol.
3.0 Implementation Examples
3.1 Network Element-Managed Server Failover Based on Application Layer Messages
In block 302, a request, which originated from a client, is intercepted at a network element. The request is not addressed to the network element. For example, network element 106 may intercept a request that client application 110 addressed to a virtual IP address associated with server applications 112A-N.
In block 304, a server is selected from among a plurality of servers that excludes servers that are on a list of failed servers. For example, assuming that server application 112A is not on a list of failed servers maintained by network element 106, network element 106 may select, from among server applications 112A-N, server application 112A.
In block 306, the request is sent toward the selected server. A copy of the request is also stored at the network element. For example, network element 106 may send the request to server application 112A.
In block 308, data packets sent from the selected server are intercepted at the network element. For example, network element 106 may intercept multiple TCP data packets that server application 112A addressed to client application 110. Payload portions of the data packets may collectively contain an application layer message that server application 112A generated in response to determining that server application 112A was unable to service the request. Because each of the data packets may contain a separate portion of the message, under some circumstances, none of the data packets independently contains the entire message. The application layer message may contain an indication that server application 112A is unable to service the request. For example, the message may be an XML document that contains a specified XML element that indicates that server application 112A is unable to service the request.
The message may be carried according to any of a variety of protocols, including HTTP, SMTP, and FTP. Thus, within the payload portions of the data packets, protocol headers used by such protocols may precede the message. The protocol headers are separate from the message itself.
In block 310, an application layer message collectively contained in the data packets is constructed at the network element. For example, network element 106 may assemble the contents of the payload portions of the data packets intercepted from server application 112A. Network element 106 may disregard any protocol headers, such as HTTP, FTP, or SMTP headers, which precede the application layer message constructed from the contents of the payload portions.
In block 312, it is determined, at the network element, whether the application layer message contains an indication that the selected server is unable to service the request. For example, assuming that the message is an XML document, network element 106 may determine whether the XML document contains a specified XML element. If the XML document does contain the specified XML element, then network element 106 may conclude that server application 112A is unable to service the request. If the application layer message contains the indication, then control passes to block 314. Otherwise, control passes to block 316.
In block 314, the selected server is added to the list of failed servers. For example, network element 106 may add, to the list of failed servers, an entry that identifies server application 112A. The entry may be associated with a timestamp that indicates a time at which the entry will be removed, automatically, from the list of failed servers. Control passes back to block 304, in which another server is selected to receive the request, which was previously stored at the network element. For example, network element 106 may select server application 112B and send the request to server application 112B.
Alternatively, in block 316, the application layer message is sent toward the client. For example, network element 106 may add the protocol headers back to the application layer message, encapsulate the application layer message and protocol headers into one or more data packets, and send the data packets toward client application 110.
3.2 Multi-Blade Architecture
According to one embodiment, an Application-Oriented Network Services (AONS) blade in a router performs the actions discussed above.
Router 600 receives packet flows such as packet flows 610A-B. More specifically, packet flows 610A-B received by router 600 are received by supervisor blade 602. Supervisor blade 602 may comprise a forwarding engine and/or a route processor such as those commercially available from Cisco Systems, Inc.
In one embodiment, supervisor blade 602 classifies packet flows 610A-B based on one or more parameters contained in the packet headers of those packet flows. If the parameters contained in the packet header of a particular packet match specified parameters, then supervisor blade 602 sends the packets to a specified one of AONS blade 604 and/or other blades 606A-N. Alternatively, if the parameters contained in the packet header do not match any specified parameters, then supervisor blade 602 performs routing functions relative to the particular packet and forwards the particular packet on toward the particular packet's destination.
For example, supervisor blade 602 may determine that packet headers in packet flow 610B match specified parameters. Consequently, supervisor blade 602 may send packets in packet flow 610B to AONS blade 604. Supervisor blade 602 may receive packets back from AONS blade 604 and/or other blades 606A-N and send the packets on to the next hop in a network path that leads to those packets' destination. For another example, supervisor blade 602 may determine that packet headers in packet flow 610A do not match any specified parameters. Consequently, without sending any packets in packet flow 610A to AONS blade 604 or other blades 606A-N, supervisor blade 602 may send packets in packet flow 610A on to the next hop in a network path that leads to those packets' destination.
AONS blade 604 and other blades 606A-N receive packets from supervisor blade 602, perform operations relative to the packets, and return the packets to supervisor blade 602. Supervisor blade 602 may send packets to and receive packets from multiple blades before sending those packets out of router 600. For example, supervisor blade 602 may send a particular group of packets to other blade 606A. Other blade 606A may perform firewall functions relative to the packets and send the packets back to supervisor blade 602. Supervisor blade 602 may receive the packet from other blade 606A and send the packets to AONS blade 604. AONS blade 604 may perform one or more message payload-based operations relative to the packets and send the packets back to supervisor blade 602.
According to one embodiment, the following events occur at an AONS router such as router 600. First, packets, containing messages from clients to servers, are received. Next, access control list-based filtering is performed on the packets and some of the packets are sent to an AONS blade or module. Next, TCP termination is performed on the packets. Next, Secure Sockets Layer (SSL) termination is performed on the packets if necessary. Next, Universal Resource Locator (URL)-based filtering is performed on the packets. Next, message header-based and message content-based filtering is performed on the packets. Next, the messages contained in the packets are classified into AONS message types. Next, a policy flow that corresponds to the AONS message type is selected. Next, the selected policy flow is executed. Then the packets are either forwarded, redirected, dropped, copied, or fanned-out as specified by the selected policy flow.
FIGS. 3B-C depict a flow diagram 300B that illustrates one embodiment of a method of balancing data traffic among multiple servers based on application layer message content. For example, one or more of network elements 102, 104, 106, and 108 may perform such a method. More specifically, AONS blade 604 may perform one or more steps of such a method. Other embodiments may omit one or more of the operations depicted in flow diagram 300B. Other embodiments may contain operations additional to the operation depicted in flow diagram 300B. Other embodiments may perform the operations depicted in flow diagram 300B in an order that differs from the order depicted in flow diagram 300B.
Referring first to
In block 324, an association is established, at the network element, between the particular message classification and the one or more criteria. For example, AONS blade 604 may establish an association between a particular message classification and one or more criteria. For example, the criteria may indicate a particular string of text that a message needs to contain in order for the message to belong to the associated message classification. For another example, the criteria may indicate a particular path that needs to exist in the hierarchical structure of an XML-formatted message in order for the message to belong to the associated message classification. For another example, the criteria may indicate one or more source IP addresses and/or destination IP addresses from or to which a message needs to be addressed in order for the message to belong to the associated message classification.
In block 326, an association is established, at the network element, between the particular message classification and the one or more actions. One or more actions that are associated with a particular message classification comprise a “policy” that is associated with that particular message classification. A policy may comprise a “flow” of one or more actions that are ordered according to a particular order specified in the user-specified input, and/or one or more other actions that are not ordered. For example, AONS blade 604 may establish an association between a particular message classification and one or more actions. Collectively, the operations of blocks 322-326 comprise “provisioning” the network element.
In block 328, one or more data packets that are destined for a device other than the network element are intercepted by the network element. The data packets may be, for example, data packets that contain IP and TCP headers. The IP addresses indicated in the IP headers of the data packets differ from the network element's IP address; thus, the data packets are destined for a device other than the network element. For example, network element 104, and more specifically, supervisor blade 602, may intercept data packets that client application 110 originally sent. The data packets might be destined for server application 112, for example.
In block 330, based on one or more information items indicated in the headers of the data packets, an application layer protocol that was used to transmit a message contained in the payload portions of the data packets (hereinafter “the message”) is determined. The information items may include, for example, a source IP address in an IP header, a destination IP address in an IP header, a TCP source port in a TCP header, and a TCP destination port in a TCP header. For example, network element 104, and more specifically AONS blade 604, may store mapping information that maps FTP (an application layer protocol) to a first combination of IP addresses and/or TCP ports, and that maps HTTP (another application layer protocol) to a second combination of IP addresses and/or TCP ports. Based on this mapping information and the IP addresses and/or TCP ports indicated by the intercepted data packets, AONS blade 604 may determine which application layer protocol (FTP, HTTP, SMTP, etc.) was used to transmit the message.
Referring now to
In block 334, the contents of the message are determined based on the termination technique that is associated with the application layer protocol that was used to transmit the message. For example, AONS blade 604 may provide the data packets as input to a procedure that is mapped to the application layer protocol determined in block 332. The procedure may use the appropriate message termination technique to extract the contents of the message from the data packets. The procedure may return the message as output to AONS blade 604. Thus, in one embodiment, the message extracted from the data packets is independent of the application layer protocol that was used to transmit the message.
In block 336, a message classification that is associated with criteria that the message satisfies is determined. For example, AONS blade 604 may store mapping information that maps different criteria to different message classifications. The mapping information indicates, among possibly many different associations, the association established in block 324. AONS blade 604 may determine whether the contents of the message satisfy criteria associated with any of the known message classifications. In one embodiment, if the contents of the message satisfy the criteria associated with a particular message classification, then it is determined that the message belongs to the particular message classification.
Although, in one embodiment, the contents of the message are used to determine a message's classification, in alternative embodiments, information beyond that contained in the message may be used to determine the message's classification. For example, in one embodiment, a combination of the contents of the message and one or more IP addresses and/or TCP ports indicated in the data packets that contain the message is used to determine the message's classification. For another example, in one embodiment, one or more IP addresses and/or TCP ports indicated in the data packets that contain the message are used to determine the message's classification, regardless of the contents of the message.
In block 338, one or more actions that are associated with the message classification determined in block 336 are performed. If two or more of the actions are associated with a specified order of performance, as indicated by the user-specified input, then those actions are performed in the specified order. If the output of any of the actions is supposed to be provided as input to any of the actions, as indicated by the user-specified input, then the output of the specified action is provided as input to the other specified action.
A variety of different actions may be performed relative to the message. For example, an action might be a “load-balancing” action that specifies one or more parameters. The parameters might include a pointer or reference to a load-balancing algorithm, such as a round-robin algorithm, a weighted round-robin algorithm, or an adaptive load-balancing algorithm. When the “load-balancing” action is performed, the load-balancing algorithm referenced by the action is invoked. Additionally, the parameters might include a pointer or reference to a session identifier locating technique. When the “load-balancing” action is performed, the session identifier locating technique referenced by the action is invoked. If a message contains a session identifier, then the message is sent towards the server application to which the session identifier is mapped.
As a result of the method illustrated in flow diagram 300B, network routers may be configured to perform data traffic load-balancing operations. Different load-balancing algorithms may be used in relation to different types of data traffic. Thus, for example, “purchase order” messages may be distributed among servers according to a first load-balancing algorithm, while “account transaction” messages may be distributed among servers according to a second, different load-balancing algorithm.
3.3 Action Flows
In other embodiments, any one of actions 402-414 may be performed individually or in combination with any others of actions 402-414.
3.4 AONS Examples
3.4.1 AONS General Overview
Application-Oriented Network Systems (AONS) is a technology foundation for building a class of products that embed intelligence into the network to better meet the needs of application deployment. AONS complements existing networking technologies by providing a greater degree of awareness of what information is flowing within the network and helping customers to integrate disparate applications by routing information to the appropriate destination, in the format expected by that destination; enforce policies for information access and exchange; optimize the flow of application traffic, both in terms of network bandwidth and processing overheads; provide increased manageability of information flow, including monitoring and metering of information flow for both business and infrastructure purposes; and provide enhanced business continuity by transparently backing up or re-routing critical business data.
AONS provides this enhanced support by understanding more about the content and context of information flow. As such, AONS works primarily at the message rather than at the packet level. Typically, AONS processing of information terminates a TCP connection to inspect the full message, including the “payload” as well as all headers. AONS also understands and assists with popular application-level protocols such as HTTP, FTP, SMTP and de facto standard middleware protocols.
AONS differs from middleware products running on general-purpose computing systems in that AONS' behavior is more akin to a network appliance, in its simplicity, total cost of ownership and performance. Furthermore, AONS integrates with network-layer support to provide a more holistic approach to information flow and management, mapping required features at the application layer into low-level networking features implemented by routers, switches, firewalls and other networking systems.
Although some elements of AONS-like functionality are provided in existing product lines from Cisco Systems, Inc., such products typically work off a more limited awareness of information, such as IP/port addresses or HTTP headers, to provide load balancing and failover solutions. AONS provides a framework for broader functional support, a broader class of applications and a greater degree of control and management of application data.
3.4.2 AONS Terminology
An “application” is a software entity that performs a business function either running on servers or desktop systems. The application could be a packaged application, software running on application servers, a legacy application running on a mainframe, or custom or proprietary software developed in house to satisfy a business need or a script that performs some operation. These applications can communicate with other applications in the same department (departmental), across departments within a single enterprise (intra enterprise), across an enterprise and its partners (inter-enterprise or B2B) or an enterprise and its customers (consumers or B2C). AONS provides value added services for any of the above scenarios.
An “application message” is a message that is generated by an application to communicate with another application. The application message could specify the different business level steps that should be performed in handling this message and could be in any of the message formats described in the section below. In the rest of the document, unless otherwise specified explicitly, the term “message” also refers to an application message.
An “AONS node” is the primary AONS component within the AONS system (or network). As described later, the AONS node can take the shape of a client proxy, server proxy or an intermediate device that routes application messages.
Each application message, when received by the first AONS node, gets assigned an AONS message ID and is considered to be an “AONS message” until that message gets delivered to the destination AONS node. The concept of the AONS message exists within the AONS cloud. A single application message may map to more than one AONS message. This may be the case, for example, if the application message requires processing by more than one business function. For example, a “LoanRequest” message that is submitted by a requesting application and that needs to be processed by both a “CreditCheck” application and a “LoanProcessing” application would require processing by more than one business function. In this example, from the perspective of AONS, there are two AONS messages: The “LoanRequest” to the “CreditCheck” AONS message from the requesting application to the CreditCheck application; and the “LoanRequest” to the “LoanProcessing” AONS message from the CreditCheck application to the LoanProcessing Application.
In one embodiment, AONS messages are encapsulated in an AONP (AON Protocol) header and are translated to a “canonical” format. Reliability, logging and security services are provided from an AONS message perspective.
The set of protocols or methods that applications typically use to communicate with each other are called “application access protocols” (or methods) from an AONS perspective. Applications can communicate to the AONS network (typically end point proxies: a client proxy and a server proxy) using any supported application access methods. Some examples of application access protocols include: IBM MQ Series, Java Message Service (JMS), TIBCO, Simple Object Access Protocol (SOAP) over Hypertext Transfer Protocol (HTTP)/HTTPS, and SMTP. Details about various access methods are explained in later sections of this document.
There are a wide variety of “message formats” that are used by applications. These message formats may range from custom or proprietary formats to industry-specific formats to standardized formats. Extensible Markup Language (XML) is gaining popularity as a universal language or message format for applications to communicate with each other. AONS supports a wide variety of these formats.
In addition, AONS provides translation services from one format to another based on the needs of applications. A typical deployment might involve a first AONS node that receives an application message (the client proxy) translating the message to a “canonical” format, which is carried as an AONS message through the AONS network. The server proxy might translate the message from the “canonical” format to the format understood by the receiving application before delivering the message. For understanding some of the non-industry standard formats, a message dictionary may be used.
A node that performs the gateway functionality between multiple application access methods or protocols is called a “protocol gateway.” An example of this would be a node that receives an application message through File Transfer Protocol (FTP) and sends the same message to another application as a HTTP post. In AONS, the client and server proxies are typically expected to perform the protocol gateway functionality.
If an application generates a message in Electronic Data Interchange (EDI) format and if the receiving application expects the message to be in an XML format, then the message format needs to be translated but the content of the message needs to be kept intact through the translation. In AONS, the end point proxies typically perform this “message format translation” functionality.
In some cases, even though the sending and receiving application use the same message format, the content needs to be translated for the receiving application. For example, if a United States-resident application is communicating with a United Kingdom-resident application, then the date format in the messages between the two applications might need to be translated (from mm/dd/yyyy to dd/mm/yyyy) even if the applications use the same data representation (or message format). This translation is called “content translation.”
3.4.3 AONS Functional Overview
As defined previously, AONS can be defined as network-based intelligent intermediary systems that efficiently and effectively integrate business and application needs with more flexible and responsive network services.
In particular, AONS can be understood through the following characteristics:
AONS operates at a higher layer (layers 5-6) than traditional network element products (layers 2-4). AONS uses message-level inspection as a complement to packet-level inspection—by understanding application messages, AONS adds value to multiple network element products, such as switches, firewalls, content caching systems and load balancers, on the “message exchange route.” AONS provides increased flexibility and granularity of network responsiveness in terms of security, reliability, traffic optimization (compression, caching), visibility (business events and network events) and transformation (e.g., from XML to EDI).
AONS is a comprehensive technology platform, not just a point solution. AONS can be implemented through distributed intelligent intermediary systems that sit between applications, middleware, and databases in a distributed intra- and inter-enterprise environment (routing messages, performing transformations, etc.). AONS provides a flexible framework for end user configuration of business flows and policies and partner-driven extensibility of AONS services.
AONS is especially well suited for network-based deployment. AONS is network-based rather than general-purpose server-based. AONS is hybrid software-based and hardware-based (i.e., application-specific integrated circuit (ASIC)/field programmable gate array (FPGA)-based acceleration). AONS uses out-of-band or in-line processing of traffic, as determined by policy. AONS is deployed in standalone products (network appliances) as well as embedded products (service blades for multiple switching, routing, and storage platforms).
3.4.4 AONS System Overview
This section outlines the system overview of an example AONS system.
Within AONS network 702, key building blocks include AONS Endpoint Proxies (AEPs) 704-710 and an AONS Router (AR). Visibility into application intent may begin within AEP 704 placed at the edge of a logical AONS “cloud.” As a particular client application of client applications 714A-N attempts to send a message across the network to a particular server application destination of server applications 716A-N and 718A-N, the particular client application will first interact with AEP 704.
AEP 704 serves as either a transparent or explicit messaging gateway which aggregates network packets into application messages and infers the message-level intent by examining the header and payload of a given message, relating the message to the appropriate context, optionally applying appropriate policies (e.g. message encryption, transformation, etc.) and then routing the message towards the message's application destination via a network switch.
AONS Router (AR) 712 may intercept the message en route to the message's destination endpoint. Based upon message header contents, AR 712 may determine that a new route would better serve the needs of a given application system. AR 712 may make this determination based upon enterprise-level policy, taking into account current network conditions. As the message nears its destination, the message may encounter AEP 706, which may perform a final set of operations (e.g. message decryption, acknowledgement of delivery) prior to the message's arrival. In one embodiment, each message is only parsed once: when the message first enters the AONS cloud. It is the first AEP that a message traverses that is responsible for preparing a message for optimal handling within the underlying network.
AEPs 704-708 can further be classified into AEP Client Proxies and AEP Server Proxies to explicitly highlight roles and operations performed by the AEP on behalf of the specific end point applications.
A typical message flow involves a particular client application 714A submitting a message to the AEP Client Proxy (CP) 704 through one of the various access protocols supported by AONS. On receiving this message, AEP CP 704 assigns an AONS message id to the message, encapsulates the message with an AONP header, and performs any necessary operations related to the AONS network (e.g. security and reliability services). Also, if necessary, the message is converted to a “canonical” format by AEP CP 704. The message is carried over a TCP connection to AR 710 along the path to the destination application 718A. The AONS routers along the path perform the infrastructure services necessary for the message and can change the routing based on the policies configured by the customer. The message is received at the destination AEP Server Proxy (SP) 706. AEP SP 706 performs necessary security and reliability functions and translates the message to the format that is understood by the receiving application, if necessary. AEP SP 706 then sends the message to receiving application 718A using any of the access protocols that application 718A and AONS support. A detailed message flow through AONS network 702 is described in later sections.
3.4.5 AONS System Elements
This section outlines the different concepts that are used from an AONS perspective.
An “AEP Client Proxy” is an AONS node that performs the services necessary for applications on the sending side of a message (a client). In the rest of this document, an endpoint proxy also refers to a client or server proxy. The typical responsibilities of the client proxy in processing a message are: message pre-classification & early rejection, protocol management, message identity management, message encapsulation in an AONP header, end point origination for reliable delivery, security end point service origination (encryption, digital signature, authentication), flow selection & execution/infrastructure services (logging, compression, content transformation, etc.), routing—next hop AONS node or destination, AONS node and route discovery/advertising role and routes, and end point origination for the reliable delivery mechanism (guaranteed delivery router).
Not all functionalities described above need to be performed for each message. The functionalities performed on the message are controlled by the policies configured for the AONS node.
An “AEP Server Proxy” is an AONS node that performs the services necessary for applications on the receiving side of a message (a server). In the rest of the document, a Server Proxy may also be referred as an end point proxy. The typical responsibilities of the Server Proxy in processing a message are: protocol management, end point termination for reliable delivery, security end point service termination (decryption, verification of digital signature, etc.), flow selection & execution/infrastructure services (logging, compression, content translation, etc.), message de-encapsulation in AONP header, acknowledgement to sending AONS node, application routing/request message delivery to destination, response message correlation, and routing to entry AONS node.
Note that not all the functionalities listed above need to be performed for each message. The functionalities performed on the message are controlled by the policies configured for the AONS node and what the message header indicates.
An “AONS Router” is an AONS node that provides message-forwarding functionalities along with additional infrastructure services within an AONS network. An AONS Router communicates with Client Proxies, Server Proxies and other AONS Routers. An AONS Router may provide service without parsing a message; an AONS Router may rely on an AONP message header and the policies configured in the AONS network instead of parsing messages. An AONS Router provides the following functionalities: scalability in the AONS network in terms of the number of TCP connections needed; message routing based on message destination, policies configured in the AONS cloud, a route specified in the message, and/or content of the message; a load at the intended destination—re-routing if needed; availability of the destination—re-routing if needed; cost of transmission (selection among multiple service providers); and infrastructure services such as sending to a logging facility, sending to a storage area network (SAN) for backup purposes, and interfacing to a cache engine for cacheable messages (like catalogs).
AONS Routers do not need to understand any of the application access protocols and, in one embodiment, deal only with messages encapsulated with an AONP header.
Application-Oriented Networking Protocol (AONP) is a protocol used for communication between the nodes in an AONS network. In one embodiment, each AONS message carries an AONP header that conveys the destination of the message and additional information for processing the message in subsequent nodes. AONP also addresses policy exchange (static or dynamic), fail-over among nodes, load balancing among AONS nodes, and exchange of routing information. AONP also enables application-oriented message processing in multiple network elements (like firewalls, cache engines and routers/switches). AONP supports both a fixed header and a variable header (formed using type-length-value (TLV) fields) to support efficient processing in intermediate nodes as well as flexibility for additional services.
Unless explicitly specified otherwise, “router” or “switch” refers herein to a typical Layer 3 or Layer 2 switch or a router that is currently commercially available.
3.4.6 AONS Example Features
In one embodiment, an underlying “AONS foundation platform of subsystem services” (AOS) provides a range of general-purpose services including support for security, compression, caching, reliability, policy management and other services. On top of this platform, AONS then offers a range of discreet functional components that can be wired together to provide the overall processing of incoming data traffic. These “bladelets™” are targeted at effecting individual services in the context of the specific policy or action demanded by the application or the information technology (IT) manager. A series of access method adaptors ensure support for a range of ingress and egress formats. Finally, a set of user-oriented tools enable managers to appropriately view, configure and set policies for the AONS solution. These four categories of functions combine to provide a range of end-customer capabilities including enhanced security, infrastructure optimization, business continuity, application integration and operational visibility.
The enhanced visibility and enhanced responsiveness enabled by AONS solutions provides a number of intelligent, application-oriented network services. These intelligent services can be summarized in four primary categories:
Enhanced security and reliability: enabling reliable message delivery and providing message-level security in addition to existing network-level security.
Infrastructure optimization: making more efficient use of network resources by taking advantage of caching and compression at the message level as well as by integrating application and network quality-of-service (QoS).
Business and infrastructure activity monitoring and management: by reading information contained in the application layer message, AONS can log, audit, and manage application-level business events, and combine these with network, server, and storage infrastructure events in a common, policy-driven management environment.
Content-based routing and transformation: message-based routing and transformation of protocol, content, data, and message formats (e.g., XML transformation). The individual features belonging to each of these primary categories are described in greater detail below.
3.4.6.1 Enhanced Security and Rellability
Authentication: AONS can verify the identity of the sender of an inbound message based upon various pieces of information contained within a given message (username/password, digital certificate, Security Assertion Markup Language (SAML) assertion, etc.), and, based upon these credentials, determine whether or not the message should be processed further.
Authorization: Once principal credentials are obtained via message inspection, AONS can determine what level of access the originator of the message should have to the services it is attempting to invoke. AONS may also make routing decisions based upon such derived privileges or block or mask certain data elements within a message once it's within an AONS network as appropriate.
Encryption/Decryption: Based upon policy, AONS can perform encryption of message elements (an entire message, the message body or individual elements such as credit card number) to maintain end-to-end confidentiality as a message travels through the AONS network. Conversely, AONS can perform decryption of these elements prior to arrival at a given endpoint.
Digital Signatures: In order to ensure message integrity and allow for non-repudiation of message transactions, AONS can digitally sign entire messages or individual message elements at any given AEP. The decision as to what gets signed will be determined by policy as applied to information derived from the contents and context of each message.
Reliability: AONS can complement existing guaranteed messaging systems by intermediating between unlike proprietary mechanisms. It can also provide reliability for HTTP-based applications (including web services) that currently lack reliable delivery. As an additional feature, AONS can generate confirmations of successful message delivery as well as automatically generate exception responses when delivery cannot be confirmed.
3.4.6.2 Infrastructure Optimization
Compression: AEPs can compress message data prior to sending the message data across the network in order to conserve bandwidth and conversely decompress it prior to endpoint delivery.
Caching: AONS can cache the results of previous message inquires based upon the rules defined for a type of request or based upon indicators set in the response. Caching can be performed for entire messages or for certain elements of a message in order to reduce application response time and conserve network bandwidth utilization. Message element caching enables delta processing for subsequent message requests.
TCP Connection Pooling: By serving as an intermediary between message clients and servers AONS can consolidate the total number of persistent connections required between applications. AONS thereby reduces the client and server-processing load otherwise associated with the ongoing initiation and teardown of connections between a mesh of endpoints.
Batching: An AONS intermediary can batch transactional messages destined for multiple destinations to reduce disk I/O overheads on the sending system. Similarly, transactional messages from multiple sources can be batched to reduce disk I/O overheads on the receiving system.
Hardware Acceleration: By efficiently performing compute-intensive functions such as encryption and Extensible Stylesheet Language Transformation (XSLT) transformations in an AONS network device using specialized hardware, AONS can offload the computing resources of endpoint servers, providing potentially lower-cost processing capability.
Quality of Service: AONS can integrate application-level QoS with network-level QoS features based on either explicit message prioritization (e.g., a message tagged as “high priority”) or via policy that determines when a higher quality of network service is required for a message as specific message content is detected.
Policy Enforcement: At the heart of optimizing the overall AONS solution is the ability to ensure business-level polices are expressed, implemented and enforced by the infrastructure. The AONS Policy Manager ensures that once messages are inspected, the appropriate actions (encryption, compression, routing, etc.) are taken against that message as appropriate.
3.4.6.3 Activity Monitoring and Management
Auditing/Logging/Metering: AONS can selectively filter messages and send them to a node or console for aggregation and subsequent analysis. Tools enable viewing and analysis of message traffic. AONS can also generate automatic responses to significant real-time events, both business and infrastructure-related. By intelligently gathering statistics and sending them to be logged, AONS can produce metering data for auditing or billing purposes.
Management: AONS can combine both message-level and network infrastructure level events to gain a deeper understanding of overall system health. The AONS management interface itself is available as a web service for those who wish to access it pro grammatically.
Testing and Validation: AONS' ability to intercept message traffic can be used to validate messages before allowing them to reach destination applications. In addition to protecting from possible application or server failures, this capability can be leveraged to test new web services and other functions by examining actual message flow from clients and servers prior to production deployment. AONS also provides a “debug mode” that can be turned on automatically after a suspected failure or manually after a notification to assist with the overall management of the device.
Workload Balancing and Failover: AONS provides an approach to workload balancing and failover that is both policy- and content-driven. For example, given an AONS node's capability to intermediate between heterogeneous systems, the AONS node can balance between unlike systems that provide access to common information as requested by the contents of a message. AONS can also address the issue of message affinity necessary to ensure failover at the message rather than just the session level as is done by most existing solutions. Balancing can also take into account the response time for getting a message reply, routing to an alternate destination if the preferred target is temporarily slow to respond.
Business Continuity: By providing the ability to replicate inbound messages to a remote destination, AONS enables customers to quickly recover from system outages. AONS can also detect failed message delivery and automatically re-route to alternate endpoints. AONS AEPs and ARs themselves have built-in redundancy and failover at the component level and can be clustered to ensure high availability.
3.4.6.4 Content-Based Routing and Transformation
Content-based Routing: Based upon its ability to inspect and understand the content and context of a message, AONS provides the capability to route messages to an appropriate destination by matching content elements against pre-established policy configurations. This capability allows AONS to provide a common interface (service virtualization) for messages handled by different applications, with AONS examining message type or fields in the content (part number, account type, employee location, customer zip code, etc.) to route the message to the appropriate application. This capability also allows AONS to send a message to multiple destinations (based on either statically defined or dynamic subscriptions to message types or information topics), with optimal fan-out through AONS routers. This capability further allows AONS to redirect all messages previously sent to an application so that it can be processed by a new application. This capability additionally allows AONS to route a message for a pre-processing step that is deemed to be required before receipt of a message (for example, introducing a management pre-approval step for all travel requests). Thus capability also allows AONS to route a copy of a message that exceeds certain criteria (e.g. value of order) to an auditing system, as well as forwarding the message to the intended destination. This capability further allows AONS to route a message to a particular server for workload or failover reasons. This capability also allows AONS to route a message to a particular server based on previous routing decisions (e.g., routing a query request based on which server handled for the original order). This capability additionally allows AONS to route based on the source of a message. This capability also allows AONS to route a message through a sequence of steps defined by a source or previous intermediary.
Message Protocol Gateway: AONS can act as a gateway between applications using different transport protocols. AONS supports open standard protocols (e.g. HTTP, FTP, SMTP), as well as popular or de facto standard proprietary protocols such as IBM Websphere MQ.
Message Transformations: AONS can transform the contents of a message to make them appropriate for a particular receiving application. This can be done for both XML and non-XML messages, the latter via the assistance of either a message dictionary definition or a well-defined industry standard format.
3.4.7 AONS Functional Modules
In one embodiment, AOS common services 806 include: security services, standard compression services, delta compression services, caching service, message logging service, policy management service, reliable messaging service, publish/subscribe service, activity monitoring service, message distribution service, XML parsing service, XSLT transformation service, and QoS management service.
In one embodiment, AOS protocol/access methods 810 include: TCP/SSL, HTTP/HTTPS, SOAP/HTTP, SMTP, FTP, JMS/MQ and JMS/RV, and Java Database Connectivity (JDBC).
In one embodiment, AOS message execution controller 808 includes: an execution controller, a flow subsystem, and a bladelett™ subsystem.
In one embodiment, AOS bladelets™ 818 and scriptlets™ 820 include: message input (read message), message output (send message), logging/audit, decision, external data access, XML parsing, XML transformation, caching, scriptlet container, publish, subscribe, message validation (schema, format, etc.), filtering/masking, signing, authentication, authorization, encryption, decryption, activity monitoring sourcing, activity monitoring marking, activity monitoring processing, activity monitoring notification, message discard, firewall block, firewall unblock, message intercept, and message stop-intercept.
In one embodiment, AOS configuration and management module 802 includes: configuration, monitoring, topology management, capability exchange, failover redundancy, reliability/availability/serviceability (RAS) services (tracing, debugging, etc.), archiving, installation, upgrades, licensing, sample scriptlets™, sample flows, documentation, online help, and language localization.
In one embodiment, supported platforms include: Cisco Catalyst 6503, Cisco Catalyst 6505, Cisco Catalyst 6509, and Cisco Catalyst 6513. In one embodiment, supported supervisor modules include: Sup2 and Sup720. In one embodiment, specific functional areas relating to the platform include: optimized TCP, SSL, public key infrastructure (PKI), encryption/decryption, interface to Cat6K supervisor, failover/redundancy, image management, and QoS functionality.
3.4.8 AONS Modes of Operation
AONS may be configured to run in multiple modes depending on application integration needs, and deployment scenarios. According to one embodiment, the primary modes of operation include implicit mode, explicit mode, and proxy mode. In implicit mode, an AONS node transparently intercepts relevant traffic with no changes to applications. In explicit mode, applications explicitly address traffic to an intermediary AONS node. In proxy mode, applications are configured to work in conjunction with AONS nodes, but applications do not explicitly address traffic to AONS nodes.
In implicit mode, applications are unaware of AONS presence. Messages are address to receiving applications. Messages are redirected to AONS via configuration of application “proxy” or middleware systems to route messages to AONS, and/or via configuration of networks (packet interception). For example, domain name server (DNS)-based redirection could be used to route messages. For another example, a 5-tuple-based access control list (ACL) on a switch or router could be used. Network-based application recognition and content switching modules may be configured for URL/URI redirection. Message-based inspection may be used to determine message types and classifications. In implicit mode, applications communicate with each other using AONS as an intermediary (implicitly), using application-native protocols.
Traffic redirection, message classification, and “early rejection” (sending traffic out of AONS layers prior to complete processing within AONS layers) may be accomplished via a variety of mechanisms, such as those depicted in
At the lowest layer, layer 902, all traffic passes through. At the next highest layer, layer 904, traffic may be filtered based on 5-tuples. A supervisor blade or Internetwork Operating System (IOS) may perform such filtering. Traffic that passes the filters at layer 904 passes to layer 906. At layer 906, traffic may be further filtered based on network-based application recognition-like filtering and/or message classification and rejection. Traffic that passes the filters at layer 906 passes to layer 908. At layer 908, traffic may be further filtered based on protocol headers. For example, traffic may be filtered based on URLs/URIs in the traffic. Traffic that passes the filters at layer 908 passes to layer 910. At layer 910, traffic may be processed based on application layer messages, include headers and contents. For example, XPath paths within messages may be used to process traffic at layer 910. An AONS blade may perform processing at layer 910. Thus, a select subset of all network traffic may be provided to an AONS blade.
In explicit mode, applications are aware of AONS presence. Messages are explicitly addressed to AONS nodes. Applications may communicate with AONS using AONP. AONS may perform service virtualization and destination selection.
In proxy mode, applications are explicitly unaware of AONS presence. Messages are addressed to their ultimate destinations (i.e., applications). However, client applications are configured to direct traffic via a proxy mode.
3.4.9 AONS Message Routing
Components of message management in AONS may be viewed from two perspectives: a node view and a cloud view.
Within AONS cloud 1010, the message is routed using AONP. Thus, using AONP, the message may be routed from AONS CP 1006 to an AONS router 1012, or from AONS CP 1006 to AONS SP 1008, or from AONS router 1012 to another AONS router, or from AONS router 1012 to AONS SP 1008. Messages processed at AONS nodes are processed in AONP format.
When the message reaches AONS SP 1008, AONS SP 1008 transforms the message into the message format used by server application 1014. AONS SP 1008 routes the message to server application 1014 using the message protocol of server application 1014. Alternatively, if AONS SP 1008 is not present, AONS CP 1006 may route the message to server application 1014.
The details of the message processing within AONS cloud 1010 can be understood via the following perspectives: Request/Response Message Flow, One-Way Message Flow, Message Flow with Reliable Delivery, and Node-to-Node Communication.
Referring to
According to one embodiment, AONP is used in node-to-node communication with the next hop. In one embodiment, AONP uses HTTP. AONP headers may include HTTP or TCP headers. AONP may indicate RM ACK, QoS level, message priority, and message context (connection, message sequence numbers, message context identifier, entry node information, etc.). The actual message payload is in the message body. Asynchronous messaging may be used between AONS nodes. AONS may conduct route and node discovery via static configuration (next hop) and/or via dynamic discovery and route advertising (“lazy” discovery).
At circumscribed numeral 4, AEP CP 1506 sends the AONP message to an AONS router 1508. At circumscribed numeral 5, AONS router 1508 receives the AONP message. At circumscribed numeral 6, AONS router 1508 sends the AONP message to an AEP SP 1510. At circumscribed numeral 7, AEP SP 1510 receives the AONP message and removes the AONP header from the message, thus decapsulating the message. At circumscribed numeral 8, AEP SP 1510 sends the message to receiving application 1504.
At circumscribed numeral 9, AEP SP 1510 sends a reliable messaging (RM) acknowledgement (ACK) to AONS router 1508. At circumscribed numeral 10, AONS router 1508 receives the RM ACK and sends the RM ACK to AEP CP 1506. At circumscribed numeral 11, AEP CP 1506 receives the RM ACK and, in response, deletes the copy of the message that is stored in data store 1512. Because the delivery of the message has been acknowledged, there is no further need to store a copy of the message in data store 1512. Alternatively, if AEP CP 1506 does not receive the RM ACK within a specified period of time, then AEP CP 1506 resends the message.
Referring to
At circumscribed numeral 10, AEP SP 1610 sends a reliable messaging (RM) acknowledgement (ACK) to AONS router 1608. At circumscribed numeral 11, AONS router 1608 receives the RM ACK and sends the RM ACK to AEP CP 1606. At circumscribed numeral 12, AEP CP 1606 receives the RM ACK and, in response, deletes the copy of the message that is stored in data store 1612. Because the delivery of the message has been acknowledged, there is no further need to store a copy of the message in data store 1612. Alternatively, if AEP CP 1606 does not receive the RM ACK within a specified period of time, then AEP CP 1606 resends the message.
At circumscribed numeral 6, AONS node 1706 receives a response message from server 1708 on the same connection on which AONS node 1706 sent the request message. At circumscribed numeral 7, AONS node 1706 reads the message, correlates the message with the request message, executes a flow, and adds an AONP header to the message. At circumscribed numeral 8, AONS node 1706 sends the message to AONS node 1704. At circumscribed numeral 9, AONS node 1704 reads the message, correlates the message with the request message, executes a flow, and removes the AONP header from the message, formatting the message according to the message format expected by client 1702. At circumscribed numeral 10, AONS node 1704 sends the message to client 1702 on the same connection on which client 1702 sent the request message to AONS node 1704.
According to the node view, the message lifecycle within an AONS node, involves ingress/egress processing, message processing, message execution control, and flow execution.
In one embodiment, executing the flow comprises executing each step (i.e., bladelet™/action) of the flow. If a bladelet™ is to be run within a separate context, then AMEC framework 1902 may enqueue into bladelet™-specific queues, and, based on thread availability, dequeue appropriate bladelet™ states from each bladelet™ queue.
3.4.10 Flows, Bladelets™, and Scriptlets™
According to one embodiment, flows string together bladelets™ (i.e., actions) to customize message processing logic. Scriptlets™ provide a mechanism for customers and partners to customize or extend native AONS functionality. Some bladelets™ and services may be provided with an AONS node.
3.4.11 AONS Services
As mentioned in the previous section, a set of core services may be provided by AONS to form the underlying foundation of value-added functionality that can be delivered via an AONS node. In one embodiment, these include: Security Services, Standard Compression Services, Delta Compression Services, Caching Service, Message Logging Service, Policy Management Service (Policy Manager), Reliable Messaging Service, Publish/Subscribe Service, Activity Monitoring Service, Message Distribution Service, XML Parsing Service, XSLT Transformation Service, and QoS Management Service. In one embodiment, each AONS core service is implemented within the context of a service framework.
3.4.12 AONS Configuration and Management
In one embodiment, an AONS node is provisioned and configured for a class of application messages, where it enforces the policies that are declaratively defined on behalf-of the application end-points, business-domains, security-domains, administrative domains, and network-domains. Furthermore, the AONS node promotes flexible composition and customization of different product functional features by means of configurability and extensibility of different software and hardware sub-systems for a given deployment scenario. Due to the application and network embodiments of the AONS functionality, the AONS architecture framework should effectively and uniformly address different aspects of configurability, manageability, and monitorability of the various system components and their environments.
The AONS Configuration and Management framework is based upon five functional areas (“FCAPS”) for network management as recommended by the ISO network management forum. The functional areas include fault management, configuration management, accounting management, performance management, and security management. Fault management is the process of discovering, isolating, and fixing the problems or faults in the AONS nodes. Configuration management is the process of finding and setting up the AONS nodes. Accounting management involves tracking usage and utilization of AONS resources to facilitate their proper usage. Performance management is the process of measuring the performance of the AONS system components and the overall system. Security management controls access to information on the AONS system. Much of the above functionality is handled via proper instrumentation, programming interfaces, and tools as part of the overall AONS solution.
3.4.13 AONS Monitoring
In one embodiment, AONS is instrumented to support well-defined events for appropriate monitoring and visibility into internal processing activities. The monitoring of AONS nodes may be accomplished via a pre-defined JMX MBean agent that is running on each AONS node. This agent communicates with a remote JMX MBean server on the PC complex. An AONS MIB is leveraged for SNMP integration to third party consoles.
3.4.14 AONS Tools
In one embodiment, the following tool sets are provided for various functional needs of AONS: a design studio, an admin studio, and a message log viewer. The design studio is a visual tool for designing flows and applying message classification and mapping policies. The admin studio is a web-based interface to perform all administration and configuration functions. The message log viewer is a visual interface to analyze message traffic, patterns, and trace information.
4.0 Implementation Mechanisms—Hardware Overview
Computer system 500 includes a bus 502 or other communication mechanism for communicating information, and a processor 504 coupled with bus 502 for processing information. Computer system 500 also includes a main memory 506, such as a random access memory (RAM), flash memory, or other dynamic storage device, coupled to bus 502 for storing information and instructions to be executed by processor 504. Main memory 506 also may be used for storing temporary variables or other intermediate information during execution of instructions to be executed by processor 504. Computer system 500 further includes a read only memory (ROM) 508 or other static storage device coupled to bus 502 for storing static information and instructions for processor 504. A storage device 510, such as a magnetic disk, flash memory or optical disk, is provided and coupled to bus 502 for storing information and instructions.
A communication interface 518 may be coupled to bus 502 for communicating information and command selections to processor 504. Interface 518 is a conventional serial interface such as an RS-232 or RS-322 interface. An external terminal 512 or other computer system connects to the computer system 500 and provides commands to it using the interface 514. Firmware or software running in the computer system 500 provides a terminal interface or character-based command interface so that external commands can be given to the computer system.
A switching system 516 is coupled to bus 502 and has an input interface 514 and an output interface 519 to one or more external network elements. The external network elements may include a local network 522 coupled to one or more hosts 524, or a global network such as Internet 528 having one or more servers 530. The switching system 516 switches information traffic arriving on input interface 514 to output interface 519 according to pre-determined protocols and conventions that are well known. For example, switching system 516, in cooperation with processor 504, can determine a destination of a packet of data arriving on input interface 514 and send it to the correct destination using output interface 519. The destinations may include host 524, server 530, other end stations, or other routing and switching devices in local network 522 or Internet 528.
The invention is related to the use of computer system 500 for avoiding the storage of client state on computer system 500. According to one embodiment of the invention, computer system 500 provides for such updating in response to processor 504 executing one or more sequences of one or more instructions contained in main memory 506. Such instructions may be read into main memory 506 from another computer-readable medium, such as storage device 510. Execution of the sequences of instructions contained in main memory 506 causes processor 504 to perform the process steps described herein. One or more processors in a multi-processing arrangement may also be employed to execute the sequences of instructions contained in main memory 506. In alternative embodiments, hard-wired circuitry may be used in place of or in combination with software instructions to implement the invention. Thus, embodiments of the invention are not limited to any specific combination of hardware circuitry and software.
The term “computer-readable medium” as used herein refers to any medium that participates in providing instructions to processor 504 for execution. Such a medium may take many forms, including but not limited to, non-volatile media, volatile media, and transmission media. Non-volatile media includes, for example, optical or magnetic disks, such as storage device 510. Volatile media includes dynamic memory, such as main memory 506. Transmission media includes coaxial cables, copper wire and fiber optics, including the wires that comprise bus 502. Transmission media can also take the form of acoustic or light waves, such as those generated during radio wave and infrared data communications.
Common forms of computer-readable media include, for example, a floppy disk, a flexible disk, hard disk, magnetic tape, or any other magnetic medium, a CD-ROM, any other optical medium, punch cards, paper tape, any other physical medium with patterns of holes, a RAM, a PROM, and EPROM, a FLASH-EPROM, any other memory chip or cartridge, a carrier wave as described hereinafter, or any other medium from which a computer can read.
Various forms of computer readable media may be involved in carrying one or more sequences of one or more instructions to processor 504 for execution. For example, the instructions may initially be carried on a magnetic disk of a remote computer. The remote computer can load the instructions into its dynamic memory and send the instructions over a telephone line using a modem. A modem local to computer system 500 can receive the data on the telephone line and use an infrared transmitter to convert the data to an infrared signal. An infrared detector coupled to bus 502 can receive the data carried in the infrared signal and place the data on bus 502. Bus 502 carries the data to main memory 506, from which processor 504 retrieves and executes the instructions. The instructions received by main memory 506 may optionally be stored on storage device 510 either before or after execution by processor 504.
Communication interface 518 also provides a two-way data communication coupling to a network link 520 that is connected to a local network 522. For example, communication interface 518 may be an integrated services digital network (ISDN) card or a modem to provide a data communication connection to a corresponding type of telephone line. As another example, communication interface 518 may be a local area network (LAN) card to provide a data communication connection to a compatible LAN. Wireless links may also be implemented. In any such implementation, communication interface 518 sends and receives electrical, electromagnetic or optical signals that carry digital data streams representing various types of information.
Network link 520 typically provides data communication through one or more networks to other data devices. For example, network link 520 may provide a connection through local network 522 to a host computer 524 or to data equipment operated by an Internet Service Provider (ISP) 526. ISP 526 in turn provides data communication services through the worldwide packet data communication network now commonly referred to as the “Internet” 528. Local network 522 and Internet 528 both use electrical, electromagnetic or optical signals that carry digital data streams. The signals through the various networks and the signals on network link 520 and through communication interface 518, which carry the digital data to and from computer system 500, are exemplary forms of carrier waves transporting the information.
Computer system 500 can send messages and receive data, including program code, through the network(s), network link 520 and communication interface 518. In the Internet example, a server 530 might transmit a requested code for an application program through Internet 528, ISP 526, local network 522 and communication interface 518. In accordance with the invention, one such downloaded application provides for avoiding the storage of client state on a server as described herein.
Processor 504 may execute the received code as it is received and/or stored in storage device 510 or other non-volatile storage for later execution. In this manner, computer system 500 may obtain application code in the form of a carrier wave.
5.0 Extensions and Alternatives
In the foregoing specification, the invention has been described with reference to specific embodiments thereof. It will, however, be evident that various modifications and changes may be made thereto without departing from the broader spirit and scope of the invention. The specification and drawings are, accordingly, to be regarded in an illustrative rather than a restrictive sense.
This application is related to U.S. patent application Ser. No. 10/991,792, entitled “PERFORMING MESSAGE AND TRANSFORMATION ADAPTER FUNCTIONS IN A NETWORK ELEMENT ON BEHALF OF AN APPLICATION”, by Pravin Singhal, Qingqing Li, Juzar Kothambalawa, Parley Van Oleson, Wai Yip Tung, and Sunil Potti, filed on Nov. 17, 2004; U.S. patent application Ser. No. 10/997,616, entitled “CACHING CONTENT AND STATE DATA AT A NETWORK ELEMENT”, by Alex Yiu-Man Chan, Snehal Haridas, and Raj De Datta, filed on Nov. 23, 2004; U.S. patent application Ser. No. 11/005,978, entitled “PERFORMING MESSAGE PAYLOAD PROCESSING FUNCTIONS IN A NETWORK ELEMENT ON BEHALF OF AN APPLICATION”, by Tefcros Anthias, Sandeep Kumar, Ricky Ho, and Saravanakumar Rajendran, filed on Dec. 6, 2004; U.S. patent application Ser. No. 11/007,421, entitled “PERFORMING SECURITY FUNCTIONS ON A MESSAGE PAYLOAD IN A NETWORK ELEMENT”, by Sandeep Kumar, Subramanian Srinivasan, Tefcros Anthias, Subramanian Iyer, and Christopher Wiborg, filed on Dec. 7, 2004; U.S. patent application Ser. No. 11/007,152, entitled “NETWORK AND APPLICATION ATTACK PROTECTION BASED ON APPLICATION LAYER MESSAGE INSPECTION”, by Sandeep Kumar, Yi Jin, Sunil Potti, and Christopher Wiborg, filed on Dec. 7, 2004; U.S. patent application Ser. No. 11/009,127, entitled “REDUCING THE SIZES OF APPLICATION LAYER MESSAGES IN A NETWORK ELEMENT”, by Ricky Ho, Tefcros Anthias, Kollivakkam R. Raghavan, and Alex Yiu-Man Chan, filed on Dec. 10, 2004; U.S. patent application Ser. No. 11/009,270, entitled “GUARANTEED DELIVERY OF APPLICATION LAYER MESSAGES BY A NETWORK ELEMENT”, by Tefcros Anthias and Ricky Ho, filed on Dec. 10, 2004; U.S. patent application Ser. No. ______, entitled “INTERPRETING AN APPLICATION MESSAGE AT A NETWORK ELEMENT USING SAMPLING AND HEURISTICS” (Attorney Docket No. 50325-0915), by Tefcros Anthias, Sunil Potti, Subramanian Srinivasan, and Nitesh Trikha, filed on Jan. 5, 2005; and U.S. patent application Ser. No. ______, entitled “DATA TRAFFIC LOAD BALANCING BASED ON APPLICATION LAYER MESSAGES” (Attorney Docket No. 50325-0916), by Hari Kathi, Subramanian Srinivasan, and Pravin Singhal, filed on Jan. 6, 2005; the contents of all of which are incorporated by reference in their entirety for all purposes as though fully disclosed herein.
