System and method of algorithmically generating a server side transaction identifier

Abstract

A system, medium and method for generating a server side transaction ID (XID′) is disclosed. A request is received from a client device to access a server. The request includes a source port, source IP address, protocol information and a client generated transaction ID (XID). An endian'ness of the client of determined and classified. A Client ID unique to the client request is then generated using the source port, source IP address, protocol information, and a Masked XID generated from the endian'ness determination and the XID. A server side transaction ID (XID′) is then synthesized by combining the Client ID and an XID halfword containing least significant bits (LSB) identified from the endian'ness determination. The XID′ is then transmitted to the file server, wherein the XID′ is associated with the XID for the corresponding client request.

Description

FIELD

The present disclosure is generally directed to a system and method of algorithmically generating a server side transaction identifier and replicating the same among one or more virtualization devices.

BACKGROUND

In a file system virtualization environment, a Network File System (NFS) protocol assigns a unique Transaction ID (XID) to each requested operation from a client device to a file server device. The XID is generated by the client device when issuing the request, whereby the XID is returned by the server in the server response to the client device. An application layer gateway device is utilized to multiplex requests between multiple clients and one or more file servers, whereby the gateway maps the XIDs received from the client devices into new, unique Transaction ID (XID′) values before forwarding the client devices' requests to the server(s). The gateway device must save the mapping information between the client device and the XID by using the identity of the client device (i.e. its network address) and XID. Further, the gateway device must also map its created XID′ to the client device. When the server returns a response with the XID′, the gateway device maps the received XID′ back to the originating client identity and XID. Therefore, the gateway device must save and retrieve the original XID included in the client device's request to be able to include it in the server response that is ultimately sent back to the client device. Accordingly, the gateway device must store corresponding XID/XID′ entries for every operation in progress in a mapping table.

Further, some protocols, including the NFS protocol, use the XID information to identify operations that are duplicates of prior operations that are being retried due to timeouts or other errors. The application gateway device must therefore map a retransmitted operation for a client device's XID to the same XID′ used the first time the operation was proxied by the gateway device. This ensures the server detects this request as a retransmitted request instead of a new request. Considering that the gateway device must store the XID/XID′ information for every operation in a mapping table, and that client devices are capable of sending thousands of requests per second, the mapping table used by the gateway device can easily grow to millions of entries. Further, if the gateway device utilizes a backup gateway device, the mapping table must also be stored in a shared memory and replicated in the backup gateway device so that the backup device can seamlessly handle communications in case the active gateway device goes off-line. This current implementation imposes a significant burden on the active and backup gateway devices.

What is needed is a file virtualization device which algorithmically generates a server side transaction identifier (XID′) and replicates the algorithm among one or more virtualization devices without requiring the needed amount of substantial storage space to store XID/XID′ mapping information.

SUMMARY

In an aspect, a method of generating a server side transaction ID (XID′) is disclosed. The method comprises receiving, at a file virtualization device, a request from a client device to access a server, wherein the request includes a source port, source IP address, protocol information and a client generated transaction ID (XID). The method comprises determining an endian'ness of the client. The method comprises generating a Client ID for the client request, the Client ID utilizing the source port, source IP address, protocol information, and a Masked XID generated from the endian'ness determination and the XID. The method comprises synthesizing a server side transaction ID (XID′) by combining the Client ID and an XID halfword containing least significant bits (LSB) identified from the endian'ness determination. The method comprises transmitting the XID′ to the file server, wherein the XID′ is associated with the XID for the corresponding client request.

In another aspect, a non-transitory computer readable medium having stored thereon instructions for generating a server side transaction ID (XID′) is disclosed. The medium comprises machine executable code, which when executed by at least one machine, causes the machine to receive a request from a client device to access a server. The request includes a source port, source IP address, protocol information and a client generated transaction ID (XID). The code causes the machine to determine an endian'ness of the client. The code causes the machine to generate a Client ID unique to the client request, wherein the Client ID utilizes the source port, source IP address, protocol information, and a Masked XID generated from the endian'ness determination and the XID. The code causes the machine to synthesize a server side transaction ID (XID′) by combining the Client ID and an XID halfword containing least significant bits (LSB) identified from the endian'ness determination. The code causes the machine to transmit the XID′ to the file server, wherein the XID′ is associated with the XID for the corresponding client request.

In an aspect, a file virtualization device comprises a network interface that is configured to receive client requests and transmit the client requests to one or more file servers. The file virtualization device comprises a memory configured to store programming instructions in executable code for generating a server side transaction ID (XID′). The file virtualization device comprises a processor that is configured execute the code, which causes the processor to receive a request from a client device to access a server. The request includes a source port, source IP address, protocol information and a client generated transaction ID (XID). The processor is further configured to determine an endian'ness of the client. The processor is further configured to generate a Client ID that is unique to the client request which utilizes the source port, source IP address, protocol information, and a Masked XID generated from the endian'ness determination and the XID. The processor is further configured to synthesize a server side transaction ID (XID′) by combining the Client ID and a XID halfword containing least significant bits (LSB) identified from the endian'ness determination. The processor is further configured to transmit the XID′ to the file server, wherein the XID′ is associated with the XID for the corresponding client request.

In one or more of the above aspects, the generated Client ID is stored as a new entry in a Client ID Table, wherein at least a portion of the Client ID Table is replicated to at least one other processor in the file virtualization device and/or another file virtualization device.

In one or more of the above aspects, the endian'ness of the client is identified as having a Big Endian format, a Little Endian format, an Unknown format or a Random format. The identified endian'ness is then stored in a Classification table in memory. In the case that the endian'ness is classified as having the Random format, a first XID halfword is converted to a zero bit value in generating the Client ID.

In one or more of the above aspects, endian'ness is determined by first receiving and storing a plurality of XIDs from the client for a set number of client requests. A plurality of upper halfwords are then compared to identify a number of changed halfwords among adjacent upper halfwords in the set. An entropy classification is then designated to the upper halfwords based on the number of identified changed halfwords among upper halfwords. A plurality of lower halfwords are then compared to identify a number of changed halfwords among adjacent lower halfwords in the set. An entropy classification is then designated to the lower halfwords based on the number of identified changed halfwords among lower halfwords. An XID classification is then assigned to the client based on the entropy classifications designated to the upper and lower halfwords.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1A is an example of system including one or more file virtualization devices in accordance with an aspect of the present disclosure;

FIG. 1B is a block diagram of a relational operation in association with one or more file virtualization devices in accordance with an aspect of the present disclosure;

FIG. 1C is a block diagram of a file virtualization device including a generator module in accordance with an aspect of the present disclosure;

FIG. 2 illustrates a block diagram of the generator module in accordance with an aspect of the present disclosure;

FIG. 3 illustrates a general diagram representative of a overall process performed by the generator module in accordance with an aspect of the present disclosure;

FIG. 4 illustrates a flow chart depicting at least a portion of the XID generating process performed by the generator module in accordance with an aspect of the present disclosure; and

FIG. 5 illustrates a flow chart depicting at least a portion of the endian determination process performed by the generator module in accordance with an aspect of the present disclosure.

DETAILED DESCRIPTION

In general, the present disclosure is directed to a system and method which algorithmically generates a server side transaction ID (XID′) using a client generated transaction ID (XID). In particular, a generator module within a file virtualization device receives a client request for access to one or more file servers, wherein the client request contains the XID. The generator module determines the endian'ness classification of the client and generates a Masked XID from the classification information and the client XID itself. The generator module thereafter extracts the source IP address, source port and protocol information from the client request and combines it with the Masked ID to generate a Client ID and a corresponding search key unique to the client-server transaction. The Client ID is stored in a Client ID Table. The generator module then combines the Client ID with the least significant bit portion of the client XID to synthesize the server side XID′. The file virtualization device transmits the XID′ along with its address information to the one or more requested file servers. The file virtualization device is also configured to share replicated copies of the Client ID Table and other transaction data to other cores in the same and/or other file virtualization devices.

Advantages of the present system and method are that they reduce the data and communication overhead associated with storing and sharing XID/XID′ mapping information. In particular, the present system and method stores and replicates entries in the Client ID Table instead of a complete XID/XID′ mapping table. The present system optimizes DRC update message exchanges between cores within or exterior to the active core by only synchronizing elements of the Client ID Table. At this reduced message rate, the DRC updates in a file virtualization device can be saved to network accessible storage instead of having to be sent to a backup file virtualization device. This eliminates the requirement for the backup file virtualization device to locally save copies of the DRC updates.

FIG. 1A is an example of a network environment including one or more file virtualization systems in accordance with an aspect of the present disclosure. As shown in FIG. 1A, the network environment includes one or more server devices 102(1)-102(n), one or more client devices 104(1)-104(n), and one or more data center sites 100. The data center site 100 within the environment comprises a network connection via network 112 between the file virtualization devices 110(1)-110(n) and client devices 104(1)-104(n) a secured or unsecured connection via LAN 114 between one or more file virtualization devices 110(1)-110(n) and the servers 102(1)-102(n). The ellipses and the designation “n” in the figures denote an unlimited number of file server devices, file virtualization devices, and/or client devices. It should be noted that although only one data center site 100 is shown in FIG. 1A, more than one backup data center sites (not shown) may be employed in the environment.

For purposes of discussion, the file virtualization system 110, when in an active state, hosts active services and operates to execute various virtualization services between client devices and virtual file servers. In particular to the present disclosure, the one or more of the file virtualization devices 110(1)-110(n) are configured to perform an algorithmic process to generate a server side XID′ between the file virtualization devices 110(1)-110(n) and the file servers 102(1)-102(n).

In this example, the network 112 comprises a publicly accessible network, for example, the Internet. Communications, such as read and write requests between client devices 104(1)-104(n) and file server devices 102(1)-102(n) take place over the network 112 according to standard network protocols, such as the HTTP, TCP/IP, request for comments (RFC) protocols, Common Internet File System (CIFS) protocols, Network File System (NFS) protocols and the like.

Further, the network 112 can include local area networks (LANs), wide area networks (WANs), direct connections and any combination thereof, other types and numbers of network types. On an interconnected set of LANs or other networks, including those based on different architectures and protocols, routers, switches, hubs, gateways, bridges, and other intermediate network devices may act as links within and between LANs and other networks to enable messages and other data to be sent between network devices. Also, communication links within and between LANs and other networks typically include twisted wire pair (e.g., Ethernet), coaxial cable, analog telephone lines, full or fractional dedicated digital lines including T1, T2, T3, and T4, Integrated Services Digital Networks (ISDNs), Digital Subscriber Lines (DSLs), wireless links including satellite links and other communications links known to those skilled in the relevant arts.

LAN 114 includes a private local area network that allows communications between the file virtualization devices 110(1)-110(n) and one or more file server devices 102(1)-102(n), although the LAN 114 may comprise other types of private and public networks with other devices.

File server devices 102(1)-102(n) are capable of performing operations such as, for example, storing files and data in a file system. In an aspect, file server devices 102(1)-102(n) are accessed by client devices 104(1)-104(n) via the file virtualization system 110. In FIG. 1A, although two file server devices 102(1)-102(n) are shown it should be understood that any number of file server devices, including one, can be used. In an aspect, file server devices 102(1)-102(n) can be heterogeneous devices provided by different independent manufacturers. Further, according to various examples, file server devices 102(1)-102(n) can be used to form a tiered storage arrangement in which high priority data and/or frequently accessed data is stored in fast, more expensive file server devices and low priority and/or relatively less accessed data can be stored in slower, less expensive file server devices. A series of applications run on the file server devices 102(1)-102(n) that allow the transmission of data, cookies, descriptor files, namespace data, and other file system data. The file server devices 102(1)-102(n) can provide data or receive data in response to requests from the client devices 104(1)-104(n). In an aspect, file server devices 102(1)-102(n) may store and/or provide other data representative of requested resources, such as particular Web page(s), image(s) of physical objects, and any other objects.

As shown in FIG. 1A, client devices 104(1)-104(n) communicate with the file server devices 102(1)-102(n) via the file virtualization device 110, whereby the client devices 104(1)-104(n) make requests to retrieve as well as send data to the file server devices 102(1)-102(n) via the network 112. Generally, client devices 104(1)-104(n) can include virtually any network device capable of connecting to another network device to send and receive information, including Web-based information. The set of such devices can include devices that typically connect using a wired (and/or wireless) communications medium, such as personal computers (e.g., desktops, laptops, tablets), smart TVs, stand alone multimedia boxes, mobile and/or smart phones and the like.

Each of the file server devices 102(1)-102(n), file virtualization devices 110(1)-110(n), and client devices 104(1)-104(n) can include a central processing unit (CPU), controller or processor, a memory, and an interface system which are coupled together by a bus or other link, although other numbers and types of each of the components and other configurations and locations for the components can be used.

Generally, the file virtualization devices 110(1)-110(n) in the file virtualization system 110 simplify storage management and lower total storage management costs. In an aspect, the file virtualization devices 110(1)-110(n) automate data management tasks and eliminate the disruption associated with storage management operations. The file virtualization system 110 provides a virtual layer of intelligence between the network 112 and the respective file server devices 102(1)-102(n). The file virtualization system 110 thus eliminates the inflexible mapping which typically ties client devices to physical file server devices. The file virtualization system 110(1)-110(n) decouples the logical access to files from their physical location, so files are free to move among different file server devices, which are now free to change without disrupting users, applications, or administrators. The file virtualization devices 110(1)-110(n) implement intelligent file virtualization that simplifies data management further by providing automated, policy-based management across heterogeneous storage environments.

An example file virtualization device can be the ARX® Series devices provided by F5 networks, Inc. of Seattle, Wash. The file virtualization device can be configured to plug directly into existing IP/Ethernet network 112 and/or LAN 114, in substantial real-time. The file virtualization devices 110(1)-110(n) are configured to virtualize heterogeneous file server devices 102(1)-102(n) that present file systems via NFS and/or CIFS, for example.

In an example, the file virtualization devices 110(1)-110(n) do not connect directly to a storage area network (SAN) but instead manages SAN data presented through a gateway or file server device, without changing the existing infrastructure of the system 100. The file virtualization devices 110(1)-110(n) appear as a single gateway device to client devices 104(1)-104(n), and as a single CIFS or NFS client to their respective file server devices. In an aspect, the file virtualization devices can be configured to carry out data management operations, although the file virtualization devices can additionally or alternative carry out storage management operations.

For example, the file virtualization devices 110(1)-110(n) may be configured to automate common storage management tasks (e.g., data migration, storage tiering, and/or load balancing), which take place without affecting access to the file data or requiring re-configuration of file system(s) on client devices 104(1)-104(n). The file virtualization devices 110(1)-110(n) manage metadata that tracks the location of files and directories that are distributed across file server devices 102(1)-102(n), which is stored in configuration data. The file virtualization devices 110(1)-110(n) use the configuration data to utilize namespace data, which is an aggregation of the underlying file systems, and as well as masked changes to the underlying storage systems from users and applications of client devices 104(1)-104(n).

Along with generating XID′(s), file virtualization devices 110(1)-110(n) of the active file virtualization system 110 continually replicate client ID information among multiple cores 116(1)-116(n) within a single file virtualization device 110(1), as shown by arrow 120A (FIG. 1B). In particular to FIG. 1B, network traffic management device 110(1) may include ‘n’ number of cores 116, whereby client ID information may be replicated from the active core 116(1) to one or more other cores 116(n) within the active network traffic management device 110(1) and/or within one or more cores 118(1)-118(n) in one or more backup file virtualization devices 110(n) in the same data center and/or in another data center. In an aspect, Client ID information is replicated to one or more file systems separate from the file virtualization devices 110, whereby the file systems can provide the Client ID information to any core in any file virtualization device.

In an aspect, the client ID data is synchronously exported from one or more active file virtualization devices to one or more backup file virtualization devices. In a particular aspect, the Client ID data is exported from one or more active cores in a file virtualization device to one or more backup cores within itself or inside other file virtualization devices. The non-active file virtualization device(s) and/or cores, upon receiving the imported Client ID data, will store the Client ID data in a memory associated with that file virtualization device. In an aspect, the replicated Client ID data is stored in a storage device exterior to the file virtualization devices 110(1)-110(n). The backup virtualization device and/or core becomes operational when the active virtualization device fails and/or if one or more file server devices 102(1)-102(n) fail. The failure can occur as a result of a catastrophic disaster, equipment breakdown, or equipment/software upgrade.

FIG. 1C is a block diagram of an example file virtualization device in accordance with an aspect of the present disclosure. As shown in FIG. 1C, the file virtualization device 110 includes one or more data planes 122, one or more control planes 132, one or more input-output devices 142 and one or more displays 144. It should be noted that the illustrated example file virtualization device 110 may include additional, fewer and/or different hardware modules.

The network input-output interface 124 is configured to allow the file virtualization device 110 to communicate with other network devices, such as one or more other file virtualization devices, via any network protocol such as TCP or UDP.

Input-output device 142 may in some examples connect to multiple input-output devices external to file virtualization device 110. Some examples of the input-output device 142 may be configured to provide storage or an installation medium, while others may provide a universal serial bus (USB) interface for receiving USB file server devices such as the USB Flash Drive. Still other examples of the input-output device 142 may be a bridge between the bus 130 (in data and/or control plane) and an external communication bus, such as: a USB bus; an Apple Desktop Bus; an RS-232 serial connection; a SCSI bus; a FireWire bus; a FireWire 800 bus; an Ethernet bus; an AppleTalk bus; a Gigabit Ethernet bus; an Asynchronous Transfer Mode bus; a HIPPI bus; a Super HIPPI bus; a SerialPlus bus; a SCI/LAMP bus; a FibreChannel bus; or a Serial Attached small computer system interface bus.

In an aspect, the data plane 122 of the file virtualization device 110 functions to provide a data path that handles non-metadata operations at wire speed. The control plane 132 of the file virtualization device 110 functions to provide handling of operations that affect metadata and migration of file data to and from file server devices 102(1)-102(n). In an aspect, the control plane 132 is configured to perform certain functions such as logging, reporting, port mirroring, and hosting Simple Network Management Protocol (SNMP) and other protocols. Control plane memory 138 can store an operating system used for file virtualization device 110, and log files generated during operation of file virtualization device 110. Each path provided by data plane 122 and control plane 132, respectively, has dedicated processing and memory resources and each can scale independently based upon varying network and storage conditions.

In this example shown in FIG. 1B, the data plane 122 includes one or more data plane processors (CPU) 126, one or more data plane memories 128, and one or more input-output interfaces 124 coupled to each other through one or more internal data plane bus 130.

Similarly, in this example, the control plane 132 includes one or more control plane processors (CPU) 136, one or more control plane memories 138 and one or more configuration databases 150, all coupled to one another via internal control plane bus.

The configuration database 150 is configured to store object relationships of the configuration data and mapping information between the various objects in the file system managed by file virtualization device 110. Additionally, as shown in FIG. 1B, the control plane 132 is able to communicate with the input-output device 142 and the display 144 via the internal control plane bus 130.

Data plane CPU 126 and control plane CPU 136 are configured to process instructions fetched from the data plane memory 128; one or more microprocessor units, one or more microprocessors, one or more microcontrollers, and central processing units with a single processing core or a plurality of processing cores 116, 118.

The data plane memory 128 and the control plane memory 138, can comprise: Static random access memory (SRAM), Burst SRAM or SynchBurst SRAM (BSRAM), Dynamic random access memory (DRAM), Fast Page Mode DRAM (FPM DRAM), Enhanced DRAM (EDRAM), Extended Data Output RAM (EDO RAM), Extended Data Output DRAM (EDO DRAM), Burst Extended Data Output DRAM (BEDO DRAM), Enhanced DRAM (EDRAM), synchronous DRAM (SDRAM), JEDECSRAM, PCIOO SDRAM, Double Data Rate SDRAM (DDR SDRAM), Enhanced SDRAM (ESDRAM), SyncLink DRAM (SLDRAM), Direct Rambus DRAM (DRDRAM), Ferroelectric RAM (FRAM), disk type memory, tape memory, spinning storage media, or any other type of memory device capable of executing the systems and methods described herein.

As shown in FIG. 1C, the data plane memory 128 includes a generator module 130 which includes non-transitory software code which, when executed by processor or CPU 126, causes the CPU 126 to perform the novel functions described below. Although the generator module 130 is shown in memory 128, it is contemplated that module 130 may be located elsewhere within or exterior to memory 128.

The system and method of the present disclosure utilizes a novel concept based on the observation that each client device does not generate a random XID, but instead utilizes an internal XID generator that operates in conjunction with a counter. Thus, for each client generated XID, only the least significant bits (LSB) of the XID actually change among multiple requests sent from the client device. Thus, the most significant bits (MSB) of the XID 306 do not typically change among multiple requests, unless a wraparound occurs (discussed below).

In general, the generation module 130 synthesizes a unique Client ID 308 from the client IP address 300, transport protocol 302, source port 304, and a generated Masked XID 314 which is determined from the XID classification 312 and the client XID 306. The generator module 130 concatenates the client ID 306 with the least significant bits (LSB) of the client's XID to algorithmically generate a XID′ 310 which uniquely identifies the client's request. The synthesized client XID′ 310 is passed from the file virtualization device 110 to the file server 102. One or more Client IDs are replicated and exported or transmitted to one or more other file virtualization devices 110(n) and/or one or more cores in the file virtualization device 110(1) (i.e. cores 116(1)-116(n)) and/or in the other file virtualization devices 110(n) (i.e. cores 118(1)-118(n)).

FIG. 2 illustrates a block diagram of the generator module in accordance with an aspect of the present disclosure. FIG. 3 illustrates a general diagram representative of an overall process performed by the generator module in accordance with an aspect of the present disclosure. As shown in FIG. 2, the generator module 130 includes a plurality of software modules such as an XID classifier 202, an endian'ness identifier 204, a client identifier 206, a client table handler 208, a client storage handler 210 and a XID′ synthesizer 212. In an aspect, the generator module 130, via the client table handler 208, is configured to store and retrieve information with a Client ID Table 214 in one or more memories 128. In an aspect, the generator module 130, via the client storage handler 210, is configured to store and retrieve information with a Classification Table 216 and/or a Client ID Database 218 in one or more memories 128. It should be noted that the illustrated example generator module 130 may include additional, fewer and/or different software modules.

The XID Classifier 202 is configured to classify XIDs based on the endian'ness of each of the client's received 32-bit XIDs. The endian'ness is determined by the endian'ness identifier 204 and allows the XID Classifier 202 to identify which portion of the XID 306 has the most significant bits (MSB) 306A and the least significant bits (LSB) 306B. The Client Identifier 206 uses a tuple of data from the client request to create a 16 bit Client ID 308 that is unique to the particular transaction associated with the client request. In an aspect, as shown in FIG. 3, the tuple includes the source IP address 300, source port 302, protocol 304, client endian'ness 312, and a Masked XID 314 which is determined from endian'ness XID classification 312 of the client 104. As will be discussed in more detail, the generated Client ID 308 forms the higher order component 310A of the generated XID′ 310.

A new Client ID 306 is generated when the requesting client is new (by IP address, protocol, or port) or when the LSB 16 bits of the client's XID halfword has overflowed and changed the MSB halfword of the client's XID (i.e. wraparound). After a Client ID 308 has been generated by the generator module 130, it is stored in the Client ID Table 214. Accordingly, the Client ID Table 214 changes when a newly generated Client ID is entered. In an aspect, at least a portion of the Client ID Table 214 is replicated to one or more cores in the same or other file virtualization devices or in one or more shared file servers/databases in case of a failover. In one aspect, only newly generated Client ID(s) 308, and not the entire Client ID Table 214, are replicated. In another aspect, the entire Client ID Table 214 is replicated.

The Classification Table 216 contains mapping information between the client's IP address 300 and the classified endian'ness of that client 104. The Client ID Database 218 contains a list of issued and available Client IDs per data plane core.

The XID′ synthesizer 212 combines the Client ID 308 generated by the client identifier 206 with the 16 lower order LSB bits of the client XID to form the synthesized XID′ 310 that is ultimately sent to the file server 102.

FIG. 4 illustrates a flow chart depicting at least a portion of the XID′ generating process performed by the generator module in accordance with an aspect of the present disclosure. It should be noted that one or more steps may be performed in a different order with respect to the other steps. It should also be noted that additional or fewer steps are contemplated in the illustrated flow chart.

As shown in FIG. 4, the file virtualization device 110 receives a request from a client 104 over network 112, wherein the client's request for the transaction is to access one or more file servers 102(1)-102(n) (Block 400). In an aspect, the client request is a remote procedure call (RPC) utilizing a Network File System (NFS) protocol (referred to as a “NFS request”), although other types of protocols are contemplated. As partially illustrated in FIG. 3, the client's NFS request includes at least a 32 bit source IP address 300, a 16 bit client source port 302, an 8 bit protocol 304 and a 32 bit client generated XID 306.

After the client request is received, the generator module 130 in general determines whether an XID classification for the client 104 is stored in the Classification Table 216 (Block 402). In particular, the generator module 130 identifies the client 104 by the source IP address 300 and uses it to perform a lookup in the Classification Table 216 to determine the classified XID endian'ness 312 of the client 104. As explained in more detail below, the classified XID endian'ness 312 is used to generate a Masked XID 314.

As shown in Block 404, the generator module 130 inquires whether an entry for the XID classification is found in the Classification Table 216. If there is no XID classification entry in the Classification Table 216, then a new table entry containing a new XID classifier is created in the Classification Table 216 and the generator module 130 initiates a process to classify the client XID (Block 406). In an aspect, the client XID 306 is classified by analyzing the endian'ness of the XID 306 using a novel process described below in FIG. 5. It should be noted, however, that the process described in FIG. 4 may utilize other appropriate methods to classify the XID while remaining enabling.

Once the XID classification of the client 104 is known, the client identifier 204 generates a search key to search the Client ID Table 214 to find the Client ID 308 in the Client ID Table 214 which corresponds to the client's request (Block 408). In particular, the search key includes the source IP address 300, port 302, protocol 304, and a Masked XID 314 which is generated based on the client's XID classification 312. In particular, the Masked XID 314 utilizes two input parameters, the client's XID 306 and the client endian'ness 312. The Masked XID is a full 32-bit word.

The XID classifier 202 is configured to classify the client XID as having a Big Endian format, a Little Endian format, an Unknown format or a Random format using the method described in FIG. 5. If client XID is classified as Little Endian, then the upper 16 bit halfword of the XID 306 is the masked XID that is used to form the search key, and the lower 16 bit halfword of the XID 306 is considered to contain the LSBs which form the lower order halfword of the XID′ 310B. In an example aspect, if the XID is classified as Little Endian, then the Masked XID=Client XID & 0xFFFF0000, in which ‘FFFF’ are non-zero bits.

If the client XID is classified as Big Endian, then the lower 16 bit halfword of the XID 306 is the masked XID that is used to form the search key, and the upper 16 bit halfword of the XID 306 is considered the LSBs which form the lower order halfword of the XID′ 310B. In an example aspect, if the XID is classified as Big Endian, then the Masked XID=Client XID & 0x0000FFFF, in which ‘FFFF’ are non-zero bits.

If the client XID is classified as Unknown, then all 32 bits of the client XID is used as the Masked XID 314. In particular, all 32-bits of the front-end XID are folded into the XID LSBs to capture as much entropy as possible. In an example aspect, if the XID is classified as Unknown, the Masked XID=Client XID & 0xFFFFFFFF, in which ‘FFFF’ are non-zero bits. In this case the Masked XID is the unmodified client XID and the Client ID Table acts like a XID to XID′ mapping table. Since each XID having an Unknown endian'ness creates a new table entry, each XID having an Unknown endian'ness must map to a new XID′. Once the endian'ness is determined then that Masked XID 314 becomes a value like 0x12340000 and that Masked XID 314 value remains constant for any Client XID that begins with 0x1234. In other words, the low 16 bits of the client XID are masked off to form the Masked XID 314.

If the client XID is classified as Random, the Masked XID 314 is given zero bits. Additionally, all 32-bits of the XID 306 are folded into the LSB halfword of the XID to capture as much entropy as possible. Thus, the Client ID 306 includes the client's IP address 300, port 302, protocol 304, and a Masked XID 314 of zero. In an example aspect, if the XID is classified as Random, the Masked XID=Client XID & 0x00000000. The generator module 130 XORs the two halves of the client's XID to form the XID LSBs that are combined with the client ID to form the XID′.

The generator module 130 extracts the 32 bit source IP address 300, the 16 bit source port address 302, the 8 bit protocol information 304, and the Masked XID 306 to create the search key that can be used to locate the assigned Client ID 308 (Block 408). In particular, the generator module 130 maintains this information in the Client ID Table 214 in memory 128 or another storage location. The Client ID Table 214 is configured to store the IP address 300, source port 302, protocol 304, and Masked XID 314 as well as endian'ness XID classification 312 when the entry was created. The Client ID Table 214 stores all of the generated Client IDs 308 in a list to ensure that any new generated Client ID 308 does not have the same identifying bits of an already existing Client ID.

If the Client ID is not found using the search key (Block 410), the generator module 130 will generate a new Client ID using the tuple information and the Masked XID and add the new entry in the Client ID Table 214 (Block 412). The generator module 130 is able to incrementally generate 2¹⁶ or 65,536 Client IDs before a wraparound occurs (i.e. previously issued Client ID are reused). The Client IDs which are available for use are kept in the Client ID database 218. Accordingly, when a new client ID is to be generated for a client request, the next incremental Client ID will be allocated from the Client ID Database 218. In an example event that there is no free Client ID available in the Client ID Database 218, the XID Classifier 202 will identify and reuse the Least Recently Used (LRU) Client ID for the new Client ID entry. The allocated Client ID is then added to the Client ID Table 214 as a new entry in which the new entry is mapped to the tuple (IP address 300, port 302, protocol 304, and Masked XID 314) of the corresponding client request.

Referring back to Block 410, if a Client ID 308 corresponding to the search key is found, the Client ID 308 is retrieved from the Client ID Table (Block 422). As shown in Block 414, the XID′ synthesizer 212 of the generator module 130 thereafter extracts the 16 bit LSB halfword of the incoming client XID 306 and concatenates it with the generated Client ID 308 assigned by the client identifier 206 to generate a 32 bit synthesized XID′ (Block 414). The synthesized XID′ is transmitted in the NFS packet to the file server (Block 416). At least a portion of the Client ID Table 214 is replicated among one or more other cores within the file virtualization device or in one or more other file virtualization devices.

FIG. 5 illustrates a flow chart depicting at least a portion of the endian'ness determination process performed by the generator module in accordance with an aspect of the present disclosure. As shown in FIG. 3, the 32 bit client XID 306 can be divided into two halfwords 306A, 306B. In particular, the higher order halfword 306A is the 16 bit portion which changes infrequently among the client requests, and is therefore referred to herein as the most significant bits portion (MSB). In contrast, the lower order halfword 306B is the 16 bit portion of the client XID which changes with every client request.

As mentioned in FIG. 4, the generator module 130 generates the server side XID′ based on correctly identifying which of the MSB and LSB halfwords 306A, 306B are the higher and lower portions. However, not all of the client XIDs are received by the file virtualization device 110 in the same bit or byte order, in which this phenomenon is called endian'ness. For instance, some clients may transmit their XIDs in reverse order (e.g. LSB-MSB) in comparison to other clients' XIDs (e.g. MSB-LSB). For example, a client's XID is classified as having a Big Endian format if the first two bits received over the transmission channel are part of the MSB halfword. In contrast, a client's XID is classified as having a Little Endian format, if the last two bits received over the transmission channel are part of the MSB halfword. In an aspect, the endian'ness identifier 204 compares changed bits among upper halfwords and lower halfwords in their respective sets to determine their entropies.

Advantages are apparent in identifying the endian'ness of the Client XID. For instance, an example aspect may include Client XIDs in which both halfwords change on several or all requests. For this example scenario, the generator module 130 can ensure that wraparounds do not occur too frequently and/or can detect a potential malicious attack and take appropriate action.

As mentioned, clients are identified by the source IP address 300 in which the IP address 300 is used to efficiently lookup the XID classification for that client in the Classification Table 216. In particular, the endian'ness identifier 204 takes the client IP address 300 as well as any other relevant information about the client request and/or client and stores that information along with the Client XID 306 in the Classification Table 216 (Block 500). In an aspect, the endian'ness identifier 204 segments the 32 bit Client XID 306 into a set of upper halfwords and a set of lower halfwords (Block 500). It should be noted that the endian'ness identifier 204 takes the same order of bits for each XID and assigns them to the upper halfword set and lower halfword set for the set number of iterations.

As shown in FIG. 5, the XID Classifier 202 determines whether the set number of stored Client XIDs has been met (Block 502). This is because the endian'ness determination process is performed for a set number of iterations, whereby each iteration is associated with a different request for the same client having a different Client XID value. The number of iterations can be set by the administrator. In an example aspect, the number of iterations can be set to a value of 12, although other values are contemplated. This iterative process allows the XID classifier 202/endian'ness identifier 204 to reevaluate a client's endian'ness each time a new XID value is received for that client.

If the endian'ness identifier 204 in Block 502 determines that the set number of iterations has not yet been met, the endian'ness identifier 204 stores each 16 bit XID halfword in its respective set (e.g. upper or lower set) in the endian'ness table 216 for that iteration (Block 504). The halfwords are stored in the order of their arrival from the client. The process then repeats back to Block 500.

Once the set number of iterations have been reached (and the desired number of that client's upper and lower XID halfwords have been stored), the endian'ness identifier 204 examines the upper and lower halfwords, assigned in their sets, separately to determine whether the bit changes in the upper and lower XID halfwords are frequently changing or not frequently changing between halfwords over the iteration set (Block 506).

The endian'ness identifier 204 compares each of the stored upper halfwords with the next halfword in the set and records the number of changed halfwords. Based on the number of changed halfwords, the endian'ness identifier 204 designates the upper halfword as having a certain entropy at the time of performing the identification process (Block 508). Similarly, based on the number of changed halfwords, the endian'ness identifier 204 designates the lower halfword as having a certain entropy at the time of performing the identification process (Block 510).

In particular to an aspect, the endian'ness identifier 204 will designate a halfword as having high entropy if 8 or more adjacent halfwords were identified to have changed within the stored set of iterations. In contrast, the endian'ness identifier 204 will designate a halfword as having low entropy if less than 4 adjacent halfwords were identified to have changed within the stored set of iterations. Further, the endian'ness identifier 204 will designate that the halfword as having an unknown entropy if 4 to 7 adjacent halfwords had changed within the stored set. It should be noted that the bit value ranges for each of the above described entropies are examples and other bit value ranges are contemplated.

The XID Classifier 202 thereafter analyzes the entropies assigned to the representative upper and lower halfwords and classifies the endian'ness of the XID based on the combined entropy designations (Block 512). As shown in Block 514, the upper halfword has a low entropy and the lower halfword has a low entropy. Based on established classification logic, the XID classifier 202 classifies the endian'ness for the XID to be Unknown (Block 516). As shown in Block 518, the upper halfword has low entropy and the lower halfword has high entropy. Based on established classification logic, the XID classifier 202 classifies the endian'ness for the XID to be Big Endian (Block 520). As shown in Block 522, the upper halfword has high entropy and the lower halfword has low entropy. Based on established classification logic, the XID classifier 202 classifies the endian'ness for the XID to be Little Endian (Block 524).

In an aspect, if both of the upper and lower halfwords are determined by the endian'ness identifier 204 to have high entropy (Block 526), the XID classifier 202 classifies the endian'ness for the XID to be Random (Block 528). In the case that the endian'ness is Random, the MSB halfword is given a value of zero, thereby reducing the entropy of the client's XID to only 16 bits.

After the endian'ness of the XID 306 is determined, the classification for the client's XID 306 is stored in the Classification Table 216. The process thereafter returns to Block 404 in FIG. 4 as the XID classification information is determined.

In an aspect, the current XID classification may be used to select the MSB halfword of the XID as the Client ID Table search key. In this scenario, the classification is changed from Unknown to some other classification after enough XIDs have been received and analyzed by the endian'ness identifier 204. This should be noted as the previous Unknown classification was used by the XID classifier 202 to create the initial entry of the client in the Client ID Table 214.

In an aspect, one or more previous XID classifications are stored in the Classification Table 216. This is to address the potential scenario where the client's endian'ness changes from one endian'ness to another. The typical scenario for changing a client's endian'ness occurs when a client's endian'ness changes from Unknown to either Big, Little, or Random endian'ness after the client endian'ness is classified by the Endian'ness Classifier. If the previous endian'ness is different from the current endian'ness then that previous endian'ness must be used first when creating a search key to look up a client ID in the client ID table. If a search of the table fails to find a match then the current endian'ness must be used. As an optimization, the previous endian'ness can be set to the value of the current endian'ness after a timeout to avoid searching the client ID table twice. This allows the generator module 130 to find a Client ID that may have been stored in the Table 214 under a previous XID classification.

Having thus described the basic concepts, it will be rather apparent to those skilled in the art that the foregoing detailed disclosure is intended to be presented by way of example only, and is not limiting. Various alterations, improvements, and modifications will occur and are intended to those skilled in the art, though not expressly stated herein. The order that the measures are implemented may also be altered. These alterations, improvements, and modifications are intended to be suggested hereby, and are within the spirit and scope of the examples. Additionally, the recited order of processing elements or sequences, or the use of numbers, letters, or other designations therefore, is not intended to limit the processes to any order.

Claims

1. A method of generating a server side transaction ID (XID′), the method comprising: receiving, at a file virtualization device, a request from a client device to access a server, wherein the request includes a source port, source IP address, protocol information and a client generated transaction ID (XID);determining an endian'ness of the client;generating a Client ID for the client request, the Client ID utilizing the source port, source IP address, protocol information, and a Masked XID generated from the endian'ness determination and the XID;synthesizing a server side transaction ID (XID′) by combining the Client ID and a XID halfword containing least significant bits (LSB) identified from the endian'ness determination; andtransmitting the XID′ to the file server, wherein the XID′ is associated with the XID for the corresponding client request.

2. The method of claim 1, further comprising storing the generated Client ID as a new entry in a Client ID Table and replicating at least a portion of the Client ID Table to at least one other core in the file virtualization device.

3. The method of claim 1, further comprising storing the generated Client ID as a new entry in a Client ID Table and replicating at least a portion of the Client ID Table to at least one other core in another file virtualization device.

4. The method of claim 1, further comprising: classifying the endianness of the client as having a Big Endian format, a Little Endian format, an Unknown format or a Random format; andstoring the classified endian'ness in a Classification table.

5. The method of claim 4, wherein determining the endianness further comprises: receiving and storing a plurality of XIDs from the client for a set number of client requests;comparing a plurality of upper halfwords to identify a number of changed halfwords among adjacent upper halfwords in the set;designating an entropy classification to the upper halfwords based on the number of identified changed halfwords among upper halfwords;comparing a plurality of lower halfwords to identify a number of changed halfwords among adjacent lower halfwords in the set;designating an entropy classification to the lower halfwords based on the number of identified changed halfwords among lower halfwords; andassigning an XID classification to the client based on the entropy classifications designated to the upper and lower halfwords.

6. The method of claim 4, wherein the endian'ness is classified as having the Random format, the method further comprising: converting the first XID halfword to a zero bit value;generating the Client ID with the converted first XID halfword.

7. A non-transitory computer readable medium having stored thereon instructions for generating a server side transaction ID (XID′), the medium comprising machine executable code, which when executed by at least one machine, causes the machine to: receive a request from a client device to access a server, wherein the request includes a source port, source IP address, protocol information and a client generated transaction ID (XID);determine an endian'ness of the client;generate a Client ID unique to the client request, the Client ID utilizing the source port, source IP address, protocol information, and a Masked XID generated from the endian'ness determination and the XID;synthesize a server side transaction ID (XID′) by combining the Client ID and a XID halfword containing least significant bits (LSB) identified from the endian'ness determination; andtransmit the XID′ to the file server, wherein the XID′ is associated with the XID for the corresponding client request.

8. The computer readable medium of claim 7, wherein the machine is further configured to store the generated Client ID as a new entry in a Client ID Table and replicate at least a portion of the Client ID Table to at least one other machine in the file virtualization device.

9. The computer readable medium of claim 7, wherein the machine is further configured to store the generated Client ID as a new entry in a Client ID Table and replicate at least a portion of the Client ID Table to at least one other machine in another file virtualization device.

10. The computer readable medium of claim 7, wherein the machine is further configured to: classify the endianness of the client as having a Big Endian format, a Little Endian format, an Unknown format or a Random format; andstore the classified endian'ness in a Classification table.

11. The computer readable medium of claim 10, wherein the machine, when determining the endianness, is further configured to: receive and store a plurality of XIDs from the client for a set number of client requests;compare a plurality of upper halfwords to identify a number of changed halfwords among adjacent upper halfwords in the set;designate an entropy classification to the upper halfwords based on the number of identified changed halfwords among upper halfwords;compare a plurality of lower halfwords to identify a number of changed halfwords among adjacent lower halfwords in the set;designate an entropy classification to the lower halfwords based on the number of identified changed halfwords among lower halfwords; andassign an XID classification to the client based on the entropy classifications designated to the upper and lower halfwords.

12. The computer readable medium of claim 7, wherein the endian'ness is classified as having the Random format, the machine is further configured to convert the first XID halfword to a zero bit value and generate the Client ID with the converted first XID halfword.

13. A file virtualization device comprising: a network interface configured to receive client requests and transmit the client requests to one or more file servers;a memory configured to store non-transitory machine executable code including programming instructions for generating a server side transaction ID (XID′); anda processor configured execute the code, which when executed by the processor, causes the processor to: receive a request from a client device to access a server, wherein the request includes a source port, source IP address, protocol information and a client generated transaction ID (XID);determine an endian'ness of the client;generate a Client ID unique to the client request, the Client ID utilizing the source port, source IP address, protocol information, and a Masked XID generated from the endian'ness determination and the XID;synthesize a server side transaction ID (XID′) by combining the Client ID and a XID halfword containing least significant bits (LSB) identified from the endian'ness determination; andtransmit the XID′ to the file server, wherein the XID′ is associated with the XID for the corresponding client request.

14. The file virtualization device of claim 13, wherein the processor is further configured to store the generated Client ID as a new entry in a Client ID Table and replicate at least a portion of the Client ID Table to at least one other processor in the file virtualization device.

15. The file virtualization device of claim 13, wherein the processor is further configured to store the generated Client ID as a new entry in a Client ID Table and replicate at least a portion of the Client ID Table to at least one processor in another file virtualization device.

16. The file virtualization device of claim 13, wherein the processor is further configured to: classify the endianness of the client as having a Big Endian format, a Little Endian format, an Unknown format or a Random format; andstore the classified endianness in a Classification table in the memory.

17. The file virtualization device of claim 16, wherein the processor, when determining the endianness, is further configured to: receive and store a plurality of XIDs from the client for a set number of client requests;compare a plurality of upper halfwords to identify a number of changed halfwords among adjacent upper halfwords in the set;designate an entropy classification to the upper halfwords based on the number of identified changed halfwords among upper halfwords;compare a plurality of lower halfwords to identify a number of changed halfwords among adjacent lower halfwords in the set;designate an entropy classification to the lower halfwords based on the number of identified changed halfwords among lower halfwords; andassign an XID classification to the client based on the entropy classifications designated to the upper and lower halfwords.

18. The file virtualization device of claim 13, wherein the endian'ness is classified as having the Random format, the processor is further configured to convert the first XID halfword to a zero bit value and generate the Client ID with the converted first XID halfword.