Patent Application
20140052822
The present invention generally relates, in a first aspect, to a system for Content Distribution Internetworking, and more particularly to a system comprising a global origin server that coordinates the formation of a global network by connecting to local origin servers of a plurality of Content Delivery Networks, or CDNs.
A second aspect of the invention relates to a method comprising using a global origin server for coordinating the formation of a global network by its connection to CDN local origin servers.
The terminology and definitions that might be useful to understand the present invention are included.
PoP: A point-of-presence is an artificial demarcation or interface point between two communication entities. It is an access point to the Internet that houses servers, switches, routers and call aggregators. ISPs typically have multiple PoPs.
Content Delivery Network (CDN): This refers to a system of nodes (or computers) that contain copies of customer content that is stored and placed at various points in a network (or public Internet). When content is replicated at various points in the network, bandwidth is better utilized throughout the network and users have faster access times to content. This way, the origin server that holds the original copy of the content is not a bottleneck.
ISP DNS Resolver: Residential users connect to an ISP. Any request to resolve an address is sent to a DNS resolver maintained by the ISP. The ISP DNS resolver will send the DNS request to one or more DNS servers within the ISP's administrative domain.
URL: Simply put, Uniform Resource Locator (URL) is the address of a web page on the world-wide web. No two URLs are unique. If they are identical, they point to the same resource.
URL (or HTTP) Redirection: URL redirection is also known as URL forwarding. A page may need redirection if (1) its domain name changed, (2) creating meaningful aliases for long or frequently changing URLs (3) spell errors from the user when typing a domain name (4) manipulating visitors etc. For the purpose of the present invention, a typical redirection service is one that redirects users to the desired content. A redirection link can be used as a permanent address for content that frequently changes hosts (much like DNS).
Bucket: A bucket is a logical container for a customer that holds the CDN customer's content. A bucket either makes a link between origin server URL and CDN URL or it may contain the content itself (that is uploaded into the bucket at the entry point). An end point will replicate files from the origin server to files in the bucket. Each file in a bucket may be mapped to exactly one file in the origin server. A bucket has several attributes associated with it—time from and time until the content is valid, geo-blocking of content, etc. Mechanisms are also in place to ensure that new versions of the content at the origin server get pushed to the bucket at the end points and old versions are removed.
A customer may have as many buckets as she wants. A bucket is really a directory that contains content files. A bucket may contain sub-directories and content files within each of those sub-directories.
Geo-location: It is the identification of real-world geographic location of an Internet connected device. The device may be a computer, mobile device or an appliance that allows for connection to the Internet for an end user. The IP-address geo-location data can include information such as country, region, city, zip code, latitude/longitude of a user.
Operating Business (OB): An OB is an arbitrary geographic area in which the provider of the CDN service is installed. An OB may operate in more than one region. A region is an arbitrary geographic area and may represent a country, or part of a country or even a set of countries. An OB may consist of more than one region. An OB may be composed of one or more ISPs. Each region in an OB is composed of exactly one region DNS server and tracker. An OB has exactly one instance of Topology Server.
Partition ID: It is a global mapping of IP address prefixes into integers. This is a one-to-one mapping. So, no two OBs can have the same PID in its domain.
Default Operating Business: OB0 is defined as a default operating business where the TLD DNS server resides. All IP prefixes that are not part of other regions default to this region. By design, the default OB0 is designed to have just one region that may be used to serve content to such IP prefixes (that are not part of any other OB).
Consistent Hashing: This method provides hash-table functionality in such a way that adding or removing a slot does not significantly alter the mapping of keys to slots. Consistent hashing is a way of distributing requests among a large and changing population of web servers. The addition of removal of a web server does not significantly alter the load on the other servers.
Overlay Network: An overlay network is a computer network that is built on top of another network. Nodes in an overlay network are connected by virtual/logical links. Each logical link may consist of a path that is made up of multiple physical links in the underlying network.
Content Distribution Internetworking (CDI): Content Distribution Internetworking is the ability to connect many independently administered CDNs to form a federation of CDNs. This allows a CDN to extend beyond its administrative domain to increase the reach of content.
Transport Control Protocol (TCP): Transport Control Protocol is one of the core protocols of the Internet Protocols. TCP is responsible of an ordered and reliable delivery of data stream between two network hosts.
Next, each component of the CDN service provider's sub-system is described. The infrastructure consists of Origin Servers, Trackers, End Points and Entry Point.
A CDN customer has two options for uploading content. The customer can either upload files into the bucket or give URLs of the content files that reside at the CDN customer's website. Once content is downloaded by the CDN infrastructure, the files are moved to another directory for post-processing. The post-processing steps involve checking the files for consistency and any errors. Only then is the downloaded file moved to the origin server. The origin server contains the master copy of the data.
In addition, the end points maintain a geo-IP database and table of a list of datacenters.
When an end user makes a request for content, the tracker uses the statistical information at its disposal to determine if (1) the content can be served to the requesting end user and if so, (2) determines the closest end point and one with the least load to serve an end user. Thus, the tracker acts as a load-balancer for the CDN infrastructure.
CDNs typically operate as single global entities; have multiple points of presence and in locations that are geographically far apart. As a result, a CDN may have multiple replicas of each piece of content being hosted. The definition of origin servers for CDN providers is generalised as follows: (1) an entity (like a server) that resides in the administrative domain of CDN customer. Content is replicated at end point(s) after the first request for content by an end user. (2) All origin servers are under the administrative control of the same CDN provider and contain content from CDN customers. These servers contain the master copy of the content and replicate it at the end point(s). Adding additional storage capacity at the CDN service provider is merely a case of adding additional origin servers under its administrative control.
There are many different designs of CDNs. For example, [2] uses a hierarchy of DNS [1] servers together with geo-location information to find a content server that is closest to a requesting end user to serve content. Other solutions like [3] rely on a small number of large datacenters or [6] a large number of small datacenters connected by a well-provisioned private network to first identify a datacenter that is closest to the requesting end user. Once the datacenter is identified, an end point in the datacenter is identified to deliver the content. Only at this final step the CDN connects to the public Internet. Further, [5] relies on extensive storage and caching infrastructure at the major peering points. Amazon [4] provides CDN service using Amazon Cloudfront together with its simple storage service allowing end users to get data from various edge locations of the Internet that Amazon peers with. Of these, only [2] connects to the public Internet and provides a global CDN service and falls under a single administrative domain. The other CDN designs fall under different administrative domains.
Regardless of administrative control, content originally stored in origin servers is replicated at end points for distribution to requesting end users. The origin servers in the CDN service provider always contain the master copy of the content obtained from a CDN customer. The CDN service is designed to work as a global CDN.
There are several reasons why a number of OBs may want to remain and operate independently and yet come together to form a global CDN.
Each OB may be an independent operating business in one country and hence, may want complete control over all of the infrastructure elements of the CDN. The OBs may yet be a part of a single global entity.
Given that an OB operates in one country, it is easier for an OB to establish a deep relationship with the content providers in that country and operate within its laws.
By allowing the content providers in the OB deciding if their content is visible only within that OB or may be shown in other OBs (or even globally), the OBs can give to content providers all the control over their content they desire.
An OB may not want to expose the detailed topological information about its network to other OBs and yet be part of the global CDN to share content and expand content reach.
The presence of several CDNs, each operating its own naming convention (i.e., having its own CDN URLs) and its own DNS infrastructure to identify the requested content, makes it impossible to extend the scale and reach of CDNs. Several proposals under Content Delivery Internetworking (CDI) have been proposed with a goal of peering CDNs. The key goals of peering CDNs are (a) increase capacity, (b) improve delivery points in the network, (c) expand reach of content to a wider customer base, (d) provide better fault-tolerance and (e) achieve better economies of scale and (f) a better overall user experience.
In [8] the authors introduce definitions for Content Distribution Internetworking (CDI) or CDN peering and define the terminology. They envision Content Internetworking as consisting of Accounting Internetworking, Content Internetworking Gateways, Request-Routing Internetworking. The authors discuss many known request-routing mechanisms in [9]. They discuss DNS based request-routing schemes including multi-level resolution, anycast and object encoding in DNS. In addition, they discuss Transport and Application layer request-routing schemes including URL rewriting and HTTP redirection. In [10], the authors present various Content Internetworking scenarios. They propose content internetworking gateways to route requests for content and accounting under a variety of scenarios, with particular emphasis on accounting internetworking.
A technique called CDN brokering is defined by the Content Alliance. Here, CDN brokering is the ability of a CDN to redirect clients dynamically between two or more CDNs. One such realization is the DNS-based system, Intelligent Domain Name Server (IDNS). The IDNS [7] is a DNS broker that uses a probability distribution in the region in which the CDNs operates to determine which CDN will serve the request. However, this requires that CDNs hold the content names and end points from which they are served in caches. Content names are identified from the HTTP request of the headers. While this works for HTTP downloads, it cannot work for live streaming of content.
Most proposals under CDI are very broad and offer only guidelines and little by way of concrete protocols for implementation. Some problems have been detected with the existing proposals:
Overall, the standardization efforts in CDI are poor with little or no activity for better part of a decade.
Here, we describe the notation that is used in the rest of the invention:
OBi: Any arbitrary operating business i may be denoted by OBi. Similarly, we have denote OBk, OBl, OBm for operating businesses k, l and m. Here, i, k, l, m etc. are all integers.
OSi: Any arbitrary operating business i (OBi) has an origin server denoted by OSi.
OB0: This is used to denote the default operating business 0.
OSG: This is used to denote the global origin server.
OS{j}: This is used to denote a list of origin servers that may contain a requested content. If origin servers at j, k, l and m contain the requested content, OS{j}=(OSj, OSk, OSl, OSm). Here, {j}=(j, k, l, m).
It is necessary to provide an alternative to the state of the art, which covers the gaps found therein, particularly those related to the above indicated problems related to the known CDI proposals.
To that end, the present invention concerns, in a first aspect, to a system for Content Distribution Internetworking, or CDI, comprising a plurality of Content Delivery Networks, or CDN, each defining an operating business having its respective local origin server, and computing means for performing the interconnection of said plurality of CDNs.
Different from other known CDI proposals, in the one provided by the system of the first aspect of the invention, said computing means comprise a global origin server that coordinates the formation of a global network by connecting to some (or all) of said local origin servers at the OBs.
Other embodiments of the system of the first aspect of the invention are described according to appended claims 2 through 14, and in a subsequent section related to the detailed description of several embodiments.
By the system of the invention, a CDI can run on a single hierarchy of DNS servers and may combine an arbitrary number of CDNs while connecting to the public Internet.
A second aspect of the invention provides a method for Content Distribution Internetworking. The method comprises performing the interconnection to a plurality of CDNs, each defining an operating business with its own local origin server.
Different from other known methods, the CDI provided by the second aspect of the invention comprises, using a global origin server for coordinating the formation of a global network by connecting said global origin server to some (or all) of said local origin servers.
Other embodiments of the method of the second aspect of the invention are described according to appended claims 15 to 18, and in a subsequent section related to the detailed description of several embodiments.
The embodiments described for the system of the first aspect of the invention are also valid for the method of the second aspect, as for the functions the different elements of the system perform.
The previous and other advantages and features will be more fully understood from the following detailed description of embodiments, with reference to the attached drawings, which must be considered in an illustrative and non-limiting manner, in which:
Next, a description of the invention for several embodiments will be done, referring to both, the system and the method of the invention.
This invention shows how to combine many independently operating but similar CDNs to come together to seamlessly form a global CDN network. As part of this, the traditional role of an origin server from a stand-alone entity in a CDN that sends content to an end point for further distribution is extended to one that is content aware. In a global CDN network that consists of a collection of Operating Businesses (OBs), each OB has its own CDN infrastructure, so the origin servers from all the OBs form an overlay network of origin servers that share content for replication at end points. Only the end points within OBs are responsible of distributing content to requesting end users.
The key to the CDI architecture is (i) the presence of a global origin server, OSG. This OSG maintains the meta-data of all the buckets that are in each OS of all the OBs in the CDNs. (ii) All the OSs and the OSG come together to form a global overlay of origin servers.
If the requested content is not in the OSi of the requesting CDN i, the location of the content is determined from OSG. Subsequently, OSi downloads the content from the set of origin servers OS{j} of the CDNs {j, |j|≧1} in which the content exists. The OSi then serves the content to the requesting end user.
Thus, all the CDNs that operate in independent administrative domains come together to form a seamless global CDN.
Next, the details of the architecture of the system of the invention, for internetworking independently operating CDNs to come together to form a seamless global CDN will be presented.
As seen from
Each OBi has a publishing point that the CDN customers within the OBi can use to publish their content on the CDN. The customers may use two techniques to upload their content to the CDN. (1) upload content into their buckets at the publishing point or (2) provide the publishing point with the address of the web-server to download content. After post-processing, the customer bucket with content is available at the origin server where it is ready to be served to the end points. At the bucket and file level, the customer may determine the geographic region where the content may be shown. The geographic region is mapped to OBs.
In addition to an Origin Server at each OB, there is also a global origin server OSG. The global origin server keeps an open TCP connection with each of the OSs in the other OBs.
There is a single top-level domain (TLD) server for the domain t-cdn.net. The DNS at each OB resolves all the IP addresses in the second-level domain for the OB (it is the authoritative server in the DNS sub-zone of the OB).
The CDN service provider consists of independent Operating Businesses (OBs) that all together form a global CDN. A few key aspects of the global CDN are:
The tracker at each OBi maintains a consistent hash ring for all content that resides in the OBi's infrastructure.
Building a seamless global CDN out of a set of individual CDNs with their individual administrative domains consists of two main steps: In the first step the DNS resolution is done to identify an end point that will serve content to the requesting end user. In the second step, the end point will get the content from the network of origin servers and serve content. Next, the DNS resolution will be discussed:
When an end user, say in OBi requests content b87.t-cdn.net/87/video01.flv, the ISP DNS resolver first resolves the TLD t-cdn.net. The TLD DNS server resolves the subzone of OBi using its geo-IP database. The ISP DNS resolver then queries the authoritative DNS sever at OBi that forwards the request to an end point in OBi.
The end point first checks to see if the requested bucket and content is part of OSi. If it is not, the end point checks with the origin server if the content is part of OSG. If it is not part of OSG either, an error is returned to the end user. If the content is either in OSi or OSG, the end point determines the closest datacenter to the requesting user's ISP DNS (called a partition ID, in this case, say 34). The end point also calculates the consistent hash of the requested URL and returns HTTP 302, moved location b87-p34-abf8.t-cdn.net to the end user. The end user now sends an address resolution request for b87-p34-abf8.t-cdn.net to the OBi DNS server. Next, the DNS server forwards the request to the tracker serving the OBi. The tracker performs a consistent hash of the received URL and identifies the end point that should serve the requested content.
If the OSi in OBi has the requested content, it is downloaded to the end point that serves the content to the end user (label d in
A logical network of origin servers (OS) is build with the global origin server (OSG) as the head. The global origin server, OSG keeps an open connection with the origin servers across all the OBs. It uses this open connection to synchronize buckets with the origin servers across all the OBs.
As shown in
The OSG may receive content from the origin servers in other OBs by one of the following methods: (1) The OSG picks one of the origin servers in the other OBs that has the content, and forwards its address to the OS of the requesting OB. The OS of the requesting OB downloads the content and forwards it to the requesting end user. (2) OSG forwards a list of OS servers in all the OBs that have the requested content. The requesting OS may use (a) a P2P protocol to download the content from OSs of OBs that have the content and forward the content to the requesting end user or (b) get it from the one of the origin servers among the list returned by OSj.
In
Next the OSG design for an embodiment of the system of the first aspect of the invention is discussed. The local origin server OSi at each OBi in CDN service provider's infrastructure contains a master copy of the data uploaded by all the CDN customers in that OB. The OSG on the other hand does not contain a master copy of any data. It is the entity that coordinates the formation of a global CDN by connecting all of the disjoint OSi.
Any end point that does not have a copy of the data can request it from the origin server. A CDN customer does not have access to the origin server. The CDN service provider's infrastructure moves content from the entry point to the origin server after performing sanity-checks on the downloaded data.
In all, seven messages that must be supported are defined as part of the wire protocol (this is oblivious to whether HTTP or message passing protocol is used). The OSG supports the following messages:
(1) Get bucket list (OS_getBucketList). This message is sent by the OSG when a TCP connection is first established between OSi and OSG.
(2) Received bucket update (OS_receivedBucketUpdate). This message is sent to the OSi that sent the updated bucket and file list.
(3) Origin Server List for requested file (OS_IistForFile). This message is sent to OSi that requested the file in response to OS_getFile message.
(4) Abort connection (OS_connectionAbort). The OSG may abort connections with any (or all) OSi if the server needs to undergo maintenance or it detects that it has not received any update. This will force the OSi to open a new connection.
The OSG has three modules: connection module, neighbourhood module and a buckets module. We describe the function of each of the modules below.
When a request for content comes to OSG, the connection manager receives the request. The bucket manager identifies the OS{i} that have the requested file. The neighbourhood manager ranks the OS{i} in order from least loaded to the most loaded origin servers. The list is then sent to the requesting OSi by the connection module.
The OSi at an OBi still stores a master copy of all the content at OBi. In addition, to get content from other OSs, it needs to support the following messages on a wire protocol:
(5) Bucket list (OS_bucketList). This message, along with a list of buckets (and files in the buckets) is sent to the OSG in response to OS_getBucketList.
(6) Update bucket list (OS_updateBucketList). This message is sent to the OSG along with a list of updates to the files and buckets since the last update. Statistics related to the OSi are piggybacked to the bucket updates sent to the OSG.
(7) Get file (OS_getFile). An OSi gets a list of IP addresses for the requested content. It gets the IP addresses in order of the least busy Origin Server, OSj first. The OSi then connects to OSj and gets the requested content.
Each of the OSi implements three modules, a connection module, a statistics module and a bucket module.
If a requesting end user is not in any of the administrative domains of the OBs and if the requested content may be shown in the geography of the end user, the request for content is forwarded to the closest OBl. The tracker at OBl then determines the end point in the OBl that may be best suited to download content to the requesting end user.
If an end point in OBl that is assigned by the tracker to serve the requesting end user does not have the requested content, the OSl first gets the address of the OS{j} from OSG that have the requested content. Subsequently, the OSl downloads the content and sends it to the assigned end point in the same OBl that will serve the content to the end user.
When a New OB Comes Online:
When a new OB (call it OBn) comes online, the OSn at OBn does the following:
As part of its initialization, the OSn is assigned the IP address of the Global origin server, OSG. When the origin server OSn comes up, it opens a TCP connection with the OSG.
The global origin server OSG receives the following information from each of the OSi for all OBi that come together to form a global inter-distribution network. Periodically, each of the OSi reports the number of outbound bytes, number of inbound bytes, number of active connections and CPU utilization. The OSG uses this information to infer the load at the OS. The OSi send this information with updates to buckets and files in each bucket.
The sequence diagram in
When an OB comes online (say OBi ) the OSi in the OBi establishes a TCP connection with OSG. Next, OSG sends an OS_getBucketList message to the OBi. In response, the OSi sends a list of all buckets and files in each of the buckets to OSG. Following the initial message exchange, the OSi updates the OSG with changes in file/bucket list every couple of minutes via the OS_updateBucketList. This message also contains the statistics information at the OSi. In response, the OSG acknowledges the receipt of the information from OBi via an OS_receivedBucketUpdate response.
When an OB fails (the connection between OSi and OSG goes down), the bucket manager at OSG removes all buckets and files associated with OSi. The neighbourhood manager at OSG removes OSi from its list of neighbours and all buckets associated with OSi. When an OSi in OBi comes online, it attempts to open a connection with OSG.
Typically, OSG is mirrored for redundancy. However, if OSG must go down for maintenance (or for any other unexpected reason) and comes back up, it starts the interconnection between the disjoint CDNs. Once each of the OSi connects with the OSG, they respond to the OS_getBucketList request with a list of buckets and files in the buckets. Subsequently, each of the OSi periodically send the updates to files and buckets to the OSG.
The global origin server OSG has the following functionalities:
Thus, the OSG knows the location of every piece of content in each OB.
When an OSi in an OBi looks for a piece of content, it gets an ordered list of OB{i} from whom it can request the content.
The key advantages of this invention are:
A person skilled in the art could introduce changes and modifications in the embodiments described without departing from the scope of the invention as it is defined in the attached claims.
| Number | Date | Country | Kind |
|---|---|---|---|
| P201130756 | May 2011 | ES | national |
| Filing Document | Filing Date | Country | Kind | 371c Date |
|---|---|---|---|---|
| PCT/EP12/58350 | 5/7/2012 | WO | 00 | 11/6/2013 |
