Electrical computers and digital processing systems: multicomput – Computer-to-computer session/connection establishing – Network resources access controlling
Reexamination Certificate
1998-08-26
2001-04-03
Lim, Krisna (Department: 2153)
Electrical computers and digital processing systems: multicomput
Computer-to-computer session/connection establishing
Network resources access controlling
C709S219000, C709S203000
Reexamination Certificate
active
06212565
ABSTRACT:
BACKGROUND OF THE INVENTION
This invention relates generally to client-server computer networks. More particularly, this invention relates to a general technique for improving the performance of a proxy server array during persistent network access.
The recent publicity and emphasis on the information superhighway has increased the awareness and acceptance of the Internet as a mass communications medium. Until recently, “cruising or surfing” the Internet was a disorienting, even a frustrating experience, something like trying to navigate without maps. The World Wide Web has made it easier to access the array of resources available on the Internet. Resources, such as web servers, ftp servers, and telnet servers, provide the user with the ability to easily find the data content or information he wants simply and easily.
The volume of World Wide Web traffic on the Internet is staggering but a significant fraction of this traffic is redundant. That is, a large number of users request the same data from the same resource, at around the same time. As a result, a significant percentage of a corporation's network infrastructure is carrying and servicing the repeated requests for same data content, day after day.
To manage this growing demand for access to the Internet and to reduce network communications costs, several system-level applications have been developed to extend caching to the client/server network environment. Two current approaches are proxy server arrays and network caches. Proxy server arrays are network server-based applications that are placed between a client application, such as a web browser, and a resource, such as a Web server. Initially proxy servers were designed to deal with problems caused by firewall issues in corporate web access. Eventually, proxy servers were also recognized as being an ideal environment to cache web data and to improve system performance, as well as to reduce the load on the network and on the servers.
In most World Wide Web based client/server applications, the proxy server receives a request to access a specific resource from a client system. The proxy server examines the request to determine if it can service the request itself. If the particular web page is stored in its cache, the proxy server will retrieve the web page and forward it to client that made the request. If not, the proxy server sends a request to the desired resource site specified by the Uniform Resource Locator (URL). The URL acts as the address of the resource and as such is unique throughout the Internet. The proxy server retrieves the web page from the resource specified by the URL address and transfers the web page to the client. In addition, the proxy server stores the web page in its cache for future use by other client systems.
It is becoming common for networks (such as intranets or the Internet) to include a plurality of proxy servers accessible by one or more clients. Requests from clients for various pages stored on servers in the network are routed through the plurality of proxy servers, which cache pages whenever possible.
Certain systems allow the client to specify to which proxy server it will send its request. Thus, a certain client can send requests to more than one proxy server. While caching performed by these proxy servers improves the overall performance of a network, having more than one proxy server store the same web pages in its cache is inefficient.
The World Wide Web operates using the http protocol. An initial version of http (http 1.0) required a separate TCP connection for each transfer of information between a client and a proxy or a server. Subsequently, newer versions of the http protocol have reduced the need to establish a separate connection for each request. For example, version 1.1 of the http protocol includes “persistent http,” where multiple http transfers can use the same connection, i.e., the same proxy server without having to establish a new connection each time. Persistent http connections are described in, for example, Request For Comments (RFC) 2068, “Hypertext Transfer Protocol—HTTP/1.1,” January 1997, available from the Internet Engineering Taskforce (IETF).
A problem arises, however, if a client can send its. requests to more than one proxy server, since clients generally do not know whether a proxy server has a persistent connection to the web server or not. Furthermore, if one or more clients can access one or more proxy servers, the clients do not know which of the proxy servers have already received requests for a particular service from another client (and may therefore have a persistent connection open).
In some conventional systems, the Internet Cache Protocol (ICP) is used to determine and select the most applicable location from which to retrieve a requested web page. In ICP, one proxy server establishes a “working” relationship with other proxy servers. Proxy servers designated as parents are on one level while child proxy servers are on lower level(s). The terms “neighbor” and “peer” refer to either a parent or a sibling which are a single “cache reference” away.
In general, in ICP the flow of a client request is up through the hierarchy of proxy servers. If a proxy server does not have a client's requested web page, it requests that a special proxy server, called an arbitrator, query the other proxy servers to see if they have the desired web page. If any of these proxy servers has the requested web page in its cache, then the inquiring proxy server enters a demand for the web page. The cached web page is either forwarded directly to the client or to the original proxy server for transfer to the client. If none of the proxy servers have the web page in their cache, the proxy server must forward the request either to a parent or back to the origin proxy server for service. Thus, if a successful request or “hit” occurs, it may fetch the web page from a peer proxy server or the requested wed page is received from a parent but if the request is unsuccessful or “missed,” it must be passed to a parent server for service. The role of a parent is complete the transaction and service the request. If necessary, a parent proxy server will open a resource directly to service a client's request.
There are several problems that arise with the ICP approach. For example, the arbitrating proxy server may be overrun with requests or the network path between proxy servers may be congested. In addition, the additional hierarchy introduces extra delays for the clients requesting uncached data.
Other conventional client/server systems, such as the CARP (Cache Array Routing Protocol), available from Microsoft, Inc. of Redmond, Washington access a variety of proxy servers to retrieve pages stored on a single server. CARP, for example, uses a deterministic hashing function in the client to allocate page accessing and caching among a variety of proxy servers. By accessing a variety of proxy servers for various pages stored on a server, the CARP system aims to achieve load balancing between the proxy servers. Unfortunately, such a system has the disadvantage that it requires a hashing function that distributes the page accesses equally among the proxy servers. If the hashing function is poorly chosen or if the URL names lead to unbalanced distribution of the URLs to various proxies, it will negatively affect the load balancing between proxy servers. Thus, CARP's use of a deterministic hashing function to distribute requests to proxies does not always achieve a good distribution of requests among proxies. Lack of good distribution leads to inefficient usage of proxies. More importantly, with CARP, two URLs for the same Web server are likely to be sent to two different proxy servers, thereby undoing the benefits of persistent connections between a particular proxy and a server.
SUMMARY OF THE INVENTION
A described embodiment of the present invention provides a method and apparatus that ensures that requests for pages for a particular domain name are routed to the same proxy server by all of a plurality of client
Lim Krisna
Squire Sanders & Dempsey L.L.P.
Sun Microsystems Inc.
LandOfFree
Apparatus and method for improving performance of proxy... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Apparatus and method for improving performance of proxy..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Apparatus and method for improving performance of proxy... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2463735