Electrical computers and digital processing systems: multicomput – Computer network managing – Network resource allocating
Reexamination Certificate
1998-03-17
2001-03-20
Rinehart, Mark (Department: 2756)
Electrical computers and digital processing systems: multicomput
Computer network managing
Network resource allocating
C709S216000, C709S223000
Reexamination Certificate
active
06205481
ABSTRACT:
BACKGROUND
Computer networks, such as the Internet, private intranets, extranets, and virtual private networks, are increasingly being used for a variety of endeavors including the storage and retrieval of information, communication, electronic commerce, entertainment, and other applications. In these networks certain computers, known as servers, are used to store and supply information. One type of server, known as a host or home server, provides access to information such as data or programs stored in various computer file formats but generically referred to herein as a “document”. While in the Internet the documents are typically primarily composed of text and graphics, each such document can actually be a highly formatted computer file containing data structures that are a repository for a variety of information including text, tables, graphic images, sounds, motion pictures, animations, computer program code, and/or many other types of digitized information.
Other computers in the network, known as clients, allow a user to access a document by requesting that a copy be sent by the home server over the network to the client. In order for a client to obtain information from a home server, each document typically has an address by which it can be referenced. For example, in the context of the Internet and within the communication protocol known as Hyper Text Transfer Protocol (HTTP), the address is typically an alphanumeric string, known as a Uniform Resource Locator (URL), that specifies (a) an address of the home server from which to obtain the information in the form of a name or a numerical address, and (b) a local information text string that identifies the information requested by the client, which may be a file name, a search request, or other identification.
After the user specifies a URL to the client computer, the address portion of the URL is sent over the network to a naming service such as the Domain Name Service (DNS) in order to obtain instructions for how to establish a connection with the correct home server. Once the connection with the server is established, the client can then retrieve the desired document by passing the local information text string over the network directly to the home server. The server then retrieves the document from its local disk or memory storage and transmits the document over the network to the client. The network connection between the home server and the client is then terminated.
Computer and network industry analysts and experts are presently quite concerned that traffic on the Internet is becoming so heavy that the very nature of the way in which it is possible to use the Internet may change. In particular, many individuals now believe that the Internet is intolerably slow and is no longer a reliable entity for the exchange of information in a timely fashion.
The present bottlenecks are no doubt the result of exponential increases in the number of users as well as in the number of complex documents such as multimedia files being sent. It might appear that the answer is simply to add more bandwidth to the physical connections between servers and clients. This will come, however, only at the expense of installing high bandwidth interconnection hardware, such as coaxial or fiber optic cable and associated modems and the like, into homes and neighborhoods around the world.
Furthermore, added bandwidth by itself perhaps would not guarantee that performance would improve. In particular, large multimedia files such as for video entertainment would still potentially displace higher priority types of data, such as corporate E-mails. Unfortunately, bandwidth allocation schemes are difficult to implement, short of modifying existing network communication protocols. The communication technology used on the Internet, called TCP/IP, is a simple, elegant protocol that allows people running many different types of computers such as Apple Macintoshes, IBM-compatible PCs, and UNIX workstations to share data. While there are ambitious proposals to extend the TCP/IP protocol so that the address can include information about packet content, these proposals are technologically complex and would require coordination between operators of many thousands of computer networks. To expect that modifications will be made to existing TCP/IP protocols is thus perhaps unrealistic.
An approach taken by some has been to recognize that the rapidly growing use of the Internet will continue to outstrip server capacity as well as the bandwidth capacity of the communication media. These schemes begin with the premise that the basic client-server model (where clients connect directly to home servers) is wasteful of resources, especially for information which needs to be distributed widely from a single home server to many clients. There are indeed, many examples of where Internet servers have simply failed because of their inability to cope with the unexpected demand placed upon them.
To alleviate the demand on home servers, large central document caches may be used. Caches are an attempt to reduce the waste of repeated requests for the same document from many clients to a particular home server. By intercepting parallel requests, a cache can be used to serve copies of the same document to multiple client locations.
From the client's point of view, the interaction with a cache typically occurs in a manner which is transparent to the user, but which is slightly different from a network messaging standpoint. The difference is that when the address portion of the request is submitted to the Domain Name Service (DNS) to look up the information needed to connect to the home server, the DNS has been programmed to return the address of a cache instead of the actual original home server.
Alternatively, a server node, acting as a proxy for the client, may issue probe messages to search for a cache copy. Once a cache copy is found at a particular node in the network, the request is then forwarded to that node. For example, under the auspices of the National Science Foundation, document caches have been placed at various locations in the United States in order to eliminate bottlenecks at cross-oceanic network connections. Generally, certain of these caches located on the West Coast handle requests for documents from the Asia-Pacific and South American countries, and a number of those located on the East Coast handle requests for documents from Europe. Other of these national caches handle requests for popular documents located throughout the United States.
However, such caching techniques do not necessarily or even typically achieve optimum distribution of document request loading. In particular, in order for the caches to be most effective, the DNS name service or other message routing mechanism must be appropriately modified to intercept requests for documents for which the expected popularity is high. The introduction of cache copies thus increases the communication overhead of name resolution, because of the need to locate the transient copies. The name service must register these copies as they come into existence, disseminate this information to distribute demand for the documents, and ensure the timely removal of records for deleted cache copies. Often times, the cache lookup order is fixed, and/or changes in document distribution must be implemented by human intervention.
Unfortunately, frequent and pronounced changes in request patterns can force the identity, location, and even the number, of cache copies to be highly transient. The resulting need for updating of cache directories means that they cannot typically be replicated efficiently on a large scale, which can thus turn the name service itself into a bottleneck.
Another possible approach to implementing caches is to change the client/server interaction protocol so that clients proactively identify suitable cache copies using a fully distributed protocol, for example, by issuing probes in randomized directions. Aside from the complexity of modifying existing protocols and message cost introduced by such
Heddaya Abdelsalam A.
Mirdad Sulaiman A.
Yates David J.
Yates Ian C.
Cardone Jason D.
Hamilton Brook Smith & Reynolds P.C.
InfoLibria, Inc.
Rinehart Mark
LandOfFree
Protocol for distributing fresh content among networked... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Protocol for distributing fresh content among networked..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Protocol for distributing fresh content among networked... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2490488