Electrical computers and digital processing systems: multicomput – Computer network managing – Network resource allocating
Reexamination Certificate
2002-04-30
2004-08-03
Eng, David Y. (Department: 2155)
Electrical computers and digital processing systems: multicomput
Computer network managing
Network resource allocating
C709S223000
Reexamination Certificate
active
06772211
ABSTRACT:
FIELD OF THE INVENTION
This invention relates to client-server computing systems operating under the Internet Protocol, and more particularly to a switch and methods for switching client IP packets in a prescribed order to a server selected from a group of servers.
BACKGROUND OF THE INVENTION
With the advent of the World Wide Web (“WWW”), a universal client-server computing platform has emerged on the Internet. A very large number of web-servers on the Internet are serving web applications, which interact with web browsers acting as clients. A web application is typically organized into a hierarchy of webpages, scripted in Hypertext Markup Language (“HTML”) and/or Extensible Markup Language (“XML”.) It operates under the HyperText Transfer Protocol (“HTTP”.) A web application could itself be a suite of applications, including access and manipulation of databases, media and other resources hosted on one or more servers.
The resources provided by the servers are called up by their respective Uniform Resource Locator (“URL”). Generally, an URL will contain an IP address that points to a server followed by additional pointers to files residing on the server. In the case of a web application, a client browser can therefore access a webpage or a link by its URL. In particular, the browser typically first accesses the web application by its website address, which is a portal address of the application, calling up a homepage with links to the hierarchy of webpages. For example, a commercial entity may create an on-line shopping site, “www.onlineshop.com”, for customers to browse and purchase merchandise on the Internet. The domain name address “www.onlineshop.com” is an alias for the IP address that points to where the application resides on the Internet.
One problem this new computing paradigm presents is the need for the server hosting a web application to meet the potentially huge demand from the clients. The global nature of the Internet has meant that at any time there could be millions of clients attempting to access the same web application. A common solution is to host the web application in a data center.
FIG. 1
illustrates a data center hosting the web application by means of a server farm. Multiple replicas of the web application are made available from a group of servers, known collectively as a “server farm”. The data center provides a multiplicity of web servers and other related servers for hosting multiple copies of a web application and related resources. This architecture allows easy scaling of resource capacity to meet increased demand. When a client request comes in, a LAN/Web switch performs a load-balancing function by connect it to one of the less busy servers in the group.
The URL of the website for the web application now points to the LAN/Web switch so that when a browser addresses the web application, the client packets are initially directed to the LAN/Web switch. The LAN/Web switch then switches the packets to one of the less busy servers in the data center based on load-balancing considerations. The switching is done using information associated with Layers 2-4 of the Open system Interconnection (“OSI”) model, or the more specific Transmission Control Protocol/Internet Protocol (“TCP/IP”).
FIG. 2
is a table illustrating the protocol layers of the OSI model, the corresponding TCP/IP protocol stacks, and the types of conventional switching and routing operable at each layer. According to the OSI model, each device on a network implements the seven OSI layers in a modular fashion. Starting with Layer 7, which is a software application at the top, each layer communicates with its immediate layers. As the layers get lower, the information to be sent out is increasing packaged for the specific hardware of the device, ending in Layer 1, which is the physical communication channel itself. Under TCP/IP, Layers 1-2 are implemented as a subnet (or MAC) stack, Layer 3 as the Internet (or IP) stack, Layer 4 as the transport (or TCP/UDP) stack, and Layers 5-7 as the Application stack. Each stack is usually implemented by a software and hardware combination. Typically, data is generated by a first network device and is processed down the protocol stacks, from Layer 7 to Layer 1, into a packet, with each stack adding a header to the packet. The packet can then be sent via a physical channel to a second network device. The second network device processes the packet up the stacks starting from Layer 1, and unwraps the respective headers after terminating them at their associated stacks. At Layer 7, the application data of the first device is retrieved for interaction with the application of the second device.
FIG. 3
illustrates the various headers of an IP packet. Each IP packet consists of a data portion for carrying the data payload and a header portion for carrying overhead information. The header portion is further partitioned into layer- or protocol-dependent headers. For example, a Layer 2 or MAC header includes a destination MAC address and a source MAC address that respectively specify the destination and source hardware addresses of a node in a subnet. On a LAN, an IP packet is directed to a destination device by its destination MAC address. A Layer 3 or IP header includes a source IP address and a destination IP address that respectively specify the IP addresses of the source and destination nodes on the Internet. On the Internet, an IP packet is directed to a destination device by its destination IP address. A Layer 4 or TCP header includes a source TCP port and a destination TCP port that respectively specify the port numbers used by the source node and the destination node. On a device, an IP packet is directed to a destination port by its port number. In general, transporting a packet from one location to other requires processing of Layers 2-4 header information.
The data portion of the IP packet contains Layer 7 information, which is data generated by the application. In web applications, the data will include HTTP headers. Since HTTP is not one of the basic OSI or TCP protocols, but a High level protocol associated with web applications, its header is therefore regarded as application data and therefore located in the data portion of the IP packet. The HTTP header includes an URL field for specifying the URL the packet is requesting. It may also include a cookie field for the application to communicate environmental information with the client.
As mentioned earlier, each device communicating on the Internet implements the TCP/IP stacks. For example, when a client computer running a browser requests a web page from a server, the client packets are typically routed by a number of routers and possibly a web switch before reaching the destination server. When a router intercepts the packets, it is processed from Layer 1 up to Layer 3, so that Layer 3 information such as the source and destination IP addresses can be extracted in order for the router to route the packet to the next device. When the packets get to a conventional Web switch, the packets are only processed from Layer 1 up to Layer 4. In general, the upper layer information can only be obtained after the all the lower layer stacks have been processed or “terminated”. Thus, the upper or deeper layer information of an IP packet requires more processing to obtain.
FIG. 2
also lists the common types of routing and switching that can take place at the various layers. LAN switches, such as an Ethernet switch for a LAN operate at Layer 2 or the subnet stack of TCP/IP. Routers, operating at Layer 3 or the network layer, allow IP switching in which IP packets may be routed to a node having an IP address on the Internet. A router basically examines the destination IP address on a packet, looks up its routing table for the output port number in order to send the packet to the next node.
As mentioned earlier, a web switch is employed to switch an incoming client packet to one of many parallel web servers in a data center. In conventional implementations, its primary task is to perform a load-balancing fun
Cheng Bo-Chao
Hsing Deh-Phone K.
Lu Leonard L.
Wu Tsong-Ho
Eng David Y.
Parsons Hsue & de Runtz LLP
Transtech Networks USA, Inc.
LandOfFree
Content-aware web switch without delayed binding and methods... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Content-aware web switch without delayed binding and methods..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Content-aware web switch without delayed binding and methods... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3273421