Steps in HTTP request service

Before we discuss further, it is important to understand how a HTTP request is serviced, so it is briefly discussed here. A client's request for desired object is fulfilled in following steps:

• Domain name to IP address mapping : The domain name present in URL must first be translated to an IP address. The client software requests its local resolver for it, if this mapping is not in its cache. The resolver in turn returns the IP address for that domain name, that it may get from Intermediate name servers (which may have cached this mapping) or from directly from authorized DNS for that domain name either recursively or iteratively. More details about DNS mechanism can be found in RFC 1034 [26] and RFC 1035 [27].

• Request for object to server with that IP address: Then client software sends request for object to server having that IP address. The server may return requested object directly or it may redirect it to other server using HTTP header options or fetch the object from other server and deliver to client or may transparently forward the request to other server which replies directly to client with address of forwarding server, etc.

Thus HTTP request service path allows us to distribute requests at two levels, first at DNS at the time of resolution of domain name to server IP address, and the other at server when request reaches at that server. Any system consisting of multiple servers and some request distribution mechanism is termed Distributed Web Server System (DWSS).

Time taken for service of any HTTP request submitted by client depends on two major factors namely network conditions and server load. Even if there is a capable server system present, but the connectivity of client in terms of delay, available bandwidth or packet loss is not good, it will sees large delays. If server system is saturated with requests, time taken for service is very large. So for keeping response time minimum, web server system should take into account both the factors.