Previous Table of Contents Next


Chapter 44
Selecting a Web Server Connection Rate

Gilbert Held

Determining the best operating rate for a WAN connection to the Internet is a common problem for organizations wishing to obtain a presence on the World Wide Web. Some simple calculations can help network managers compare and balance user requirements against the cost of providing an Internet connection.

PROBLEMS ADDRESSED

If the operating rate of the Internet connection is too slow, anyone trying to access an organization’s server from the Internet may get frustrated and terminate their access of information from the corporate Web server site. At the opposite extreme, if an organization’s Internet access connection operating rate exceeds the bandwidth required to support an acceptable level of access, you may be wasting corporate funds for an unnecessary level of transmission capacity.

As this chapter shows, with knowledge of the ways in which a Web server can be connected to the Internet, as well as knowledge about some of the transmission constraints associated with a Web server connection, it is possible to determine an appropriate Web server connection rate.

BASICS OF CONNECTING TO THE INTERNET

Exhibit 1 illustrates the typical method by which Web servers are normally connected to the Internet. A Web server resides on a local area network, with the LAN connected via a router to an IAP. The IAP has a direct connection to a backbone network node on the Internet, commonly using a full T3 or SMDS connection to provide Internet access for a large group of organizations that obtain Internet access through its connection facilities.

Although an Ethernet bus-based LAN is shown in Exhibit 1, in actuality any type of local area network that can be connected to a router (and for which TCP/IP drivers are available) can be used by the Web server. Thus, other common LANs used by Web servers include Token Ring and FDDI as well as the numerous flavors of Ethernet, such as 10Base-T, 100Base-T, and 100VG-AnyLAN.


Exhibit 1.  Web Server Connection to the Internet.

Analog vs. Digital Leased Lines

The actual WAN connection between the IAP and the customer can range in scope from low-speed analog leased lines to a variety of digital leased lines. Only a few access providers offer analog leased line connection options. When offered, the actual operating rate of the WAN connection is commonly limited to 19.2K bps or 24.4K bps, based on bandwidth constraints of a voice-grade analog leased line that limits modem-operating rates. Concerning digital leased line operating rates, IAP recommend and offer 56K-bps, Fractional T1 in increments of 56K- or 64K-bps, full T1, fractional T3, and full T3 connectivity.

Connection Constraints

Although the WAN operating rate can constrain users from accessing information from an organization’s Web server, another less recognized but equally important constraint exists — that is, the traffic on the local area network on which the Web server resides. Although the focus of this chapter is on determining an appropriate WAN operating rate to connect a Web server to the Internet, it also examines the constraints associated with LAN traffic that affect the ability of the server to respond to information requests received from the Internet.

WAN Connectivity Factors

Three key factors govern the selection of an appropriate operating rate to connect a Web server to the Internet through a wide area network transmission facility. Those factors include:

  The composition of the Web pages residing on a server.
  The types of pages retrieved by a person accessing the Web server.
  The number of “hits” expected to occur during the busy hour.

A typical Web page consists of a mixture of graphics and text. For example, a university might include a picture of “Old Main” on the home page in the form of a GIF file consisting of 75,000 bytes of storage supplemented by 500 characters of text that welcomes Internet surfers to the university home page. Thus, this university home page would contain 75,500 bytes that must be transmitted each time a member of the Internet community accesses the home page of the university.

By computing the data storage requirements of each page stored on the Web server and estimating the access distribution of each page, it is possible to compute the average number of bytes transmitted in response to each Internet access to the organization’s Web server.

For example, assume an organization plans to develop a Web server that stores four distinct Web pages as well as a home page, providing Internet users with the ability to access two types of data from the home page. The construction of a two-tier page relationship under the home page is illustrated in Exhibit 2.

This example used to compute an appropriate WAN operating rate is for illustrative purposes only. Although the Web home page is always initially accessed, from the home page users typically access other server pages using hypertext links coded on the home page. Similarly, upon accessing different server pages, a user who wants to jump to other pages on the server is constrained by the links programmed on each page. Thus, the data transmitted in response to each page an Internet user accesses, as well as the sequence of pages accessed, will more than likely differ from organization to organization.

PERFORMING THE REQUIRED COMPUTATIONS

Assume that an organization has already determined that when Web pages are arranged in a tier structure, access to a home page at the top of the tier represents 40% of all accesses, while the remaining 60% is subdivided by remaining tiers. Furthermore, the organization’s Web page structure is to be constructed in two tiers below the home page, with the data storage associated with each page to include text and graphics as well as the access percentage of each page, as listed at the bottom of Exhibit 2.


Exhibit 2.  Web Page Relationship.

After determining the data storage required for each Web page and the distribution of Internet access by page, it is possible to compute the average number of bytes that will be transmitted from the Web server in response to each “hit” on the organization’s server. Here the term “hit” refers to an access request to a Web page on the server via the HTTP using a URL that represents a file stored on the server, which equates to the contents of a Web page.

Using the information from Exhibit 2, the average data transmission rate resulting from a hit on the organization’s server is computed as follows:

120,500 × .40 + 80,500 × .30 + 60,500 × .30 = 90,500

Thus, each hit on the organization’s Web server results in a requirement to transmit 90,500 bytes of data from the server to the Internet via the WAN connection to the IAP.


Previous Table of Contents Next

Copyright © CRC Press LLC