Sometimes the bandwidth on an enterprise network is like a piece of cheap clothing: After a few washes, it seems too tight and restrictive and you want something with more room. But bandwidth doesn’t really shrink, it just seems that way. What really happens is that network traffic grows to consume the available bandwidth.
In the old days–really just a few years ago–businesses would throw more bandwidth at the problem and buy faster network links. But the tough economy and tight IT budgets make it hard for companies to justify adding another fat pipe. The problem isn’t bandwidth per se; it’s having the predictive tools to understand network problems and putting in place preventive measures to deal with congestion. IT departments are trying to find an easier way to use existing bandwidth without constantly upgrading and make smarter upgrade decisions. That’s why more businesses are turning to a variety of bandwidth-optimization technologies and techniques to get better performance out of their networks and make the most of the bandwidth they’re already paying for.
Optimization technology offers the opportunity to decrease bandwidth needs and monthly costs while providing better service to users. Because bandwidth optimization devices are relatively inexpensive, return on investment (ROI) is quite fast, allowing these devices to provide "something for nothing." As a coarse example of bandwidth optimization’s rapid ROI, consider a simple case where two small bandwidth optimization appliances, costing $6,000 each plusmaintenance and other overhead expenses, are installed at the two ends of a WAN link. If they reduce the traffic on the link by 50%, thereby avoiding the need to increase capacity by adding an additional link, they will probably save the enterprise over $1,000 per month in WAN charges. Payback therefore occurs in less than one year.
In addition to the savings in bandwidth costs, there are other savings in IT staff and equipment that are made possible by moving servers and backup systems to a central facility. These cost savings may be greater than the savings in bandwidth costs alone. Most bandwidth optimization techniques can fit into one of three categories: Efficiency, Compression, and Omission. These are as follows:
Efficiency techniques involve changing the web content in order to minimize the number of bytes that need to be sent. For example, use external files (which will cache) instead of inline styles and scripts, reuse icon images, use semantic markup. Fix any broken images, since these often send a verbose 404 error page.
Use compression on the server to squash files before they are sent. Compression is a well-established technique in telecommunications; since without significant bandwidth compression, the telephone grid could not handle the amount of data that passes through it. On the web the most popular compression algorithm for real-time compression is gzip. The topic of compression also includes image compression e.g. JPG, PNG, GIF.
Omit unneeded bytes. Remove comments, whitespace, and don’t send tags.
At the most basic level, moving the data closer to the user improves response time. Service providers offer different services that let businesses cache frequently accessed Web pages or set up mirror sites for in-demand content in locations that are closer to the user seeking the information. Companies also have improved performance by minimizing the number of bandwidth-intensive elements, such as real-time images or video clips that are offered on their Web pages and by keeping each page well under a 2-Mbyte threshold.
Another proven approach is to assign priority levels to different types of traffic, applications, departments or even individual IP addresses. Those with the highest priority get first dibs on bandwidth. This technique goes by different names: quality of service, policy networking and traffic shaping. In addition, companies often use load balancing in their data centers to improve transaction speed, both internally and externally. Many businesses use technology to implement a thin-client application model to optimize both – network and systems performance as well as the user experience. More sophisticated approaches include compression appliances that sit in front of routers or servers and treat incoming or outgoing traffic sort of like a Zip file–compressing or decompressing as needed.
Many companies use a combination of these techniques because every enterprise network is unique, reflecting the number of locations that need to be connected and the types of applications and traffic needed to conduct business.