Web services: pump up the volume: bandwidth consumption and network latency challenges must be addressed

Communications News,  Feb, 2004  by Zohar Pearl

• E-mail
• Print
• Link

In contrast, current WAN optimization solutions employ next-generation compression and caching technologies that handle all application types and protocols, including non-IP protocols. They use a variety of traffic reduction techniques, including byte-level caching, header reduction and adaptive packer compression.

Due to the significant overhead and repetitive nature of XML data, it lends itself particularly well to this type of optimization. This multifaceted, multiprotocol approach enables WAN optimization products to deliver an average WAN capacity increase of 100% to 400%, with peaks of 1,000%+, depending on the mix of applications and protocols. A 300% capacity increase, for example, is equivalent to upgrading a 128 Kbps frame relay link to 512 Kbps.

Application visibility enables network managers to see all traffic flows on a WAN and to identify the applications generating the traffic. Without such traffic monitoring and analysis capability, WAN optimization is just "a shot in the dark." Network managers have to understand application traffic patterns and volume in order to make the best use of their WAN bandwidth. Some of the most important uses of traffic monitoring and analysis include:

* baseline analysis of traffic at the beginning of a WAN optimization initiative;

* measure the results of WAN optimization vs. the baseline data to ensure that the desired effects are being achieved;

* detection of unauthorized or rogue applications that might be "stealing" bandwidth; and

* capacity planning to stay ahead of the bandwidth demand curve as traffic patterns and volume change.

To perform these tasks, WAN optimization products must capture traffic statistics, identify the applications that are generating the traffic, and then provide both real-time and historical reporting.

QoS enforcement is the process of proactively allocating bandwidth to applications based on business requirements. Without QoS, TCP/IP networks provide only a best-effort service that provides bandwidth on a first-come, first-served basis. This approach works reasonably well on the Internet, bur on enterprise networks all applications are not equal. Enterprises need the ability to align bandwidth utilization with business priorities-what QoS enforcement is all about.

The two most commonly used QoS techniques are prioritization and partitioning. Prioritization uses well-known mechanisms, such as weighted fair queuing, priority queuing and custom queuing, to control the relative priorities of applications. When the demand for bandwidth exceeds the supply, the highest priority applications are serviced first.

The other QoS mechanism, partitioning, is used to allocate specific amounts of bandwidth to a specified application. For example, a network manager could ensure the performance of VoIP by creating a logical partition to protect the voice traffic from other applications.

Prioritization and partitioning are complementary techniques for ensuring that the bandwidth is allocated to mission-critical applications, not just bandwidth-hungry ones.