advertisement
On TV.com: JESSICA ALBA photos
Find Articles in:
all
Business
Reference
Technology
News
Sports
Health
Autos
Arts
Home & Garden
advertisement

Content provided in partnership with
Thomson / Gale

The rise of the `interprise' network: audit four key application traffic-management elements - Special focus: testing & monitoring - quality-of-service criterion

Communications News,  April, 2003  by Todd Krautkremer

The Internet's effect on enterprise networks has been profound, enabling businesses to extend their online reach to customers, partners and suppliers with unprecedented efficiency. Concurrently, corporate LANs and intranets have seen exponential growth, with both increased power on the desktop, and size and power of applications. As the wall between the corporate intranet and the Internet dissolves, a wave of new Internet-inspired technologies are finding their way into enterprise networks, including Web-based applications, enterprise portals, streaming media and voice over IP (VoIP). Welcome to the new "Interprise" network.

Most Popular Articles in Technology
An overview of continuous data protection
Why all those current ratings?
Many countries now have a mobile penetration rate above 100%, report says
The Tata Group's big telecom gamble: VSNL's recent acquisition of Tyco ...
MEASURING BANK BRANCH EFFICIENCY USING DATA ENVELOPMENT ANALYSIS: MANAGERIAL ...
More »
advertisement

New users, applications, traffic and media types demand different performance levels from the network and compete for bandwidth. Creating an infrastructure to control application performance requires a systematic approach to quality of service (QoS) that includes four key application traffic-management elements: discovery, analysis, control and service-level reporting.

The first step in delivering application QoS is to understand what traffic is running on the network and its impact on business-critical applications. Once network managers understand what is on their networks, they then need to understand how this traffic is utilizing bandwidth and its impact on the company's most critical business applications.

Traffic- and application-performance analysis functions traditionally have been performed using network "sniffers" installed at various points in the network, or by application-performance management (APM) software installed on clients and servers. As network traffic becomes more dynamic, due to increasing volumes of Internet-derived traffic, sporadic sniffer analysis has become inadequate in painting a complete and accurate performance picture.

Application-level bandwidth-management platforms pick up where these other technologies have left off, providing a comprehensive view of all traffic on the network and ongoing analysis of bandwidth utilization, latency and application performance. Once an IT or network manager understands how bandwidth is being consumed and how traffic is impacting the performance of business-critical applications, control over bandwidth utilization and application performance can be achieved by systematically applying QoS technologies.

Although many of today's routers and switches have integrated QoS features that are effective in managing network-level congestion on specific ports, they lack several critical capabilities necessary for control of end-to-end application performance. These shortcomings include: limited Layer-7 application discovery and classification; no inbound control of traffic; and no per-flow, rate-based control of application sessions.

A relatively new category of equipment, application-level bandwidth-management systems, is specifically designed to overlay existing routers and switches to create an application-performance infrastructure. These devices adapt the underlying network to specific needs of applications and users by applying a wide range of traffic prioritization, shaping, metering and marking techniques on a per-application and per-flow basis.

Because most of these products provide deep Layer-7 classification, they can segregate traffic into discrete classes. Because bandwidth managers have a finer level of visibility into the different types and priorities of traffic, they also have a finer level of control.

Each class of traffic can be assigned a unique set of QoS parameters to ensure optimal performance. Moreover, these controls can be implemented on a per-flow basis to ensure no single user is under- or over-served. The most advanced products can provide bidirectional control of traffic, ensuring that inbound traffic flows are appropriately controlled.

Policy-management software platforms represent a new category of network-management solutions that are specifically designed to centralize and ease the provisioning and management of complex network and application QoS policies. The elements that together form a policy-based QoS system include the definition and distribution of policies (policy definition), the calculation and control of policies (policy control), and the execution of policies (policy enforcement). Policy-management software provides a centralized vehicle from which to define, distribute and, in some cases, control QoS policies in conjunction with a policy-enforcement system.

Since the policy-management software does not sit in the direct packet path of IP traffic, it relies on an external router or bandwidth-management system to enforce policies. Policy control may or may not be part of the policy-management software's role, depending on the type of policy-enforcement platform.

For example, more advanced bandwidth-management systems actually incorporate the policy-calculation-and-control elements with enforcement. This ensures that the active elements of the QoS policy system that must respond quickly to changing network dynamics are reliable, efficient and effective. In this model, the policy-management software simply provides a means to centrally define and distribute policies to a large number of remote bandwidth-management systems.