Bandwidth Management - Introduction

Bandwidth management describes the use of software tools to put into place policies that govern how network traffic runs should carry on when passing a WAN (wide area network) or Internet connection. These tools enable detailed regulation over traffic, often on a per-flow basis. Network managers can apply policies that the network will apply according todestination address, protocol type, user and/or source address, application type, and other factors.

Enterprises can build a business case for bandwidth management by first estimating the cumulative savings from delayed bandwidth upgrades. The next step involves examining the high cost of network and application downtime to their organizations' bottom lines, then monitoring the status of their networks to see if any core business applications are creating productivity losses during periods of congestion. The aggregate savings from implementing a bandwidth management solution is almost always a compelling business case with a rapid ROI for the organization. Enterprises would like to gain the benefits of converged WAN services while still maintaining the application response times that will suit users and make application deployments successful. Monitoring only bandwidth management systems provide the dual benefits of providing this baseline business case data, and potentially replacing current or planned network monitoring platforms. To achieve a win-win situation, it is likely that some level of bandwidth management will be required.

The first step to successful bandwidth management is monitoring applications, protocols and users by using products that inspect network layer behavior and conditions as well as specific application performance levels, and have the ability to correlate the two. Before you can successfully manage application performance across a network by setting network policies and classifying traffic, you must have a way to discover the various applications, protocols, and users on the network and evaluate how they are behaving.

Bandwidth management involves classifying and marking traffic as to its priority. For example, delay sensitive applications like voice over IP (VoIP) usually would have a small amount of bandwidth guaranteed to them, and VoIP packets would be marked for placement in the top-priority queue. Citrix traffic requires that a small (but consistent) amount of bandwidth always be available in order for sessions not to break.

Real-time reports are invaluable in troubleshooting immediate network problems, while long-term reporting is best for analyzing network trends, such as daily congestion conditions on a given circuit. The most useful and versatile systems also provide policy-based bandwidth management capabilities since you will want to address the problems identified by your reports. A high-quality seven-layer monitoring tool should also provide easy access to a wide variety of real- time and long-term reporting features. Use of a full-stack (seven-layer) device is especially powerful, because it pinpoints issues down to individual sessions, applications, protocols and users.

Together, classification and assigning actions to traffic classes form a network policy that will optimize your: bandwidth utilization, application performance and your business/IT resource alignment.

Setting enterprise-wide network policies should be a joint effort between the executive staff, departments and business units and the IT organization. Establishing a policy framework for your bandwidth management system should be closely tied to your strategic corporate goals, and affords you the opportunity to align your IT and networking resources with those goals. When you are comfortable that your traffic is being classified and grouped according to your objectives, policies can be created that direct your bandwidth management system on the proper handling of each traffic type. Once your policy framework is defined, most systems make it easy to create service groups to help classify traffic based on variables such as application, user, server, time of day, and time of week.

Once you have discovered what's running on your network, you can classify traffic as to its priority during periods of congestion. This type of monitoring requires DPI so that applications can be identified that don't utilize fixed port assignments, or use fixed ports but carry different traffic types.

Converged networks afford many efficiencies and application innovation. Bandwidth management has become a fundamental element of network management in an era where enterprises and service providers are merging all traffic onto packet-switched networks.But they also require monitoring and control to ensure that the various applications (all contending for a common pool of bandwidth) don't negatively affect one another.