The need to rethink routing

Our last newsletter summarized some of the heated discussion that took place at a panel that Jim recently moderated at Interop in Las Vegas that was entitled: “Is Routing Undergoing a Mid-Life Crisis?” This newsletter will continue the discussion about what, if anything, is driving the need for IT organizations to rethink their approach to routing.

Our last newsletter summarized some of the heated discussion that took place at a panel that Jim recently moderated at Interop in Las Vegas that was entitled: "Is routing undergoing a mid-life crisis?" This newsletter will continue the discussion about what, if anything, is driving the need for IT organizations to rethink their approach to routing.

When IT organizations deploy new routers and switches they typically want those devices to have a very long life cycle. The challenge, however, is making sure that these devices can support not only today's requirements and the ones that are likely to be real in the near term, but also the requirements that will not manifest themselves for several years.

One current trend that is impacting routing is that most IT organizations are removing servers out of branch offices and placing them into centralized data centers. Many of these IT organizations are also reducing the number of data centers that they support and implementing server virtualization wherever possible. We believe that this creates an "all of your eggs in one basket" phenomena whereby the corporate data center becomes even more critical to business operations and hence must deliver the highest levels of availability, while accommodating much higher levels of traffic aggregation and scalability. 

To validate that assertion, we talked to the vice president of architecture for a large financial services firm. He stated that his organization has implemented a number of forms of virtualization and that the fundamental driver of these initiatives was cost optimization. He added that as a result of virtualization that their previous goal of five 9s of availability for their data centers is no longer acceptable and that their new goal is 100% availability.

There are many techniques that can be used to increase router availability. This includes the use of redundant hardware subsystems that can support non-stop operations in spite of component failures, and high availability network operating system software that supports features such as in-service software upgrades, hitless fail-over among redundant route processors, and hitless process and protocol restarts.

When discussing the scalability of routers, it is important to verify that performance is not adversely affected when a wide array of features is enabled. For example, within the consolidated data centers previously mentioned it may be necessary or desirable to enable a number of security features, such as access control lists, firewall feature sets, and traffic event logging. Additional router features and services that are frequently required include application recognition and QoS policy enforcement.

Our next newsletter will continue the discussion of routing and what, if anything, is driving the need for IT organizations to rethink their approach to it. In addition, Jim is spending this week in Silicon Valley getting a sense of what technologies a wide range of vendors find hot, and what technologies they find to be over-hyped. You can follow Jim on Twitter: AshtonMetzler

Learn more about this topic

Six worst Internet routing attacks

IETF to explore new routing technique

Join the Network World communities on Facebook and LinkedIn to comment on topics that are top of mind.

Copyright © 2009 IDG Communications, Inc.

IT Salary Survey: The results are in