The Changing Face of Network Management

Keeping up with Changes
The Only Thing Certain is Change
Today’s complex network infrastructures can contain hundreds or thousands of businesscritical devices. Unauthorized or incorrect changes to even a single device can have a rippling effect on the health and availability of the network infrastructure. A recent article in Network World’s Network/Systems Management Newsletter suggested that the impact of configuration change on network stability is significant. It stated, “It has become common knowledge that when systems go down, applications stall and networks fail it's usually due to unknown, undocumented or unauthorized changes to network and systems configurations — and not because of technology failures. In fact, it has been estimated by numerous industry watchers that up to 80% of application performance problems and network downtime can be attributed to some configuration change or error.” Even if only half of all catastrophic network failures are caused by configuration change, shouldn’t you make network configuration change management a priority?

Exact Change Only
There are many reasons why configuration changes are so often incorrect.
- There are simply so many of them. Router changes are numerous and networks teem with routers and switches and other devices. They represent multiple vendors, each with their own command line structures. The number of changes is daunting and the variety of syntax makes managing with multiple tools difficult.
- Network management and configuration is highly detailed and is often handled by a few seasoned techies. Manual input is still common and even seasoned techies can make mistakes when keeping track of so many details.
- Discrepancies develop between the start-up and running configurations. This can happen when the configuration change is correct but is not saved to non-volatile RAM. In the event of a device reboot, the device reverts to the old configuration.
- Time delays caused by manual input can cause problems. If five routers need a similar change, some of them will be done before others. The time discrepancy may cause incompatibilities while the entire task is being completed.
- If configuration changes are handled manually, in the data center, then there is opportunity for less knowledgeable staff to input a change incorrectly.
The potential impact of configuration changes is evident. A quick Internet search shows that universities advertise their expected outages to network users when they have configuration changes planned, since the campus depends on the network and outages generate numerous complaints from users. Concern over outages is even greater for businesses whose revenue depends on services that run over their network. The greatest problem for businesses is when the change and the outage come as a surprise.

An Opportunity to Change for the Better
A good network fault and performance management solution is a necessity for keeping the network up and running. If your solution provides end-to-end visibility you will know when network operations are in jeopardy. Preventing network problems, such as those stemming from incorrect configuration changes, are opportunities to increase network uptime, provide better business services and improve business continuity. There are several things you should think about when incorporating configuration change management into your solution. With Tools, Less is More One decision to make regarding network configuration and change management is what solution you are going to deploy — a niche stand-alone application or one incorporated into your fault and performance management solution. Here is an opportunity to simplify your overall network management, thereby relieving your IT staff of significant error-prone, laborintensive effort. The more tools you have, the less integrated the management of various technology domains and the less efficient your IT staff will be. A niche tool only exhibits its excellence in its native environment. With many different vendors and platforms, the amount of manual correlation is significant and you lose the opportunity to speed up and clean up change processes. A single tool for fault and performance management, with a centralized control point, will not only enable better network configuration and management in the first place, but it will enable you to rapidly spot and resolve configuration conflicts. For example, intelligent thresholds are essential to performance management and should include proactive performance alarming on key performance indicators for a particular service, such as VoIP.