I have also noticed the changing tide toward a smarter way to understand and address performance problems. The trend has been slow, but it is accelerating because people have no choice. Traditional methods and tools cannot possibly fulfill the need.

It is good to see new technology solutions - and new thinking - beginning to take over!

– Glenn –

As you clearly articulate, the performance management problem can no longer be handled by throwing more people at it and relying on static threshold-based alerting and tribal knowledge-based correlation. Of course, new technologies like virtualization and SOAs, which provide undeniable benefits, compound the existing management problem tremendously. A new approach to performance management is the only way to scale in the face of this growing complexity. You’ve really captured the key elements required to address the problem here: dynamic threshold-based alerting and real time correlation of alert and metric behavior.

Too many IT Operations teams are mired in monitoring events and spend tremendous amounts of manual effort figuring out what alerts are important and which they can ignore. Correlation is limited to simple rules based on tribal knowledge. I was talking to a VP, IT Operations at a large online financial who told me that he had 10 highly paid individuals on his team who do nothing but look at graphs all day. They are comparing week over week metric behavior to identify abnormalities and manually correlating them to solve performance issues. I’ve since seen them doing this, printing out graphs of metric data and holding them up to the light to compare them. It is a slow process that seems archaic, but it is common practice in many Operations teams I’ve talked with. This particular team needs to scale their application infrastructure by a factor of three over the next two years while keeping budget relatively flat. Without automation of these manual tasks, there is no way that will be possible. Of course, this is why they are looking at Integrien’s solution.

I look forward to continued discussion on the differing approaches to analytics-based performance management and how adoption and operationalization are proceeding.

Steve

Your point on the dichotomy of the enterprise and the service provider sides are particularly notable. I’ve always found this separation a bit distressing, but as you note, the differences in scale and maturity are the main reasons. Traditionally, service providers were far ahead of enterprises in their operational sophistication (hmm, I smell another blog posting here!), which forced them to take a different approach.

Enterprises were more enamored by “eye candy” than capability and automation. This is thankfully beginning to change because it must. The complexity and scale of many enterprises rival service providers and the stronger focus on operational efficiency (a la ITIL) is accelerating their transformation into internal service providers.

This is great news, albeit a shift that is painful for many of those involved. No pain, no gain! Besides, it’s far less painful than punitive outsourcing.

I see two silos in this area still - the service providers and the enterprise. Partly a matter of sheer scale, but automation, analytics, reporting and visualization advances are happening on a faster pace on the service provider side best as I can tell. I just see too many tools doing their own graphing, charting and performance management like things which continue to chip away at the goals of a consolidated performance management solution. We have no fewer than four or five (maybe more) tools in our IBM Tivoli portfolio that play in these areas on both sides of the fence. What we lack is a OOB capability as you’ve described above similar to ProactiveNet, Netuitive, and Integrien offer. This is often left to sophisticated programmatic rules engine development, suppression and correlation engines within our suite.

I think that plain old performance management is tainted as you’ve described. Being the BSM guy that I am, I think the real innovation may come when the performance of key business services, applications and systems can be looked at from end-to-end and directly tied back into the business goals and objectives. The blending of the IT management side, Business Service Management and Business Intelligence (specifically Operational BI). Innovation here is needed and may be one way to put the grove back in performance management and monitoring.

I look forward to following your new voice outside EMC!

Doug