Kanban Capacity Management Made Easy

Kanban Capacity ManagementBy: Kent Erickson>>  Is “over-provision and forget” your approach to capacity management? A recent article from The Register suggests that it’s all too common. According to the article, only about 15% of organizations proactively manage resource capacity, and only half even have threshold alerting.

There’s nothing like a good fire drill, trying to figure out which server in a key application is hurting, is there? Why do we rely on instinct and fire alarms? The article named executive understanding, budget, and work effort as the top challenges.

Kanban Capacity Management

How about a nice simple solution you can implement in just a few minutes? A solution that focuses on warning you of looming issues a few days in advance, in areas you can actually fix quickly. Something that works equally well for servers at AWS or in VMware?

And that you have complete control over, and works with your existing monitoring system. Nothing new to buy, no new agents.

Well, if you’re running Zenoss Service Dynamics, you’re in luck! Let’s look at Kanban (just in time) Capacity Management with Zenoss!

Zenoss Kanban Capacity ManagementThe key is the Predictive Thresholding ZenPack available with Zenoss 5.1. With just a couple minutes work, you can days of warning that a key virtual machine needs a CPU upgrade and find capacity in underused virtual machines at the same time!

Starting incrementally is a great approach and helps keep my blog articles to a reasonable length, too.

I suggests starting by focusing on CPU utilization for VMs. Add your predictive threshold to the standard VM template using the average CPU metric, keeping in mind that the normal range is 0-9999, for 0-99.99%. Add an upper limit at, say 75%, and a lower limit at 10%. You’ll get a 10 day prediction either way. That’s enough time to resize the VM during a normal maintenance window. Ah, planning.

Capacity Planning Management ZenossIf you’re using service impact and change the event class to /Perf the capacity warnings will show up in the root cause analysis, and you’ll know which applications might be affected, too.

After that’s been working a few weeks, add CPU utilization for hypervisors at 75% or so, AMIs at the VM value; memory utilization for VMs and hypervisors; and for network interfaces at 50%. Watch and tweak as necessary.

It won’t take long to make just-in-time capacity management part of a regular routine, and eliminate most of those nasty surprises!

How do I get the Predictive Thresholding ZenPack?

The latest Predictive Thresholding ZenPack is available for Zenoss Service Dynamics customers with an active subscription.

Learn more about Zenoss products and solutions and schedule a demo at http://www.zenoss.com

Join and engage with the Zenoss Community at http://www.zenoss.org