The Tech Side of Server Monitoring & Capacity Planning

My colleague and our fearless CTO Alex Bewley recently wrote about the impact of technology on our environment in his piece titled Reductionist Mindset. There was also an article recently in the New York Times Bits Blog Data Centers Are Becoming Big Polluters, Study Finds discussing how by the year 2020 it is expected that data centers will contribute more greenhouse emissions than the airline industry.

The following quote from the NYT article sums up how until now we have been looking at efficiency within the data center.

For example, computer servers are used at only 6 percent of their capacity on average, while data center facilities as a whole are used at 56 percent of peak performance. In other words, if data centers were hotels, they would be bankrupt and shut down instead of growing like kudzu.

So it is not all doom and gloom. If we can increase the utilization of our infrastructure even nominally, the reduced impact we can as data center operators have on the environment can be significant. Through virtualization (VMware, LDOMs, LPARs, etc) we are now starting to see utilization/efficiency rise on servers to levels that the mainframe days enjoyed.

As the management tools for virtual infrastructure mature, I think that we're going to see more and more capabilities built in around managing data center power and cooling as part of the overall virtualization strategy. Distributed power management and localized cooling, rather than cooling the entire data center. Also as power density rises, centralized cooling will become an increasingly difficult proposition to implement.

Labels: Virtualization Power Environment

Michael Bailey (Director PM) and I presented at the Toronto VMware user group on Tuesday in the Glen Gould studio at CBC to an audience of about 110 people. We discussed the management challenges faced by IT that are caused by virtualization.

There are very clear and obvious benefits to introducing virtualization into the datacenter, the obvious being consolidation and rationalization. It's estimated that by 2012 over 90% of large enterprises will consolidate their IT assets through virtualization.

VM growth in the marketplace (not just VMware, but LPARs, LDOMs, Containers, etc.) is rapidly increasing with the VM installed base to hit 4.1 million VMs in 2009. That's almost an 8x increase from the 540k installed in 2006. The good news is that we are building dynamic services that can easily adapt to constantly chaging business drivers and pressures. The bad news is that a whole new basket of problems are introduced by virtualization.

• Problem Isolation
• Licensing and Compliance
• Change Management (Tracking, Automation, Control)

In general the tooling around VMs have not kept up with this growth. Which is where up.time comes in. While up.time is not a "silver bullet" for every single VM related challenge today, we do address what the industry sees as key problems being faced today. Specifically

• Determination of VM candidates
• Controlling Sprawl
• Identifying VM Configuration Information
• Problem Isolation
• Workload Trending

After our "slideware" presentation, we gave a "software" product demo presentation to show up.time in action. We specifically showed the following stories

• The big picture from 10,000 feet : "Managing from above"
• Trending the workload of my guests : "Am I growing, shrinking or steady state?"
• Identifying overresourced VMs : "Where did my memory go?"
• How do I look from a data center capacity perspective : "Am I well utilized or under utilized?"
• Automated & Ad Hoc Virtualization Reporting : "Become the ESX superstar"

The presentation went well and we actually ran out of time at the end due to the number of questions. The stage lights came up, the band started playing and the big long cane scooped us off the stage.

Labels: Monitoring, Server Reporting, Virtualization, VMware

I love where I work! I don't think that there are many software vendors that provide their employees with the freedom and flexibility to execute on whatever is needed to get the job done in the way that uptime software lets me. Because of this freedom, uptime as a company is very agile when it comes to providing rapid solutions to customers needs in order for their use of up.time to be successful.

In this case I'm thinking specifically about the fact that we have now added monitoring AS400 to our capabilities. While these monitors have not been publically released yet, I know that they will be. The new monitors provide monitoring, alerting and reporting capabilities for CPU, Memory, Disk, Job & Message Queues as well as Users, ASP and PTFs. As these monitors get closer to release I'll update my blog about them. If you would like to beta these, please contact support and they will redirect you to me. (support@uptimesoftware.com)

For anyone who is interested I will be presenting up.time at the VMware user group in Toronto on April 8th. During the presentation I will be covering the following topics

• Monitoring ESX servers and their guest VM workloads
• Guest VM Application and service monitoring
• Reporting on Vmotion/DRS enabled server farms
• Identifying virtualization candidates within your infrastructure

For anyone who is interested in attending the Toronto event, you can register at the following URL

http://campaign.vmware.com/usergroup/invites/Toronto_4-8-08Invite.html