When I step back and think about how we did monitoring in the past it amazes me how we survived in our infrastructure roles in IT.  More specifically when server virtualization was introduced was there really any monitoring available?  We were closing our eyes, and crossing our fingers hoping that our design could meet the test of time.Virtualization

Early Server Virtualization

For me it was about 13 years ago when I was first introduced to any enterprise actually leveraging hosts, and deploying virtual servers to the organization.  I will be honest I didn’t want anything to do with it.  It was too new, and I didn’t trust that it would support my servers the way I would want.  Also, many applications that I was architecting such as Exchange Server, SharePoint and Lync also didn’t have true support models for it yet.  Without a true support model from Microsoft getting any type of assistance from Microsoft during a real application problem was near impossible.  So for the short-term I wasn’t using the technology, but keeping a watchful eye on its development.

Also, in the beginning of the enterprise adoption of server virtualization; workplace implementation strategies were to only put Tier 3 applications on virtual servers.  So for many organizations there were initially few virtualized servers in their enterprises, and for me I typically didn’t leverage this technology option much until a couple years later.

As Time Went On

As time went on it became clear that this server virtualization thing was going to catch on, and improve, to the degree that any all applications have support statements.  With this came the importance of improving server virtualization monitoring.

Early Monitoring

Early virtualization monitoring included installing that same monitoring tool agent from your physical servers, and simply monitoring the windows server performance, and applications the same way you would with a physical server.  For host monitoring we were leveraging the out-of-box tools that came with the virtualization platform we chose.  We quickly learned that this was not enough.  The hosts within our virtualization platforms required proper monitoring, so it was only a matter of time and there were great options available.

The other issue with early monitoring is that we just let the flood gates open, and alerted on everything.  Monitoring turned into what should have been a proactive toolset, to the kind of toolset that was useless and ignored.  If an alert isn’t actionable and we receive it anyways, well we will typically just start ignoring them all.

Top 5 Lessons Learned in the Past

So what lessons have we learned from early monitoring of our virtual servers?  There are many, so let’s take a look here:

One:  We needed proper tools for monitoring with the right monitoring options to be successful, and for many years those tools just didn’t exist.  Now that they do, it’s time to research your options, and choose a robust solution that meets your organizational needs.

Two:  Alerting must be setup properly to be useful or everything gets ignored.  It’s very important to understand how to properly dedicate resources to your monitoring project.  I recently wrote about how to take monitoring gone wrong and turn it into monitoring gone right.  This post can be found here.

Three:  Virtualization monitoring should minimally include the following configuration items:

  • Server Level: disk space free, CPU, memory, network traffic, and core server services
  • Overall System Performance Statistics at the Host Level
  • Storage Usage/Disk Performance
  • The ability to analyze usage trends and patterns, reclaim VM sprawl resources due to incorrectly provisioned VMs

Four:  Keeping track of the existence of server Snapshots in your environment.  Depending on your design there can be issues with these existing for too long for a couple of reasons.  One issue I have seen is that when the snapshot is removed it can cause performance issues on the host.  Another problem can come from the consumption of disk being too high for the environment design leading to server downtimes.

Five:  Licensing and Renewals – While this doesn’t sound all that critical it should be looked at closely.  As staff and responsibilities transition it is too easy for this information to become lost.  Having your monitoring tool monitor and alert on this detail is extremely beneficial.

Virtualization Monitoring Today

One of the best things about technology is it’s never the same, and it almost always get better with the right direction and desire for it to improve.  With new programs and software, we have the opportunity to start with something new, and improve it with our experiences.  Server virtualization of 13 + years ago isn’t the same, and neither are the monitoring tool options we can use to support our virtualized environments.

We have learned and applied the lessons to how we leverage technology today.  Watch for my next post on this topic which covers “The Present – Monitoring Your Virtual Server Environment”, coming soon.