5 Common Server Virtualization Management Mistakes

November 3rd, 2014 by

There are 5 common server virtualization management mistakes I often see. VMTurbo will help you avoid them and much more.

1. Over Provisioning

One of the main reasons you switched or are considering switching to a virtualized environment is to cut costs. However, to assure performance, you will typically leave some “room for error”. So you will purchase more hardware than what you actually need, a common mistake in server virtualization management.

Often this leads to A LOT more hardware than what you need. Using VMTurbo you can understand exactly how much infrastructure you need in order to support your workload and assure application performance. VMTurbo will notify you when you need to provision new hardware, and if you can suspend existing hardware and still support the same workload without any performance degradation.

Problem: How much of your resources are actually used?

Server Virtualization Management Over Provision Problem

Solution: Journey to a desired state. The grey bars are physical machines that can be suspended once VMTurbo controls your environment

Server Virtualization Management Over Provision Solution

2. Under Provision

Other side of that coin. Not all workloads are created equal and some VMs need more resources than others. The other day a customer mentioned that they are going to move an application from the virtualized environment back to a physical one. The reason for the move was bad performance.

The mistake was, the VM was not getting all the resources it needs. While the physical machine that was supporting this virtual machine was running at less than 40% CPU utilization the virtualized machine was suffering from constant 100% vCPU utilization. VMTurbo suggested to increase the capacity of that VM’s compute.

Problem: Multiple alerts and warnings for stressed resources. How do you handle these?

Server Virtualization Management Under Provision Problem

Solution: Holistic demand driven actions. VMTurbo suggests increasing vCPU where there is a demand and the ability to meet the demand on the underlying infrastructure.

Server Virtualization Management Under Provision Solution

3. Wrong Placement

Like I already mentioned, not all workloads are created equal. Some need more compute power, others memory, IO, network etc. If you place the VMs correctly, according to their demand, you will get the most out of your environment (performance and efficiency wise).

The mistake is, to do this based on the capacity and not based on the demand. For example, configuring your environment across the board with 30% overprovisioning of memory. Meaning, if you have a 64 GB server, it will run VMs with 83 GB (64 x 130%) vMem on it.

However, what happens if on one of these servers all the VMs actually require that memory yet on another one they demand a fraction of the memory allocated? VMTurbo will make placement decisions based on the actual demand of your workload and not based on the capacity.

Solution: Placement decisions based on demand and supply of each element and its resources in your virtualized data center

Server Virtualization Management Placement

4. Static Placement

Even if you have the world’s best placement Excel worksheet. And you take into consideration all the moving parts of the hundreds of metrics that can affect your environment; by the time you finish inputting the details your workload changed. Now, you need to redo everything.

The mistake is, assuming there is a right placement. There isn’t. There is only a good placement for this point in time. VMTurbo will continuously drive your environment to a healthy state. As your workload demand changes, we will update our actions accordingly.

Solution: Actions represent what is needed at this point in time in order to get to a desired state

Server Virtualization Management Static Placement

5. Waste

There are a lot of moving parts within your virtualized environment. And sometimes you don’t control all of them – users have self-service portals, many people on your team etc. There is (plenty of) room for human error.

A few examples where VMTurbo can assist: too big of reservations driving efficiency down. Identifying wasted storage due to VMs that were deleted without deletion of their files. Identifying VMs that still have files taking space in your system that are powered off and have been for a while.

Solution: Wasted storage due to “garbage” files or powered off VMs

Server Virtualization Management Waste

Bonus one!

Trying to do it yourself.

If I asked you what is the square root of 72437121, would you try and solve that without a calculator? Probably not. Because some problems are solved better with software. Controlling your virtualzied environment is a much harder question. The answer isn’t 8511, it’s get VMTurbo.

Triangle: Efficiency
This article is about efficiency. Read more like it at the server virtualization Performance, Efficiency, Agility series.

Leave a Reply

Your email address will not be published. Required fields are marked *