Over-provisioning and the associated cost

A typical customer environment will facilitate VMs in a Blade chassis, with access to storage across an FC SAN. How can you ensure that you sweat your current assets and avoid CapEx/Opex overage?

The short answer is to optimize and forecast based upon trends. Let’s look at common costs attributed to this model:

  1. Hardware costs attributed to the purchase of more blade servers or even a new blade chassis
  2. Additional costs attributed to more network equipment/infrastructure support (this includes AC/Electrical costs)
  3. Software license/ongoing license/maintenance subscription costs, per-annum

To negate this, quite frankly, excessive CapEx/Opex overhead, these are steps that will ensure you reduce this as much as possible:

Baseline you environment - “Never leave that till tomorrow which you can do toady” - Benjamin Franklin. Identify the issues on a per-cluster basis. Clean up major problems and grab the low-hanging, easy fix fruit.

Utilize the analytics in VirtualWidom - The VM Coordinator/Seasonal Trend/Capacity Forecaster analytics can empower you to balance and predict, with firm accuracy, what you should do to alleviate problems now.

Evolution - VirtualWisdom will allow you to catch changes prior to an issue, using a precautionary approach. Virtualization patterns change as application changes are made, and rebalancing will need to be done over time. Staying on top of this, pro-actively, can negate fire drill events. No-one wants that 2am phone call!

Reassess purchasing - Based upon the recorded data/time comparison shown in VirtualWisdom and how much overhead each component has in the environment that the application has access to, can I 2x my current workload/etc. Virtana even has a service where we can record your environment, then play it back in a lab using WorkloadWisdom. We can stress test your lab using your real-world data. Not synthetic, not generic replay of data, actual application data pattern testing.

My question to all is; What have you found, you spend the most of your datacenter CapEx/Opex budget on?