The push toward virtualizing bigger and more mission critical applications in today’s IT environments is emphasizing the importance of maintaining high availability. After all, if mission critical applications go down it could range from a loss of revenue to a sudden halt in business productivity depending on the workloads. Thankfully, most hypervisors support HA features that allow us to create rules or impose limitations on resource utilization so that in the event of a failure, there exists spare capacity. While HA is great for uptime, maintaining it is a difficult problem.
Is Your HA Compliance Just Another Threshold Creating Yet More Alerts?
We often wait for our environments to leave HA compliance; whether it’s an alert that VMs are unable to power on, or insufficient capacity notifications, the cluster has drifted to an undesirable state where HA configurations are not satisfied. So the question becomes among all the tasks IT operators perform daily, how do we maintain HA at all times? This problem only compounds in environments that are rapidly growing because of how quickly HA rules can be violated.
Maintain HA Configurations and Performance
VMTurbo tackles this problem by controlling environments within HA constraints while delivering the highest performance across every virtual machine perpetually. Rather than looking at HA constraints as a boundary waiting to be crossed, the constraints become the new “cliff” or “end of the world” in our eyes. Dynamically allocating and placing workloads to maintain HA configurations and performance so that IT can always support the failover of a node(s). Here is an example on compute. Let’s say we have a 4 host cluster and running it with an N+1 failover configuration. So under Admission Control I have limited CPU and Memory utilization to 75% to ensure there is excess capacity should a host fail or I have to toss a server into maintenance mode.
HA Configuration of Hosts – Use Case 1
Typically without VMTurbo, we wait for an environment to breach these values and we arrive in a state where we cannot power on VMs nor tolerate the failover of a node. VMTurbo will prescribe decisions in automation to adjust workload configurations and placement to assure host utilization never exceeds 75%. In fact, if no combination of sizing or placement decisions can limit the risk of a growing environment exceeding these values, VMTurbo will preventively issue a decision to provision another host.
Provisioning more capacity before HA values are reached means that even during a failover IT organizations can deliver high performance. Think about it, if four hosts ran at 75% and one went down, the other three would be running at 100%; a state that cannot assure performance, obviously. By provisioning capacity before environments reach their HA limitations we can always assure performance of mission critical applications even during a failover.
HA Configuration of Storage – Use Case 2
The same logic of preventatively controlling an environment below boundaries imposed by operational logic can be applied elsewhere. Another practical example of this policy is seen on reserving a certain amount of space on storage. As a safety mechanism to ensure there is always sufficient space for the growing storage demands of the apps, service, and data IT provisions, I often see customers reserving a certain amount of space on each datastore. With a simple shift of a utilization constraint, VMTurbo will keep space utilization of datastores below the specified value.
VMTurbo will always keep space utilization below 80% through placement, sizing, and/or capacity decisions; thus, removing the complexity and operational expense of managing resource utilization.
No More HA Guessing Games, Just Control
By allowing VMTurbo to control environments within HA or operational boundaries it eliminates the pressure on IT Operations to support the availability of mission critical applications. Rather than guessing what the headroom of a cluster is or how far you can push utilization now or two months down the road, VMTurbo can control performance with HA constraints automatically without any operator intervention. VMTurbo will eliminate the over/under-provisioning of hardware and time spent managing operational compliance.