Have we just noticed that our infrastructure is not optimized and that oversized, undersized and unnecessary virtual machines are running in the infrastructure? Here are some tips to make your infrastructure more efficient: reducing the use of resources improving the performance and decreasing the costs.
Have you put in place a complete virtual machine lifecycle management?
When we talk about lifecycle management we talk about the complete cycle of a virtual machine and not just the creation of the virtual machine. The lifecycle of a virtual machine starts with the creation phase, the configuration of the virtual machine and the allocation of virtual resources, followed by the exploitation and monitoring phase of the virtual machine (its use and status) and concluded by the elimination phase of the virtual machine and the release of virtual resources.
Generally, and thanks to virtualization, organizations are very good at the first phase of the lifecycle “the creation of the virtual machine”, proof of this is the constant increase in the number of virtual machines in their infrastructures. However, the following steps of the lifecycle are usually ignored until performance problems start to increase.
In a logic of good management and best practices, the correct control of the park of virtual machines must include a proper lifecycle monitoring strategy from the creation until the deletion of the virtual machines, that allows to identify virtual machines that need to be resized (reducing or adding resources) during their exploitation phase, or deleted (releasing their virtual resources) in the final phase. (However, in reality this is not the case, as organizations used to forget about the lifecycle or in other cases because they do not know what it means for its virtualization.)
Here are some tips to help you to go through the three phases in the lifecycle of the virtual machine.
The 3 phases:
Phase 1: Creating the virtual machines
Regarding the creation phase of virtual machine, sometimes the development teams and/or the editors make unreasonable demands, with very largely oversized requirements (“belt and braces” requirements: we do not know what we need, so we ask for more, just in case).
Tip 1: restrict from the beginning the allocation of resources to virtual machines. It is always easier to add that to remove resources (the feeling of “take back is steal”,). If ever during the first days of life of the virtual machine, performance problems are noted, it is always possible to add resources and thus have an allocation adapted to the virtual machine.
In parallel, do not forget to take note of the following information: creation date, for which project, the owner of the virtual machine, end date (in the case of POC: proof of concept), and any other information that allows to understand the lifecycle and use of the virtual machine (seasonality of virtual machine, dependency, environment, etc.). This information will be useful for the following 2 phases.
DC Scope® and its virtual machine Synthesis View allows to centralize all this information effortless
Phase 2: Resizing virtual machines
Do not wait until the lack of resources in your infrastructure to resize the virtual machines. virtual machines well sized mean a better usage of resources and it translates in a best comprehension of the real capacities of the infrastructure. The cherry on the top of the cake is that the virtual machines behave better (subject of an upcoming article: “too much VCPU less performance”).
Tip 2: we advise our clients to schedule an optimization day (or even only a few hours) every quarter with the aim to resize and delete virtual machines.
In contrast to the vCenter which aggregates the information of your infrastructure in daily periods, DC Scope® keeps on its own database the information per minute on all VMware counters, so there is no loss of information. This fluidity over the time allows DC Scope® to make precise recommendations by considering all particularities of the virtual machine, as the recommendations are not based only on averages, but also considering peaks.
Tip 3: we also recommend to keep an historical of the lifecycle of the virtual machine which log all the changes in the resources allocated to the virtual machine, as well as changes on its status such as change of servers, clusters, etc. DC Scope® logs automatically any change in the virtual machines and the virtual machine Synthesis View shows easily the historical changes of virtual machines.
Phase 3: Removing virtual machines.
It is always tricky to delete virtual machines (unless you are very well organized or you have DC Scope® running in your infrastructure ;)).
For this final step, the information logged in the creation phase such as the owner of the virtual machine and the expected deletion date of the virtual machine are fundamental. In fact, it is the owner of the virtual machine who determines whether a virtual machine has completed its lifecycle according to the initial deletion date and is he who will define whether the virtual machine can be deleted or not.
Before the final deletion of virtual machines, some organizations prefer to shut down these virtual machine for a short period of time. It is extremely important to take note of this information, as it could be really easy to forget the final deletion of the virtual machines. We regularly see the unhealthy tendency to accumulate unused or turned off virtual machines. The optimization day should also be an opportunity to finally delete these virtual machines.
It is extremely hard to identify whether a virtual machine is not being used as there is always activity reported from the virtual machine which does not correspond to the real use of the virtual machine. Such activities are proper activities of the virtualization as antivirus, backup operations, VMware activity, etc. DC Scope® has a unique analysis tool to identify virtual machines that report very little use (Idle) and allows to classify them virtual machines automatically.
In addition to the management of the lifecycle of virtual machines, there are other axes to optimize virtualized infrastructures… and we will see them in future posts.