What to consider when planning virtualization and server consolidation

Virtualization and server consolidations are increasing and beneficial as they reduce OPEX and give you ROI and reduce TCO however if these projects are not planned there are some hidden costs which will end up your support Team working on the support and operational issues after project finishes and performance and availability of services may get effected. Real life examples show that many IT organizations jump in virtualization environment without planning and testing and start the migration just on the directions from pre sales Virtualization/consolidation team from vendor from whom you are purchasing the hardware or 3rd party consultancy. Make sure you evaluate Virtualization and server consolidation business case in detail. IT managers think, as virtualization is hot and helps in going green it will help their IT environment too..well it will but if planned, executed and consolidated well else you will run yourself in support/operation issues leading to reduced availability and performance of your services. Also keep in mind that as it is easy to provision servers and desktops in virtualized environment sometimes it leads to unnecessary exponential increase of your computing resources as you don’t have to wait for new hardware for a new workload.



A few points which should be considered while planning your virtualization and consolidation project.



 Which existing server models in your IT environment qualify for virtualization.

 Which applications running in your IT environment are supported on the Virtualization platform you have chose.

 Which applications running in your IT environment are not supported on the Virtualization platform you have chose but your testing shows that they run fine in virtualized environment. [Caution : Question you need to ask yourself is that are you willing to Risk a critical app on a virtual machine which is not supported by app vendor. There are some old apps for which support has already ended but it is still better to run them on physical hardware however another perspective is that Virtualization of such apps can be done after proper testing and performance baselines can be compared on physical and virtual environments and if testing results are positive Go Ahead. ]

 Which machines are running applications and services which you want to be highly available. [ caution: Do you want the Virtual machine [VM] to be highly available or the service/applications inside VM to be highly available]

 Do virtualization in phases and not in one GO!

 What backup and disaster recovery strategy you are following for the VM’s and the Hosts running the VM’s. you need to identify critical services for which disaster recovery needs to be planned. Both host and Guest based backup strategy needs to be defined and corresponding backup environment. Does your existing backup solution capable to absorb virtualization and backup and restore VM’s within the defined SLA.

 Placement of your Domain controllers and you may not want to Virtualize all your domain controllers.

 Capacity planning for host and guest servers. How much physical ram, storage, Network cards, Guest VM density on Host machine needs to be planned.

 You also need to consider that if one of the host fails, many guest will have a downtime and how soon you can bring them online on a different host and what strategy will be used.

 Which Guest machines you want to club together and which you don’t and which you want to be part of cluster hosts.

 License requirements for the operating systems and applications that will be run on your guest machines and if running Hyper-v Host which Windows Edition [Standard, Enterprise or Datacenter]/ [serve core, Hyper-v server or Windows server hyper-v edition] to be used.

 One of the most crucial steps which most of the project managers don’t do as they think this is part of operations support is Performance baseline capture of your virtualized servers and applications running on those. Once the virtualization testing has been done and the machines have been migrated to virtualized production environment make sure to capture a perfmon based performance baseline of the workloads for atleast 2 weeks covering different loads and peak load. This will help in future if you see a system/application degraded performance as then you can compare the present performance with past using perfmon tool.

 Consider booting from san using thin provisioning technologies for deduplication [caution : ISCSI boot provisioning may be economical for you]

Hope you enjoyed reading this and feel free to share your experiences with us. Thanks for staying with us and visiting our Blog.