Insufficient data from Andrew Fryer

The place where I page to when my brain is full up of stuff about the Microsoft platform

Meteorology and Cloud Control

Meteorology and Cloud Control

  • Comments 1
  • Likes

The word we us for predicting the weather came from the guys who used to try and predict when meteors and comets would appear which in turn was supposed of influence events here on earth. Nowadays the Met Office are actually getting a lot better at it and attempts have also been going on for the last 50 years to also try and control it e.g. limiting the power of tornados or encouraging rain in arid areas. 

When it comes to IT cloud control might seem irrelevant as the provider does all the administration for you when you provision a service like Azure EC2, Google, or V-Cloud.  However while the tin is taken care of the rest is still down to the good old IT professional and this was bought home to me recently during an excellent talk at the London Vmware user group by a speaker who I will refer to as  Stu (he wishes to remain anonymous and I did get his permission to blog this of course)

The key points of his Virtual First presentation that apply to pretty well any cloud provider are..

  • Need to have well planned checklist to deploy a vm to cloud
  • Need to understand human logic in the current process e.g. manual process for deciding IP and DNS settings
  • Self service doesn't not mean no governance
  • Challenge convention
    • Cloud is seen to be much cheaper because there is no support (e.g. EC2). so offer a “no support” option for vms internally so that like can be compared with like.
    • Don't have Server Naming convention use the tools in Vmware or Microsoft’s  System Center Virtual Machine Manager (which can manage Vmware and Hyper-V vm’s)
  • Think about offering local admin support but be aware other support teams may be unhappy with this e.g. those supporting the applications on top of the operating system.
  • There’s no standard unit of compute so you don't know what to provision in the virtual world i.e. user needs to prove they need to stay physical so measure and decide on the basis of evidence (old hardware vs new hardware) A good resources is rpe 2  which gives a performance index on a huge range of physical servers.
  • Keep it Simple. Don't go mad with complexity – the chances are the business will bite your arm off anyway as in a virtual world provisioning a workload is possible in a couple of hours rather than 2 months. For example provide Simple  fro your offerings e.g.  high medium low performance vms with or without support.
  • Think about decommissioning workloads to keep the costs of cloud down or the local data centre operating at maximum efficiency.
  • Think about scripts (PowerShell/Vmware PowerCLI) and requesting a new or change to VM to meet change in demand.
  • External issues don't think about physical stuff remember ip address for example tag workloads that have a regulatory requirement like data protection HIPAA, PCI

So I go to the Vmware user group to find out best practice pick up any issues the community have with Microsoft solutions running in a virtual world and thus support the IT professional in the UK even if we differ on the virtualisation platform we use.

Comments
  • Hi Andrew,

    Thanks for the kind words! i'm really sorry I ran off after only half a conversation with you afterwards, i looked for you but couldnt find you again :(. Anyways, I've kick off a series of posts on my blog over on www.vinternals.com to try and put that talk into words, any feedback from you is most welcome :)

    Stu

Your comment has been posted.   Close
Thank you, your comment requires moderation so it may take a while to appear.   Close
Leave a Comment