3 CPUs and 12GB RAM is the New Tier-1 App

In what can only be regarded as a breakthrough in hypervisor performance technology, it is now possible to run tier-1 applications on minimal resources and without regard for high-availability of any kind!

A fascinating new report from Enterprise Strategy Group reveals that one particular hypervisor is suddenly capable of running tier-1 applications on just 3 virtual CPUs and a mere 12GB RAM — power and efficiency at its finest.  Not only that, the New Tier-1 App no longer requires application-level clustering to meet stringent SLAs!  Simply throw your mission-critical SQL Server database on a stand-alone hypervisor and skip the shared block storage entirely — you just don’t need it anymore.

Comprehensive lab scalability testing has demonstrated and proven once and for all that it is indeed possible to put not one, not two, but eight mission-critical tier-1 applications on a single stand-alone hypervisor host with no need for HA protection of any kind.

Beyond the obvious cloud computing use cases one would envision NASA and other high-performance compute environments benefitting immensely.  Customers using other hypervisors can only hope their vendors are feverishly working to provide similar performance and availability features so that the remaining universe of tier-1 applications can also be effortlessly virtualized on just 3 CPU cores and 12GB of memory.

The timing for this breakthrough is actually perfect, as SQL Server 2012 introduces a new per-core licensing model that would have sent costs through the roof if not for this remarkable advancement — bullet dodged!


(Visited 342 times, 1 visits today)
This entry was posted in Virtualizationism and tagged , . Bookmark the permalink.

9 Responses to 3 CPUs and 12GB RAM is the New Tier-1 App

  1. Alex Harden says:

    Hey, at least they configured RAID groups on the disks. 😉 Yes, it’s pretty sad to see a non-clustered Tier-1 recommendation. They probably tried to cluster Hyper-V and failed (like I did when I last looked at it).

    • tonyr says:

      hmm whats so hard about clustering hyper-v, I have a single powershell script that completes the task quite easily, of course the SAN\NETWORK work needs to be done first, other then that its a 5 minute task and completes every time.

    • tonyr says:

      as a sidebar to this since you had such a difficult time creating a cluster with w2k8r2, well guess what with windows 2012 nothing has to be shared and live migrations and storage migrations just work no special requirements… and its all free.

  2. KJL says:

    Talk about taking something out of context…the ESG report is NOT a best practices or deployment guide. Perhaps if you read the title, “Workload Performance Analysis” you might understand the intent of the document.

    Of course no one is going to run Tier-1 applications without some type of HA. They clearly call out that HA is available in every version of Hyper-V…including the free one.

    Is that what VMware is stooping to now? Times must be hard…about to get a whole lot harder when Server 2012 hits the streets… How’s your resume looking? 🙂

    • Eric Gray says:

      The report essentially tries to demonstrate that because they were able to run several small VMs on a blade server, Hyper-V would have no trouble with a tier-1 application.

      If they really wanted to show how well a tier-1 application performs, they could have configured a small number of large VMs. E.g., 8 or 16 CPU cores with corresponding memory. The problem here is that Hyper-V only supports 4 CPU cores.

      Furthermore, without shared pass-through disks, in-guest clustering must be done with in-guest iSCSI. Nothing to do with hypervisor HA – think Patch Tuesday. Not the first thing that comes to mind for business critical applications.

      Anyway, I don’t agree that it was out of context – the “tier-1” reference was splattered all over this paper.

      • sam i am says:

        why would I need to configure my exchange workload with 8 or 16 CPU cores? or a majority of my SQL workloads?

  3. KJL says:

    I understand the point you are ‘trying’ to convey. That would be valid if there weren’t already thousands of companies already doing what you are mocking. One of the case studies my own company looked at when putting Hyper-V in (along side VMware, for now) was Target since their implementation is simliar to what we were looking to do. Every pack of M&M’s you buy goes through a Hyper-V server – including some of the applications which run Linux.


Comments are closed.