Nutanix

Discussion in 'Business & Enterprise Computing' started by GooSE, Oct 4, 2013.

  1. scottath

    scottath Member

    Joined:
    Aug 16, 2007
    Messages:
    2,575
    Location:
    Sydney
    Does anyone running VMWare on Nutanix have a clustered Windows file share?

    Looking at consolidating some file servers rather than doing a P2V on them in addition to the other working servers.
    Rebuilding them would be the plan with transparent failover on the file shares between servers(hosts).
    Mixed Windows/OSX environment currently accessing S2008R2 servers (non clustered - server per share).

    I built a demo cluster using shared disks it was a POC, and not production, however that isnt at all best practice, also cannot be backed up by Veeam in that configuration (vmdk shared between VMs using bus sharing)

    Thanks,
     
  2. millennia

    millennia Member

    Joined:
    Dec 10, 2013
    Messages:
    19
  3. scottath

    scottath Member

    Joined:
    Aug 16, 2007
    Messages:
    2,575
    Location:
    Sydney
    I have read through that article a number of times, the article threw me when it said about having the Quorum on an SMB accessed NDFS container.
    Just trying to get my head around the how.

    Windows VMs speccd as needed.
    2x Nic (one on the 192.168.5.x, other on production network)
    Add NFS/Clustering roles, create/mount a new container - inline dedup + compression likely
    Create folder structure on the mounted NFS, make quorum apart of that structure.
    Add VMs to the cluster, setup the file shares as you would normally then?
     
  4. Iceman

    Iceman Member

    Joined:
    Jun 27, 2001
    Messages:
    6,647
    Location:
    Brisbane (nth), Australia
    You *can* build many things. The question is; should you?

    What are you trying to achieve by building a virtual cluster within a hardware cluster?
     
  5. PabloEscobar

    PabloEscobar Member

    Joined:
    Jan 28, 2008
    Messages:
    14,638
    Fault Tolerance at the application layer I assume.
     
  6. scottath

    scottath Member

    Joined:
    Aug 16, 2007
    Messages:
    2,575
    Location:
    Sydney
    This is what im figuring atm, 1) if its a good idea 2) Practicality.
    Correct, Move the file share from being reliant on 1 machine (currently a physical server) to being redundant.
    Not saying its needed exactly - as being a school i can reboot things nightly if i wanted - but looking to best practice with FT file shares. And further to this, FT file shares on Nutanix.
    Im aware that this is overkill for the scenario it's being applied to.
    The same will be applied to an SQL server in the coming months. Similar use case/load/requirement to the file shares.

    With HA running on the cluster, it'd be running again in a few min at most on another host, but having the service not needed to be rebooted would be better.
     
    Last edited: Jun 5, 2015
  7. PabloEscobar

    PabloEscobar Member

    Joined:
    Jan 28, 2008
    Messages:
    14,638
    Terms I'm using;

    Fault Tolerance (FT) - Guest/Host/Service running on server(s) can Die in the Ass - Application keeps working with no interruption to the user

    High Availability - Guest/Host/Service dies in the ass, New Service is bought up as soon as practicable in an automated manner. Minimal (but non-zero) user interruption.

    For things that can be made fault tolerant at the application layer... then I'd be doing it at that layer.

    I would use Hypervisor HA... so that if a host dies, and takes out one of my application server instances... Application Layer FT keeps the service alive. Hypervisor HA bring up that application server instance on another host, and Fault tolerance is established again. HA Rules can be used to ensure that each part of a Application FT Cluster are not on the same host.
     
  8. scottath

    scottath Member

    Joined:
    Aug 16, 2007
    Messages:
    2,575
    Location:
    Sydney
    Sorry yea, i typo'd in the above.

    Looking to run a FT file share(s) (and sql eventually), on HA'd VMWare hypervisor across the 3 node cluster, with as you said, appropriate rules to keep servers apart from the same node.

    No interruption is the goal im looking for. We have more than enough hardware/resources to cater for it, so i figured as we are moving the servers from physical anyhow, making them a FT service should be looked at. Thus my above question on the how. Sorry for the roundabout.
     
  9. scottath

    scottath Member

    Joined:
    Aug 16, 2007
    Messages:
    2,575
    Location:
    Sydney
    Sitting up watching the twitter feed go nuts as the Nutanix keynote is on.
    Going to have plenty of reading to do tomorrow, but it looks really good :D

    EDIT - Wow, soooo many things, and VMWare seems to have gotten a bit of a hiding from it/lack of coverage. Migration from/to hypervisors and public cloud with one click too!!!
     
    Last edited: Jun 10, 2015
  10. closed_gate

    closed_gate Member

    Joined:
    Oct 21, 2004
    Messages:
    737
    Location:
    Brisbane
    Yeah, it's pretty sweet.

    Also, the first lot of NPX's were announced. Congrats to Josh Odgers from Melbourne on NPX#1.
     
  11. PabloEscobar

    PabloEscobar Member

    Joined:
    Jan 28, 2008
    Messages:
    14,638
    From the Rants thread...

    I'm not sure the Nutanix/VMware pissing contest does either of them any favors... If anything, for me at least, it goes to show that there is actually such a thing as 'bad publicity'.

    Both companies ask to 'view' independent benchmarking before publication, makes me mistrust any performance figures that I see thrown about.

    From what I've been reading, there seems to be a big push on by Nutanix away from VMware, which may leave quiet a few shops that bought in when Nutanix was friendly with VMware, with a bad taste in their mouth now that Nutanix seem to focus development towards KVM.

    If you mention Nutanix in a thread 3 times, or say VMWare out performs them in any way, Reps get summoned to the thread as-if by magic :).
     
  12. bsbozzy

    bsbozzy Member

    Joined:
    Nov 11, 2003
    Messages:
    3,925
    Location:
    Sydney
    VMWare may have started that by not inviting Nutanix to certain VMWare events.....

    VMWare claim they have it better as their hyper-converged solution is part of the kernel whereas Nutanix is a separate memory hungry VM.

    I've seen their KVM stuff in action, nice and easy, all from the one web console (and much better than the VMWare web interface, grrr...)
     
  13. PabloEscobar

    PabloEscobar Member

    Joined:
    Jan 28, 2008
    Messages:
    14,638
    I've seen things that a better than the VMWare web interface after my morning constitutional.

    It's not as bad as it was though :). A bit like a polished turd...
     
  14. scottath

    scottath Member

    Joined:
    Aug 16, 2007
    Messages:
    2,575
    Location:
    Sydney
    Its still hardware abstraction either way, so they will still work, but i think the general focus of Prism will be to KVM, and the storage platform underpinning it all will continue to be agnostic as it is currently. Keeps it open to choice of hypervisor, gives people with enough requirements the ability to pick what they like for hypervisor/server when they want, and add/remove nodes as they wish.
    Kernal vs User space is the whole debate. Both have advantages, however Nutanix is set on saying that user can be as good as kernal, and VMWare will not admit it - and want to fight about it.
    Im liking the prospect of flexibility, and cannot wait to play with CE on some older servers i just replaced with Nutanix. Now just need Veeam to come onboard with KVM backups. From their attendance at .Next though, and the announcement of v9 of their product, bringing DRaaS guessing that wont be far off - and maybe with Nutanix as the back end for it?
     
  15. PabloEscobar

    PabloEscobar Member

    Joined:
    Jan 28, 2008
    Messages:
    14,638
    A neither side wants independent testing done.
     
  16. Iceman

    Iceman Member

    Joined:
    Jun 27, 2001
    Messages:
    6,647
    Location:
    Brisbane (nth), Australia
    Why do you continually assign sinister motivation to this?
     
  17. PabloEscobar

    PabloEscobar Member

    Joined:
    Jan 28, 2008
    Messages:
    14,638
    Because I don't see any non sinister reasons for it.

    Free speech, freedom of the press and all that jazz. Both companies have well publicised blogs and evagalists that offer them a much wider reaching right of reply should they wish to point out any issues with the testing methodology.
     
  18. Daemon

    Daemon Member

    Joined:
    Jun 27, 2001
    Messages:
    5,471
    Location:
    qld.au
    It's to "control" the results which are searchable on the Internet. VMWare have been quite aggressive in the past by sending legal letters to anyone who published results without their approval.

    Nutanix it seems is no different in this regard, which is disappointing.

    I can understand the want to ensure the test environments are correct but at the same time they could easily provide a rebuttal post listing what they did wrong. A good tech journo / blogger will want to correct any mistakes, nobody (unless paid off by another company) will deliberately go out and published falsified data.

    Trying to hide the opinions and data from others just screams "we have something to hide".
     
  19. scottath

    scottath Member

    Joined:
    Aug 16, 2007
    Messages:
    2,575
    Location:
    Sydney
    Both sides afraid of this, or both sides expecting each other to make that happen as they are directly competing in the same (SAN) space with differing models?

    As said prior though, VMWare have been doing this prior too... not sure how long though
     
  20. Iceman

    Iceman Member

    Joined:
    Jun 27, 2001
    Messages:
    6,647
    Location:
    Brisbane (nth), Australia
    Why is it unreasonable to want to avoid any tom dick and harry publishing misleading information about your product because they didn't understand it well enough to benchmark it properly?

    Do you really believe these companies have time to; locate every review of their product, identify who wrote it, communicate with them, get that person to detail their testing environment and methodology, work with that person to *understand* the niche product/app/case they're testing, review what might be causing that behavior (say vmware or the shit way the app has been coded), convince the guy to re-run the test, convince the guy to re-publish the test saying his methodology was wrong for instance and then attempt to get that person who read the original review??

    If so, I believe you're taking a stroll in fantasy land.


    HAH even if you were able to go as far as getting them to give you the information to prove them wrong; getting someone to publish a retraction stating they fucked up their testing methodology and are in fact to blame for the results making themselves less credible in their readers eyes?

    Good luck with that one.

    In your opinion. Which would be based on how many times you've been knocked back?

    Seriously. Point us all to someone who has a valid test case that was refused the right to publish the results?
     

Share This Page

Advertisement: