- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
What are the VMware or Microsoft best practices that I need to know to make sure that the NTFS file server that I have created as 2x Windows Server 2008 Enterprise SP2 MSCS (Active / Passive Cluster) can work reliably to be accessed by more than 1000 people?
The underlying data disk is RDM because I need to replicate the LUN into the DR site using the SANCopy array tool.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
I do not know if you have reviewed VMware's guide for MSCS - http://pubs.vmware.com/vsphere-51/topic/com.vmware.ICbase/PDF/vsphere-esxi-vcenter-server-511-setup-...
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Does the following enhancement will be noticeable or muchly improve reliability and performance of the production VM:
- vNIC changed from e1000 into VMXnet3
- SCSI controller changed from LSI Logic SAS into Paravirtual
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Well you will need to have RDMs in physical compatibility and the scsi controller configures with physical bus sharing, anyway if you want to cluster across boxes.
Also its worth noting and has caught me out when first moving to 5.1 is if your storage is presented as an active/active pathing. It will automatically configure round robin. unfortunately round robin is not supported with use on MSCS. and you will see terrible performance so they will have to be set to fixed. (vSphere 5.5 fixed this and round robin is supported for MSCS RDMs)
I don't think you will see or notice a performance difference making those 2 changes. its just better practice. vmxnet3 is specifically designed NIC and driver for visualization and the SCSI controller wont make too much of a difference from what i've seen.
I have about 300 MSCS clusters in total running on one of my larger client sites, with differing configurations as we have moved through versions over the years.
but its always best to follow the current recommendations if possible.
Cheers
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
- SCSI controller changed from LSI Logic SAS into Paravirtual
Not really an option..... This isn't supported by VMware in a MSCS setup..
/Rubeck
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Cool, thanks all for the comments and the suggestion.
I'll check on the multipathing policy since I have successfully migrating the VM from ESX 4.1 into ESXi 5.1u1
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
NuggetGTR,
Thanks for pointing me the article that specifically said that Round Robin is not supported in MSCS VMware KB: Microsoft Clustering on VMware vSphere: Guidelines for supported configurations ?
I just found out that it works well in the existing configuration (client site that I work on).
in what way that it is considered not supported ?
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
In all MSCS setups I have seen when its on round robin you will find that disk cmds queue in the kernel every time is swaps paths. It will still work and function it just does not perform as well as it could.
If you have the MSCS virtuals setup for cluster across boxes and have round robin PSP applied on the RMDs look in ESXTOP and see if you get periods of queuing.
If it doesnt have this effect on your hosts then great, but either way its one of those things where the vendor wont support a particular setup but doesnt mean it wont work. If you were to call VMware up because you were having issues they most likely wouldnt look at it until it was running on a "supported" setup.