StarWind Software

I am ~10 days into a trial of vSAN standard and unfortunately haven't got it configured yet. I called my sales rep yesterday to try to bounce a few technical questions off of him and help get me moving in the right direction, but only got his VM & haven't connected with him yet.

So here are my questions (and I apologize in advance for how elementary they are, as this will certainly reveal my great degree of ignorance with the product):

1.) When I first installed the software, it immediately prompts me to setup the storage pool. It tries to defaul this to "My Computer\E". It also allows me to select an alternate path.
a.) Is v.SAN going to reformat the path/volume that I point it to?
b.) Doesn't it actually want a physical disk (block device)? Or does it only want/need an existing NTFS partition to sit on top of?

2.) Does StarWind vSAN work directly with blocks (as its back-end storage), or does it just write its virtual disk files (.img & .swdsk) on top of existing NTFS formatted volumes?
a.) The same question is of particular interest in relation to LSFS.

3.) Can the storage pool ever be changed (or could we have multiple storage pools)? I'm basically asking for the purpose of whether we were starting off using an existing QNAP iSCSI as our block device on the back-end, but later we may use something else (including a direct-attached storage enclosure like I've seen sometimes recommended with StarWind).

4.) Is there really any reason to not use LSFS if I am going to be running many VMs? These VMs would run a combination of application servers, file servers, backup servers, domain controllers, SQL servers, Exchange servers, etc. (the backup servers are the only VMs where I would expect to see more common sequential i/o. Most of the rest I would expect to be pretty random i/o other than during system boot time, etc.).
a.) Wouldn't LSFS really give us the best performance for this type of workload?

5.) Is it recommended that we do use hardware RAID on the underlying physical storage, before providing our block storage devices to v.SAN? I think I have seen some conflicting (or at least confusing) information between the forums, various discussions with your sales reps, etc. in relation to JBOD vs Dell Direct Attached Storage. And I have also spoken with engineers of a competing storage hypervisor that stated in some cases he prefers to *not* us RAID for the underlying storage, as long as he has 2-3 replication nodes. This is because it eliminates RAID rebuild time, and instead you can just replace the single disk and the adjacent storage nodes will replicate the necessary blocks to the replacement disk (this would be much faster than waiting for a RAID array rebuild). So I would really like some clear and strong recommendations related to RAID vs. JBOD for the underlying disk configuration.

6.) Can we use RAM & SSD caching with only a single v.SAN storage node? Of course this would be more dangerous, and in reality having a replicated storage node (at least 2 nodes total) makes RAM & SSD caching much safer to do.

Thank you so much for assistance answering these questions. I am really looking forward to becoming proficient with StarWind vSAN and hope it will be a fantastic storage platform for my business as well as several of my customer's own storage environments.

Sincerely,
-
Doug Mortensen
Impala Networks, Inc.

Hello impaladoug,

I've had the same question when I've started to use StarWind. I will try to answer some of them.

1. a.No, StarWind doesn't reformat the path/volume. StarWind creates image file on top of existing file system.
b. Starwind needs an existing NTFS partition.

2. The answer to these questions is one - StarWind writes virtual disk files on top of your NTFS formatted volumes.

3. In the begining you choose default location of your storage pool. You can change it anytime you want. Also, you can choose another path while you creating a device.

4. I think it depends on your IOs. Different number of IOPS requires different configs. Try to look at Veeam One Report and share it with me and I will try to help you.

5. Speaking about RAIDs, using it or not depends on your system requirements. Your choice should meet your RPO/RTO objectives.

6. No. I can't see any reason to do it

lxlive , thank you very much for your contribution.

Any further questions left here?

So just for clarification, StarWind VirtualSAN always sits on top of an existing NTFS volume? Even when using LSFS? Being new to the product, I assumed that you work with raw blocks like DataCore. And given that LSFS is called "filesystem", I assumed it woukd be sitting on raw blocks as its own filesystem (as opposed to something like NTFS). So no matter what, you always sit on top of NTFS (even when using LSFS)?

Yup, it is.

No performance drawbacks here and more reliable to me since you can treat these starwind virtual disk files like normal files.

darklight wrote:Yup, it is.

No performance drawbacks here and more reliable to me since you can treat these starwind virtual disk files like normal files.

Interesting. Thanks for sharing!!

Thanks for your priceless contribution, guys.

Any questions left here?