Some of the answers will be around personal preference / environment / etc.
But a few points here:
Q1: Our datastores are starting to fill, and I need to present more storage from the Unity array. Should I be looking to extend the existing datastores, or is 3TB and 20VMs per datastore a nice size to standardise on (outside of the larger datastores for single VMs)?
A1: There is no size recommendation that I can remember. This is personal preference mostly (and best practices as well). There are pros and cons on both sides of larger vs smaller (but more quantity). Example: one might be better for fault tolerance/redundancy, or isolating perf issues, etc, than the other.
Q2: There is a lot of good information out there on data reduction, however am I worrying too much about this, and should just enable this for all datastores?
A2: Better not, this depends on many factors. Please follow the recommendations provided in the best practices docs. Enabling on all without knowing if they are suitable might result in (possibly high) overheads that return no real benefits to you (for your specific environment).
Q3: How much, if at all, should I think about matching workloads (OS, application type, etc.) to certain datastores?
A3: The underlying pool is likely the same so it may not matter too much. If you have different pool for different workloads (if very different) then yeah. Again, it depends a lot on the setup/environment.
Q4: Finally, is block iSCSI best for this environemnt, or should I be looking at NFS or even VVols?
A4: I can't possibly answer that as I don't know your environment. From experience working with block, FC is a lot simpler and less problematic than iSCSI. If there is no FC, then iSCSI may be simpler than NFS/vVOLs, yes. But again, this depends a lot on your specific usage and requirements. We have a lot of users on FC/iSCSI, and substantially less using vVOLS, and only a very small fraction use NFS datastores for VMware.
Your questions here are good questions, and show you are looking well after your environment. We have specialized teams that can take a deep look into your environment and provide further feedback. You can work with your local Dell EMC team to engage these teams.
AndreD Dell
Moderator
•
239 Posts
0
March 12th, 2019 05:00
Hi mistersparky,
Some of the answers will be around personal preference / environment / etc.
But a few points here:
Q1: Our datastores are starting to fill, and I need to present more storage from the Unity array. Should I be looking to extend the existing datastores, or is 3TB and 20VMs per datastore a nice size to standardise on (outside of the larger datastores for single VMs)?
A1: There is no size recommendation that I can remember. This is personal preference mostly (and best practices as well). There are pros and cons on both sides of larger vs smaller (but more quantity). Example: one might be better for fault tolerance/redundancy, or isolating perf issues, etc, than the other.
Q2: There is a lot of good information out there on data reduction, however am I worrying too much about this, and should just enable this for all datastores?
A2: Better not, this depends on many factors. Please follow the recommendations provided in the best practices docs. Enabling on all without knowing if they are suitable might result in (possibly high) overheads that return no real benefits to you (for your specific environment).
Q3: How much, if at all, should I think about matching workloads (OS, application type, etc.) to certain datastores?
A3: The underlying pool is likely the same so it may not matter too much. If you have different pool for different workloads (if very different) then yeah. Again, it depends a lot on the setup/environment.
Q4: Finally, is block iSCSI best for this environemnt, or should I be looking at NFS or even VVols?
A4: I can't possibly answer that as I don't know your environment. From experience working with block, FC is a lot simpler and less problematic than iSCSI. If there is no FC, then iSCSI may be simpler than NFS/vVOLs, yes. But again, this depends a lot on your specific usage and requirements. We have a lot of users on FC/iSCSI, and substantially less using vVOLS, and only a very small fraction use NFS datastores for VMware.
Your questions here are good questions, and show you are looking well after your environment.
We have specialized teams that can take a deep look into your environment and provide further feedback. You can work with your local Dell EMC team to engage these teams.
Hope this helps