What is the general consensus on LUN set up for specific use cases?

We're going through a similar exercise utilizing a Dell PowerStore SAN but nearing the finish line. I must have asked Dell, VMWare, and Google 1 million different ways about setting up volumes (LUNs) best practice. Each time I was given a non-answer and I had to dig, research, and reach the conclusions below myself. Take them for what they are worth-- hopefully they are helpful. ​ Volume sizing / setup in our environment came to the following. For my intents and purposes, how we carved up the volumes (LUNs) did not make a heck of a lot of performance difference. However I boiled down our considerations to the following. ​ 1) Contain Failure - If a single LUN / Volume becomes overrun it will only crash the VMs contained on that LUN / Volume. ​ 2) Logical Organization - How do you want to manage / maintain these volumes? Is there a logical structure that lends itself to serviceability and ease of understanding for other engineers who may work on the SAN? ​ 3) Manageability - Do you have space to vMotion VMs as needed? Space for your templates, ISO files, etc? Can your thinly provisioned disks grow if needed? ​ 4) Reserve Resources - Do you need to reserve space for an important application, database, etc? Is it so critical that you want to reserve the space ahead of time in case of future growth? Note that w/ the PowerStore we can easily grow volumes but not shrink them. So best practice is to make the volumes only as large as needed (with a little headroom perhaps). But if you want to earmark storage for a critical VM(s) ahead of time you can. ​ 5) Protection Policy - Do you wish to take advantage of storage snapshots? You should, they also are immensely helpful with backup products such as Veeam. You can apply different protection policies (snapshot frequency, schedule, and retention) to a given volume. For VMs needing similar protection it is a natural choice to lump them together. For VMs that are low risk / impact you can dial back the snapshots quite a bit saving space. For VMs that are critical you can get several snapshots per day, these can really save the day in a pinch. ​ 6) Performance Policy - On the PowerStore this is really dumbed down to high, medium, and low resource priority. Essentially you just dictate the pecking order that only takes effect if there is resource contention. Not sure of the equivalent in the HP world. ​ As stated above we are on the last mile of our migration and we did have one LOB application giving us latency / lag where it hadn't before. We're now converting all of our VMS to use VMXNet3 NICS and Paravirtual SCSI adapters. We're also back peddling to correct the SAN being connected via iSCSI vs NVMEoF. If we had a mulligan I would have made the Dell engineer implement NVMEoF from go instead of glossing over it. Not sure how painful the conversion is going to be just yet. There is also some muttering from Dell engineers that you want to avoid cohabitating several high I/O VMs on the same volume. But I've yet to see any difference in performance between my varying volumes. ​ The de-dupe and compression on the PowerStore is insane. I couldn't believe how much it was able to squash the data down! ​ Good Luck!!