Great article, and very timely for me.
In the Virtual Disk presentation illustrations, it shows a disk cylinder with OS, Data and Logs all in one cylinder, seemingly attached to one LUN.
In researching system optimization, I've come to believe that we need to distribute each of these on their own LUN (along with TempDB, paging, and file groups.)
The next graphic showing the bottleneck scenarios shows the distributed file groups attaching to a few LUNs, though not specifically to their own LUN.
Is my general belief flawed? Is the deeper concern just to ensure that the Data, Logs, tempDB, etc. are transiting via separate HBAs? Should this happen automatically if they all live on their own separate virtual drives (vmdks)? Or is it good enough to have each virtual drive attach to individual data stores that reside on one LUN?
Thanks again for a great treatment of a complex topic!