I just read a great post on the Virtual Geek’s blog about the combination of VMware’s VMDK thin provisioning and array based provisioning such as offered by TP leader 3PAR. (disclaimer, I work for 3PAR). It is a fantastic asessment of the potential benefits and risks of using different combinations of thick/thin VMDKs in conjunction with thick/thin arrays.
Lots of people have been asking about the merits of using VMware’s thin provisioning on top of 3PAR’s array based thin provisioning. Virtual Geek said it very well, so I’ll give credit where credit is due:
So… What’s right – thin provisioning at the VMware layer or the storage layer? The general answer is that is BOTH.
The flexibility and efficiencies that are achievable when using both is fantastic. You can cram more VMDKs into a VMFS without consuming tons of disk. Array based TP will very quickly dedicate on write so disk is consumed only when it is actually required.
By reading the disclaimer on Virtual Geek’s blog, he is an EMC employee – I can’t tell if he works for EMC proper or for the VMware “arm” of the company. I only say that because I am impressed with his VMware subject matter expertise. Nice writeup VG!
I was a bit suprised at the end of the article though – it started raising issues that local (or host-based RAID direct attached) has with IO locality and data layout. Anyone considering array based TP must be thinking about a pretty slick networked storage solution, not DAS. I was also suprised by the first comment posted by Storage Anarchist (another EMCer – and that’s 100% OK by me). What surprised me is that Anarchist states: “That latter “exception” [Rob: IO locality] doesn’t really apply to an intelligently cached disk array like the Symmetrix DMX and V-Max.”
While I completely agree with Anarchist’s premise that IO locality is not a material issue for an intelligently cached disk array, I am surprised by the unwritten message that only uber-expensive high-end arrays alleviate this issue. Only Symmetrix DMX and V-Max offer this? I can think of several mid-range priced arrays that are leaders in array based thin provisioning (and have brilliant caching and data layout schema) that perform extremely well in VMware environments. [remember I work for 3PAR 
]
What say you? Is Thin Provisioning from monolithic [DMX] or neo-monolithic [V-Max] the only way to go here? I disagree.
