Claus's Blog

One of the things I love to see is integration. We’re constantly seeing new technology coming to market, and one of my favorites is our Hitachi Dynamic Provisioning (HDP), which includes thin provisioning, ease of provisioning, and great performance improvement for random reads.

On the thin provisioning front, we’re certainly not unique, since most other storage vendors have some form of thin provisioning, or at least claim to. But just like HDP, thin provisioning can be broken down into multiple components, and this is where implementations vary.

Firstly, TP in most implementations, works as advertised for newly provisioned storage. It doesn’t always work as advertised for existing data moved into a TP environment, except for HDS, which has a feature called Zero-Page-Reclaim (ZPR, or “zipper”) that can return unused pages of storage to the free space pool. We’re seeing a 30%-40% reclamation for previously provisioned storage moved into an HDP “pool”.

Secondly, and this also varies by implementation and vendor, is what happens when you add drives the pool of TP storage. In the HDS case, we initiate a background task to rebalance all the hard drives but at a LOGICAL LUN LEVEL (pardon the all-caps, I don’t want to appear to be yelling, but it’s an important point). It’s critical that this be done to maintain the performance benefits of HDP. I think we’re totally unique on that.

But thirdly - and this is where “a plan comes together” – is integration with the file systems. Of all the great things TP brings to the table, it does not inherently address what happens when data is deleted. Yes, that happens. By integrating with the file system, we can maintain the TP environment.

This week, (December 7th), Symantec announced their “completing the thin provisioning ecosystem” that includes the necessary API calls for the file system to “notify” the storage array when space is “deleted”. The interface is a previously disused and now revised/reused/repurposed SCSI command (called Write Same) which was jointly worked out with Symantec, Hitachi, and 3PAR. This command allows the file systems (in this case Veritas VxFS) to notify the storage systems that space is no longer occupied. How cool is that! There is also a subcommittee to INCITS T10 studying the standardization is this and SNIA is also studying this. It won’t be long before most file systems, databases, and storage vendors adopt this technology.

So here’s the deal, and help me out here, folks. So HDS supports “Write Same”, now. And although 3PAR have announced their support, it was not clear to me whether they actually support this in a deliverable product yet. 3PAR, do you currently have a GA product that supports the VxFS feature? I hope so! Same with IBM XIV, do you also currently support this feature? Information on your respective websites is a bit fuzzy. I doubt EMC supports “Write Same” now. But help me out, here, guys. Let’s see the line up of which storage vendors currently support “Write Same” and which ones will.

But the command, and support by VxFS and HDS, is the first step in bring thin provisioning to its final form. Other file systems will follow as will other storage vendors. Watch this space over the next months and watch this finally “come together”.