Tiering is Like Tables, or Storing in the Cloud Tier

We take tables for granted. Really take them for granted. We cover them with our stuff, we sit down to eat on them, we put them in front of the TV to hold our computers, we put them pretty much  everywhere, and we use them for everything from holding collections of important papers we will eventually get around to sorting to using them as workbenches to hold stuff down while we saw. In Lori and my house alone I could fill a blog with the things we use tables for.

And yet we never see them when we’re not interacting with them. Unless we bought really cheap ones or have gone literally years without maintaining them or put them together wrong when we first got them, we never fear that they will fall apart randomly on us, and we never fear that they just won’t be there tomorrow. We interact with them on a daily basis, we put them to uses they were not designed for, and still we don’t worry that they won’t be there. And while tables have been specialized a zillion different ways, and yet we interact with them much the same way no matter which style a given table is.

And that is exactly where we need our storage to be moving forward. I should not have to care about NAS vs. SAN anymore, we’re in the twenty-first century after all, and storage is as important to our systems as tables are to our lives. And in many ways they serve the same purpose – a place to store stuff, a place to drop stuff temporarily while we do something else… You get the analogy. Let us face basic facts here, an Open Source group (or ten) was able to come up with a way to host both proprietary OS’s and Open Source OS’s on the same hardware, determining what to use when, and that’s not even touching on the functionality of USB auto-detection, so why is it that you can’t auto-detect whether my share is running CIFs or NFS, or for that matter in an IP world, iSCSI?

For all of my “use caution when approaching cloud” writing – that is mostly to offset the “everything will be in the cloud yesterday!” crowd - I do see one bit that would drive me to look closer – RAIN (Redundant Array of Independent Nodes) plus a cloud gateway like that offered by nasuni offers you highly redundant storage without having to care how you access it. Yes, there are drawbacks to this method of access, but the gateway makes it look like a NAS, meaning you can treat your cloud the same as you treat your NAS boxes. No increase in complexity, but access to essentially unlimited archival storage… Sounds like a plan to me.

There are some caveats of course, you’d need to put all things that were highly proprietary or business critical on disks that weren’t copied out to the cloud, at least for now, since security is absolutely less certain than within your data center, and anyone who argues that point is likely selling cloud storage services. There aren’t other people accessing data on your SAN or NAS boxes, only employees. In the cloud, others are sharing your space. People you don’t know. That carries additional security risk. End of debate.

But with a File Virtualization product like our ARX, you could easily position proprietary or sensitive info on disk that has no third tier, while everything else is on disk that has three tiers – primary, secondary, and tertiary… With the tertiary being provisioned from the cloud through a box like the nasuni gateway. Of course, storage has some rocket science bits, and thus it is not guaranteed that ARX works with nasuni… Since they’re a newer player and I’ve not heard anyone else propose this type of solution, I’m guessing our test team hasn’t yet run compatibility testing with them – or even considered doing so. Though if the LAN side of the gateway is standards based, there is no reason ARX doesn’t just plug-n-play with them.

Why would you bother? Simple. The tertiary tier could be your offsite backup for less critical data. No need to build a data center, no need to put a bunch of custom cloud interfaces into place, farm it off to the cloud and forget about it. No worries about transporting it to an abandoned mine… Or that they’ll misplace it. Then you have your infrequently accessed stuff on slower media (it’s in the cloud, latency alone says slower), and it need not be backed up. That’s a huge chunk of your data that is automatically cared for.

Just blue-skying how you can have your table and eat at it too, it sure would be cool if cloud started making our storage standardized, it’s 2010 and we shouldn’t be excited that major vendors are finally implementing unified look-n-feel management interfaces for NAS/SAN, so I’m looking for what will be truly exciting instead of a decade or more overdue. This might just be it.

If you could back up a huge chunk of your unstructured data simply with a shadow copy or robocopy and be as assured as possible that it was protected simply by nature of how RAIN technology works… And you could do that without having to work through a specific cloud API, and you could do that through a device that would locally cache frequently accessed files until you could move them off to alternate storage… Well, that wouldn’t be a panacea, but it would certainly be a step in that direction. Lori, ever the cloud buzzword manager, immediately coined it “the Cloud Tier” when I was discussing this with her… So I herewith dub it.

Table pictures from www.treehugger.com,

Array picture CC-BY-SA Michael Moll