How to build resilient, scalable storage clouds and turn your IT department into a profit center!!
If you’ve been living under a rock for the last year the topic of Cloud based computing might be new to you. Don’t worry about it at this point, there are CLEARLY more questions than answers on the subject. I get asked at just about every meeting what my interpretation of “cloud” is. I will normally describe it as an elastic, utility based environment that when properly architected, can grow and shrink as resources are provisioned and de-provisioned. It’s a move away from “silo based” infrastructure and into a more flexible and scalable, utility based solution. From a 30,000 foot view, I think that’s probably the best way to describe it. Then the conversation usually rolls to “so, how do you compare your solution to others” relative to cloud. Here is what I normally talk about.
First and foremost we have sold solutions that are constructed just like everyone else’s. Our Magnitude 3D 4000 product line is built with pretty much the exact same pieces and parts as does Compellent, NetApp FAS, EMC Clariion and HP EVA etc. Intel-based controller motherboards, Qlogic HBAs, Xyratex or other SBOD drive bays connected via arbitrated loops. Like I’ve said in prior posts, just line each of these up, remove the “branding” and you wouldn’t be able to tell the difference. They all use the same commodity parts. Why is this important? Because none of those solutions would work well in a “Cloud” based architecture. Why? Because of all the reasons I’ve pointed out in my “Performance Starved Application” post, as well as my “Cost per TB” post. THEY DON’T SCALE WELL and they have horrible utilization rates. If you really want to build a storage cloud you have to zero in on what are the most important aspects of it, or what I like to refer to as “The Fundamentals”.
First you MUST start with a SOLID foundation. That foundation must not require a lot of “care and feeding” and it must be self healing. With traditional storage arrays, you could end up with 100, 200 or even 1000 spinning disks. Do you really want to spend the time (or the HUGE maintenance dollars) swapping out, and dealing with bad disks? Look don’t get me wrong, I get more than a few eye rolls when I bring this up. At the end of the day, if you’ve never had to restore data because of a failed drive, or any other issue related to failed disks then this is probably not something high on your list of worries. For that reason, I’ll simply say why not go with a solution that guarantees that you won’t have to touch the disks for 5 years and backs it up with FREE HARDWARE MAINTENANCE (24/7/365/4hr)!! Talk about putting your money where your mouth is. From a financial point of view, who cares if you’ve never had to mess with a failed drive, it’s freaking FREE HARDWARE MAINTENANCE for 5 years!!
Secondly, it MUST have industry leading performance. Not just “bench-marketing” type performance, I mean real audited, independent, third party, validated performance numbers. The benchmarks from the Storage Performance Council are a great example of a third party solution. You can’t just slap SSD into an array and say “I have the fastest thing in the world”. Here is a great example; if you are looking at designing a Virtual Desktop Infrastructure then performance should be at the top of your design criteria (boot storms). Go check out my blog topic on the subject. It’s called “VDI and why performance matters”
Finally, you need the glue that holds all of this together from a management and a reporting point of view. WebServices is that glue. It’s the ubiquitous “open standard” tool on which many, many application solutions have been built on. We are the only company who builds its storage management and reporting on Web Services, and have a complete WSDL to prove it. No other company epitomizes the value of WebService than Microsoft. Just go to Google “SANMAN XIOTECH” and you’ll see that the folks out in Redmond have developed their own user interface to our solution (our WSDL) to enable automated storage provisioning. HOW AWESOME IS THAT!! Not to mention, WebServices also gives you the ability to do things like develop “chargeback” options which turns the information technology department into a profit center. We have a GREAT customer reference in Florida that has done this very thing. They’ve turned their IT department into a profit center and have used those funds to refresh just about everything in their datacenter.
So those are the fundamentals. In my opinion, those are the top 3 things that you need to address before you move any further into the design phase. Once your foundation is set, then you can zero in on some of the value added attributes you would like to be able to offer as a service in the cloud. Things like CDP, CAS, De-Duplication, Replication, NAS etc.
@StorageTexan <– Follow Me on Twitter !!!