NetApp’s Cloud Strategy

We presented NetApp’s Cloud Strategy at our annual Industry Analyst Meeting a few weeks ago. To summarize, our goal is to be the storage provider of choice to companies building cloud-compute environments.

We have a strong customer base in all different types of cloud computing from clouds for consumer apps like e-mail to clouds for business apps like ERP and CRM. Customers include Oracle on Demand, Yahoo! Mail, SAP by design, Photobucket, AT&T, IBM Hosting, and many, many more. Although nobody formally measures it yet, we believe that we are already the number one cloud storage vendor in the industry.

Analysts had surprisingly varied reactions. On the positive side, one said, “NetApp nailed its cloud positioning. It gets the requirements & relates them well to its products,” and another said, "During the two-day event NetApp provided a substantial amount of evidence to support its enterprise cloud storage story and is one of the better stories I have heard in this vein.” With the opposite point of view, a third said, "I would probably go as far as to say they do not have a cloud strategy.” Wow. At least we can’t accuse them of copying each other.

Part of the disagreement comes because NetApp’s strategy is not to build clouds ourselves, but to help other people build them. Some people view this as an excellent strategy, while others think it doesn’t even count as cloud a computing strategy at all. In response, I would ask this simple question: “Does Microsoft have a PC strategy?” It’s true that Microsoft doesn’t actually build any PCs, but it seems obvious to me that they are a key driver in the PC industry and absolutely have a PC strategy.

Still, it’s a fair question to ask why we chose the strategy we have. The Microsoft analogy helps answer the question. Microsoft’s core competency is to develop software. That is a very, very different set of skills for a company than running efficient manufacturing lines to produce PCs, so Microsoft leaves that part of the overall solution to other vendors like Dell and HP. Microsoft designs an important piece of the PC – arguably the single most important – and they have strong opinions about how PCs should be used and how they should evolve, but they don’t actually build the PC themselves. Likewise, NetApp’s core competency is storage and data management. This is a very different skill set from running large data centers efficiently. I won’t go so far as to claim that storage and data management is the single most important element in cloud computing, but it certainly ranks up there. So even though we don’t provide cloud computing directly, we have strong opinions about how clouds should be built and used and how they should evolve. To me, that’s a cloud strategy.

(Here are some links for more thoughts on cloud computing and how to build efficient data centers from Dave Robbins, our CTO in IT.)

Comments

[Anand]

Thanks for the article, the microsoft example was really worth thinking!

Regards,
Anand

[Dan Orchard]

I agree that its good to have a cloud strategy as its what customers are talking about but isnt cloud computing just another buzz word to help IT manufacturers sell more?
Ive always enjoyed reading NetApp articles due to them being more hip, and more of the today. Other vendors have always seemed like old men with pipes talking about strategies as if it was World War II.
But my only problem with the new Cloud strategy is that although as you say Microsoft have a PC strategy, they are also heavily involved in how the PC architecture is built and how it compliments their software, how will NetApp do similar to make sure they can be influential to making sure the jigsaw fits together as it should, providing the customer with a perfect solution?

Regards

Dan Orchard

[Kevin Stay]

How about coming up with an actual flash strategy? (Hey, it's great you post on this whole "cloud" claptrap especially since I guess from your position you feel kind of obligated to say something about it.) Meanwhile in the day to day business of virtualizing whole environments your inability to competitively meet those shared storage requirements is making life for NetApp "evangelists" like myself very difficult.

I appreciate your hesitancy to "eat your own children" in the form of your 15K so-called enterprise aggregates, but the worn out arguments for it in TODAY'S reality are worse than threadbare:
1. Your critical business data really has no business on SATA drives. Sorry, but with the field service number backed reliability of the NetApp platform and the quality of the SATA drives being shipped you really just need to stop saying this. Ever.
2. To get any kind of performance at all SATA is just not an option. This is where 15K aggregates are really between a rock and a hard place. Anyone who actually measures sustained throughput and IOPS needs ends up being surprised how far upstream you get with a properly done SATA layout. Meanwhile flash prices and capacities continue on their respective steep curves making it a more and more obvious choice for that smaller and smaller set of data in such a class of performance.

So, as a response we get PAM 2 and flash behind a V series. Time will tell, but from my position here in NDA land the next PAM looks like it is going to be outrageously overpriced and only do a fair job with my next wave - virtual apps and virtual desktops. That leaves me with putting a V series in front of a small amount of real flash storage. Now you seriously have my attention. In fact you would have yourself an out of the park home-run if it weren't for one little problem; even with significant discounts there is NO way in today's economy I could possibly pay for the switch from FAS to V even of the exact same model and keeping all my current NetApp shelves.

If NetApp is serious about meeting all of the emerging storage needs, especially for thin provision/de-dupe yet jumbo throughput/IO virtual apps and desktops, then you need to consider a couple of things. First, offer a 1 year straight tradeover from FAS to V (or simply unlock V functionality on FAS). Second, get really-Really-REALLY serious about your partnerships with folks like Texas Memory Systems and get off the launch pad quickly with Fusion-io and the likes.

I want nothing more than to use NetApp for all my storage, but if you cannot come up with something a lot better than what is currently/soon on offer I will simply have to look elsewhere to meet my "tier 0/new tier 1" needs.

[Val Bercovici]

Hi Kevin,

It was a pleasure chatting with you yesterday to better understand your requirements!

Later this year, we’re very much looking forward to augmenting our existing SSD solutions based on TMS RamSan today with a radical dedupe-aware cache strategy which will offer VMware ESX & VDI server farms <2ms response times for *all* I/O’s via dozens of TB’s of effective storage controller cache at a very competitive flash price point!

Stay tuned to Dave’s and my blog for more details soon :)

Val Bercovici
Cloud Czar, Office of the CTO
Vice-Chair, SNIA Solid State Storage Initiative

[David Vaughn]

Dave,

In your quote "Part of the disagreement comes because NetApp’s strategy is not to build clouds ourselves, but to help other people build them. Some people view this as an excellent strategy, while others think it doesn’t even count as cloud a computing strategy at all. In response, I would ask this simple question: “Does Microsoft have a PC strategy?” It’s true that Microsoft doesn’t actually build any PCs, but it seems obvious to me that they are a key driver in the PC industry and absolutely have a PC strategy."

I feel personally that this is an unfair comparison. Even now, Microsoft is scrambling to develop a "Cloud OS". As a former NetApp employee, and now working for IBM, I still hold a vested interest in the NetApp family product line.

I think NetApp should embrace Cloud environments as a PS, it would be an equitable market for NetApp to utilize individuals like Val not just internally, but externally as well. Look at your Data Center in RTP, and how you guys designed it from the ground up. That is an equitable service offering in the making.

I think Kevin had some interesting points with the SSD comments as well and I look forward to what is coming.

[Donald]

What standards are you using for Cloud computing? How can you guarantee security and data protection withthis approach?