Contrary to conventional wisdom, the cloud-storage-as-a-service industry is doing quite well, thank you. There are more and more great stories to tell.
As just one example, EMC just issued a press release that PEER 1 Hosting has launched a rich, robust cloud storage service based on EMC's Atmos: CloudOne
When I saw the story, I was intrigued.
I was lucky enough to speak with John Hamner, the product manager for the new service. I wanted to know more about what customer needs he was targeting with the service, and -- of course -- what he liked about the Atmos approach :)
A Bit About PEER 1
PEER 1 has been in business since 1999, and grown to to be one of the top five hosting providers in the world: their core backbone (the FastFiber Network) spans 17 data centers and 15,000 miles of fiber spanning the US, Canada and the UK. This year, it looks like they'll easily break the $100m revenue mark.
As a hosting provider, they've long offered more traditional SAN and NAS storage as part of their hosting services for some time. Not too long ago, their customers started asking for a carrier-class cloud storage service: elastic, predictable, secure, transparent, etc.
Meeting that need fell to John Hamner and his co-workers.
Where Is The Demand Coming From?
Although they had a few specific customer use cases to consider, John and his team weren't 100% sure who'd end up eventually using their service. Although they're mostly targeting sysadmins (vs. developers or consumers), they didn't want to be too restrictive on specific use cases.
As a result, they've built their service to be consumed and managed in a variety of ways -- opening up a dizzying set of use cases and associated target markets -- content storage, backups, archives, streaming -- the list goes on and on.
Sure, there are plenty of 10-20 GB smaller customers. But they've also found plenty of opportunity at the multi-terabyte level.
One intriguing example is working with medical imaging companies. More and more, they'd prefer to use an enterprise-class service with predictable per-use pricing vs. standing up their own storage farms.
Another example are people who are essentially building a service on a service. XYZMP3.com is in the business of offering digital media streaming services to artists, producers and media companies. They can offer a great service, and not have to worry about the storage or delivery infrastructure.
Positioning The Offering
John and the PEER 1 team did their homework at the outset; surveying what was out there, and finding some clear gaps in the storage services that were already out there.
First, PEER 1 is all about providing high performance, never-fail, no-excuse carrier-class services. John felt that many of the existing storage services out there didn't meet that exacting standard.
Second, PEER 1 wanted a service with *no* hidden costs. If you check out their rate card, bandwidth is free if you're on their network, unlike other services. You choose your desired protection level, and that's about it. It's about as transparent as it gets.
Third, they wanted a service that was the ultimate in simplicity to evaluate, purchase, monitor and manage. He saw that many of the other services were way too complex to evaluate and consume -- especially when it came to pages of legalese and pricing schedules.
If you take a look, you'll see the service is about as simple as it gets. My personal favorite was the AtmosFox browser plug-in, a short video of which appears here. Cool.
The Search Begins
According to John, PEER 1 evaluted no fewer than *eight* different cloud storage approaches from different technology vendors over a period of time. I didn't know there were eight different cloud storage approaches available to evaluate :) Paper evaluations were followed by lab evaluations and then a few trials at decent scale.
When I asked John what made the EMC Atmos approach stand out from the others, he unleashed a torrent of enthusiastic comments. I've done my best to summarize them here.
The functionality was rich, advanced -- and mature.
Atmos is not a version 1.0 product -- it's been out there for a while, and has the benefit of not only field production experience, but multiple rounds of enhancements and extensions.
Setting up a variety of storage policies for tenants -- check. Integrations via either API or filesystem access -- check. Powerful management tools for both PEER 1 and their customers -- check. Plenty of experience in working with service providers to stand up a multitenant secure storage-based service -- check.
That sort of maturity only comes with time and experience.
In particular, PEER 1 liked the GeoProtect feature -- basically, geographically distributed parity RAID. That gave them the ability to offer better protection than most customers could do themselves -- and at a lower cost than simply making complete copies of objects.
The product worked as advertised.
Everything we said the product did, it did -- and then some. That wasn't always the case with the other approaches they looked at.
EMC Atmos provided a complete and integrated hardware and software solution for service providers.
John was amazed at how quickly PEER 1 was able to stand up the service and integrate it with their existing operational processes: engineering, sales, support, etc. He states that it was about two months from the "go" decision to accepting customers -- a feat he equates with "light speed" in the product management realm.
Time-to-market matters in this business.
On the support side, there's one (and only one) vendor to deal with -- EMC. Most of the other approaches were assemblies of different technologies from different vendors -- a big consideration for the PEER 1 team.
EMC was obviously committed.
John just couldn't say enough about the positive experience he's had with EMC on this project. I'd share it all, but then I'd blush, and you probably wouldn't believe me anyway ...
From the early phases to deployment to joint development of new tools to help differentiate PEER 1's offering -- he and his team saw the ongoing investment and commitment to PEER 1's success with their service over time. More importantly, he felt confident that EMC was committed to continue to invest in Atmos over the long haul.
Positioning EMC As Part Of The Offering
If you take a look at the CloudOne materials, you'll notice that EMC's name and product capabilities are prominently featured throughout. I asked John about the rationale for doing so.
His answer was simple: people know EMC, and trusted EMC with their data. The fact that CloudOne is built and delivered in partnership with EMC helps greatly in positioning the service with demanding customers.
Compare that choice with ummm, using less-well-known vendors -- and the decision makes sense.
I couldn't have said it better myself :)
So, How Are Customers Reacting?
Like any hosting service, there's a bit of customer hesitancy when you're handing over the family jewels to a service provider, but -- once through that -- customers really like what they see, and are coming back for even more. The initial 200 TB in two locations is quickly filling up, making a case for additional Atmos nodes outside the US.
Plans For The Future?
Inital reaction has been so positive that there's now a very long list of things PEER 1 wants to go do around the core service they've created. One growingly important area is moving from file-oriented access to tighter API integration, and they're leveraging the thriving EMC Atmos development community to make that easier for their customers and partners.
John also thinks the whole mobile device/tablet thing (iApps?) presents a stunning opportunity to offer cloud storage services to developers and app service providers who want to tap into this explosive market.
When I pressed him a bit further, he was predictably cryptic: "the best is yet to come".
Final Thoughts
Back to my original premise, the market for cloud-based storage services is alive and thriving. The market has quickly evolved from a strictly cost-based focus to a service-level-delivered focus. PEER 1 certainly is emblematic of that shift.
Customers' needs are evolving quickly, and there's plenty of room to run for service providers who want to offer these kinds of services to their cusotmers. With Atmos, we're not really building a product, we're helping our partners build a business.
I was mightily impressed by what PEER 1 had done with EMC's technology. I think we'll see much more of this in the near future.
Thanks, guys, we're proud to be working with you!
Thanks for telling the PEER 1 story, Chuck. I echo your comments about helping customers build a business. While there is certainly great value in storage that "just works", there is even greater value in storage that enables the customer to release new products, with enough intelligence and programmability to match business models like freemium / premium, active / archive, etc. We hope to have many more stories in the near future!
Posted by: Leo | July 19, 2011 at 08:49 AM