What if you could sell customers cloud storage with SLA-guaranteed latency of less than two milliseconds plus comprehensive security and built-in disaster recovery? That’s the proposition of ClearSky Data, which is bundling primary storage, backups and disaster recovery into a unified, fully managed service in select metro areas.
The company, which exits stealth mode on Tuesday, says it’s latency that’s keeping CIOs back from fully embracing cloud storage. Anyone who’s ever used an application dependent on a database half a continent away can relate. Its business plan is straightforward: Deliver an end-to-end storage infrastructure that can match the performance, manageability and security of a local array while maintaining the cloud’s scaling and cost benefits. The back-end strategy is more Akamai than S3.
ClearSky’s founders, Laz Vekiarides and Ellen Rubin, know cloud, storage and data management. Rubin co-founded CloudSwitch, which was acquired by Verizon in 2011, and was vice president of marketing at Netezza. Vekiarides was executive director of software engineering for Dell’s EqualLogic Storage Engineering group.
ClearSky is based in Boston and backed by VCs General Catalyst and Highland Capital Partners. The service is launching initially in three metro areas: Boston, Las Vegas and Philadelphia, but in a briefing, Rubin told Channel Partners that ClearSky expects to expand its footprint rapidly as demand dictates. The company can move quickly, said Rubin, because the back end is commodity gear.
“It’s one cabinet,” she said, so as interest develops it will be a simple matter to add PoPs, and not just within North America. She cited colocation and data-center host Digital Realty’s European presence as a selling point for ClearSky’s partnership with the provider.
Encryption is baked into the provided edge appliance, data-center infrastructure and transport layer; keys remain at the customer site.
“Customers never lose control of data,” said Rubin. “It’s a founding ideal.”
One early customer is Xtium, a provider of managed cloud hosting, online backup and virtual disaster recovery services. Tim Vogel, CTO and one of the company’s founders, says Xtium has been a traditional cloud MSP since 2007, with traditional storage — up to now.
“ClearSky means capacity is now completely on demand,” said Vogel. “That removes the need for big hardware purchases while giving us more control over the location where data sits.”
Xtium was an early adopter of ClearSky’s technology and says it’s paid off.
“The economics are just stunning,” he said, clicking off a list of benefits: very low maintenance thanks to the appliance model, performance equivalent to on-site storage, availability of …
… unlimited snapshots, cutting-edge deduplication and compression and data mobility.
“We can save customers 60 to 70 percent compared with physical arrays,” Vogel said.
Xtium consumes the ClearSky cloud for its own business and also aggregates the service to resell to customers that don’t have the volume of data to work with ClearSky directly. Rubin says the company is not targeting SMBs; the sweet spot is enterprises with at least 2 TB of data under management. This is an opportunity for CSPs to deliver a service that would otherwise be out of customers’ reach.
While Vogel says plenty of vendors provide appliances with links to cloud back-ends, such as S3 or Glacier, these offerings suffer from performance problems and tend to be best suited to archival or disaster recovery. In contrast, he says ClearSky is a unified storage service that is akin to Akamai’s local caching model.
“That really lets you deliver a full production workload,” he said. “We’ve moved SQL always-on clusters, and we’ve had very good success. It’s performed just like production disk.”
While the service is not about to replace specialized, on-premises hardware for extremely low-latency applications – where access is measured in microseconds, not milliseconds – it can likely handle many production scenarios.
And ClearSky is not just gunning for the likes of EMC, NetApp and AWS S3. DRaaS vendors are also on notice.
“Because the local gateway can be virtualized, we can do multi-cloud implementations for customers,” said Vogel. “Now, I can do a traditional spin-up of a new app stack in our, or AWS’, cloud and in a couple of minutes dismount and remount and have access to the full data store on ClearSky storage.”
At that point, he can instantly replicate and move it. “Now I can put data in multiple regions, and have multiple snapshots,” he said. “It’s very elegant.”
As ClearSky adds more metro areas, the ability to geographically disperse data will increase.
“Given multi-PoP redundancy, we see this displacing a good …
… bit of our production storage,” he added. “It really is a game changer for us.”
Amazon, Google, Microsoft and other large cloud providers also want those storage dollars. Their approach is to build mega data centers in regions where they can scale at a reasonable cost and get access to large power grids.
Apple is spending $2 billion on a new data center in Mesa, Arizona, while Google recently paid $400 million to expand its footprint in Council Bluffs, Iowa. Thus, data is often stored hundreds or thousands of miles away from end users. An exception: S3 is well-suited for applications that are also running in the AWS cloud. Amazon will make sure the app and data sit …
… in close proximity. And, these services hold their own for customers with massive volumes of archival data and for pure disaster-recovery services, where low cost trumps performance. But business applications that leverage primary data and require less than 10 milliseconds to work properly are a different story.
Cloud providers have various schemes to address latency, including cloud exchanges or network hubs in major colocation sites. And, Avere and Google have partnered on hybrid cloud NAS. However, Rubin says these are capital-intensive and don’t free customer IT teams from managing a complex storage and DR infrastructure. In the Avere service, IT or a partner would still need to provision disaster recovery and manage the migration of data onto archival storage.
In contrast, ClearSky Data’s Smart Tiered Caching technology automatically distributes data across a network of geographically distributed cache layers with built-in redundancy. Frequently accessed “hot” data is cached at the edge next to customer applications, on all-flash arrays that can deliver hundreds of thousands of input/output operations per second. As data cools, it’s cached in a ClearSky PoP within 120 miles of the customer. For disaster recovery, customer data is protected with multiple copies in the ClearSky Backing Cloud. The company provisions WAN access between customer sites and its PoPs via redundant private networks and uses advanced deduplication and optimization techniques to minimize network traffic. The flash-based, on-premises caching appliance is included in the service. “If a drive fails, we replace it,” said Rubin.
ClearSky is also challenged by some well-funded software-defined storage startups, including Atlantis Computing, Hedvig and Primary Data, whose value proposition is enabling companies to build hyperconverged, highly manageable storage infrastructures on top of their existing gear and, often, the cloud services they have in use. For many enterprises, that ability to leverage sometimes massive sunk storage investments will be a selling point.
In contrast, ClearSky uses a capacity pricing model, charging per gigabyte, per month for the total amount of storage under management. The company said in a statement that the service “is one-third of the cost of customers’ previous storage environments.” While Rubin would not get more specific on pricing, in technology, it’s generally a matter of “simplicity, performance or low cost. Pick any two.”
While ClearSky will deal direct, Rubin told Channel Partners that it’s interested in partnering with providers in the metro areas it’s targeting for the service.
“We see a partner-enabled model with deal registration,” said Rubin. “Partners are important in building relationships.”
Follow editor in chief @LornaGarey on Twitter.
.@qosnetworks recently expanded its team. dlvr.it/RJJ8Zb
November 14 2019 @ 20:57:31 UTC