M+E Daily

Cloud data storage is a key component of any multi-cloud strategy but major cloud platforms today tend to offer solutions that are “too intertwined with their ecosystem, limiting options while increasing costs,” according to Wasabi.

After passing the first wave of cloud adoption, many organizations now want to establish multiple options and select the best-of-breed platform for each individual project – all while improving compatibility, interoperability and cost. That means leveraging a data storage infrastructure worldwide that’s accessible, compatible, consistent, predictable and cost-conscious over time, Wasabi says.

Wasabi executives and Enrico Signoretti, an analyst at GigaOm Research, discussed theses issues and focused on how to evaluate the data foundation layer to build a successful second generation on Oct. 14, during the webinar “The Role of Cloud Data Storage in Second Generation Multi-Cloud Strategy.”

“Backup and archiving is probably the most popular” use case for Wasabi’s 23,000 or so customers, according to David Friend, its CEO and co-founder.

“But that doesn’t necessarily mean that’s the biggest consumer of storage,” he said. “Probably the biggest consumers of storage are things relating to video and things like Hollywood movie studios that are storing feature-length films with us. One feature-length film might use 10 or 15 petabytes of storage, whereas backups and archiving typically 100 terabytes and down would be more typical of the size. But in terms of just the number of customers, Wasabi seems to be becoming kind of the default place to store off-site backups and archives.”

The company is “seeing a lot more cloud tiering of primary data to Wasabi these days,” according to David Boland, director of product marketing at the company. Companies like NetApp and its ONTAP operating system “can offload a cloud tier to Wasabi and Isilon or PowerStore from EMC Dell connects to Wasabi as well,” he noted. Additionally, “we’ve got a good relationship with the folks over at Tintri and their Cloud Connector helps offload their IntelliFlash and VMstore to Wasabi,” he said. “So definitely primary storage offload is a growing solution, especially because those on-prem storage providers have the foresight to add the S3 connector to allow them to send snapshots and cold data to the cloud. That’s definitely a good use case.”

Agreeing, Signoretti said just about “all the primary storage vendors have solutions to realize hybrid cloud infrastructure, so this is a growing use case.”

The Challenges

But “there is a problem with multi-cloud,” according to Signoretti, explaining:

“All of us started with data on premises many years ago and then the cloud came in and we started [sending] some of the data, some of the applications, some of our workloads to the cloud. And at the very beginning it was not even a strategy. Multi-cloud was a tactical thing [and] we were forced to choose a secondary cloud for other reasons.”

That “creates a problem because this proliferation of clouds, of course, create a series of data silos,” Signoretti pointed out.

“What do you do? You can make copies of data or you can access data from your remote application…. But there is a problem. Sometimes there is latency. Not always. But if the clouds are not well-connected, this can create latency or there are some costs that you didn’t think of at the beginning like egress fees. So every time your data exits from one cloud and [goes] to another, you’re paying for that transaction. And sometimes it’s very expensive,” Signoretti said.

There are “very good reasons why people have sort of a hybrid approach of some combination of on-prem and cloud-based data storage” today, according to Friend. “If I go over to my Bank of America branch and I stick in my ATM card and do some kind of a transaction, the storage behind that is probably some very hot flash storage that’s sitting behind an Oracle database or some kind of custom application,” he noted.

But he continued: “That kind of storage is extremely expensive and when I come home and go to the Bank of America application and I want to pull up my bank statement from three years ago, I’m sure that that information is not still sitting in some kind of very expensive flash storage that’s been pushed off to something more like Wasabi, where you’ve got spinning disk and you’re talking about maybe 10 to 50 milliseconds of access time, which is still, for a person who’s pressing a button and wants to see the report, it’s for all practical purposes, instantaneous. But for a transaction processing system that might be too slow.”

There are “certainly a lot of applications where the storage and the compute associated with that storage… needs to be very close together because we’re processing millions of transactions a second and there’s just no room for latency in there,” he said.

“However, that’s not where most of the world’s data lives,” he pointed out, explaining: “If I go over to Mass General [Hospital] and want to look at my X-rays, they’ve got 20 years of my X-rays and MRIs and so forth, available at the click of a mouse on the doctor’s desktop…. It’s not dead cold storage like something like [Amazon S3] Glacier or a tape in a box sitting in a warehouse somewhere and it’s not the super-hot storage that would go behind transaction processing. It’s what I would call the everything else storage. And to me that storage has all the characteristics of a utility. It’s like electricity or bandwidth and it should just be there available to store whatever you’ve got – whether it’s a picture of a night sky from a telescope in Chile or [an] MRI or my bank records from three years ago.”

He added: “It can really reside anywhere within reasonable geographic distance from the compute, whether that compute is on-prem or it’s in one cloud or another cloud…. And, in fact, out of our 23,000 customers, 23,000 of them are doing their compute somewhere else because we don’t offer compute, only storage. And so I think that comprises the bulk – and I don’t know whether that’s 90 percent or 95 percent. It’s the vast majority of all the data that exists online in the world.”

The “Dark Side” of Multi-Cloud

The “dark side” of a multi-cloud solution “really is the cost associated with moving your data back and forth between one cloud and another,” according to Boland.

But he made a prediction: “We will get to the point in the next decade or so where egress fees must go away because in order to have a best-of-breed solution, a best-of-breed compute, a best-of-breed storage, best-of-breed analytics that are sitting in different clouds and moving the data back and forth, customers can’t afford to pay them 9 cents a gigabyte over an Internet connection or 2 cents a gigybyte for a direct connect. The vision of a data fabric that we hear from different vendors really does require that egress doesn’t exist.”

Wasabi has “a few customers now that have a multi-cloud architecture,” he said. “Luckily for them, they’ve nailed up big pipes between all the clouds and the cost associated of moving data from storage to compute to analytics and back to storage isn’t that bad for those guys at 2 cents a gigabyte for the most part.”

But h predicted that “you’ll see agreements being made between folks like Wasabi and… next-generation 2.0 companies that eliminates the egress and makes a true multi-cloud solution much more attractive to customers.”