Mathias Rosenthal - Fotolia
Cloudian CEO Michael Tso said he has noticed favorable trends in object storage that could signal it is finally starting to fulfill its promise as a hot storage technology.
By submitting your personal information, you agree that TechTarget and its partners may contact you regarding relevant content, products and special offers.
Tso said more "general-purpose" companies are interested in Cloudian object storage. He said starting capacity points are rising, with deployments of more than 10 PB becoming common. He estimated 50% to 60% of Cloudian's 150 customers are enterprise users, and 40% to 50% are service providers.
Tso said Cloudian HyperStore object storage customers are also looking to make better use of their data. To that end, Cloudian recently formed a partnership with Skymind to collaborate on artificial intelligence and machine-learning systems to address use cases such as fraud detection and suspicious network traffic identification.
In this interview with TechTarget, Tso discussed additional Cloudian object storage trends and industry cloud storage developments, including the increasing interest in multicloud strategies.
Is there still confusion in the market about object storage and the best use cases for it?
Michael Tso: I think the confusion is still somewhat there. But from what we're seeing in terms of the growth in both expansion and big starting points, we are now getting more and more people coming to us asking for [Amazon] S3 [API]-compatible object storag, whereas several years ago, we were teaching people what it means to have S3 [Simple Storage Service]. So, it's quite different.
We felt when we started the company that object as a space is not even worth going after unless there is going to be a standard. And we bet the company that Amazon S3 APIs were going to be that standard. We don't have any translations, any gateways. We only support S3 as our API.
What's happened in the past 12 months is that every vendor in our space has come out and said they're also going to be supporting the S3 API. That's actually great news for the industry, because a market is only created if we all can agree on the standard. S3 is really becoming the new kind of NFS for scale-out storage.
Now that we all agree on this, it makes the choice from the end user a lot easier. They're just saying, 'I want to have S3-based object storage either on-prem or in the cloud. And I know I can use hundreds of thousands of different applications that are written to that API that I know is going to work.' So, they're protected in terms of their future investments. That is the biggest single change that's enabling people to make bigger bets, rather than just dabbing their toe in the water.
Do most of your enterprise customers with cloud storage use Amazon, or do they also store data with other service providers?
Tso: The No. 1 thing that CIOs are asking for is to be on more than one cloud -- being able to move from one cloud to another, or using different clouds for different workloads. A lot of people are realizing that it really pays to have a choice in vendors.
So, while Amazon clearly has the vast majority of the market share, and the vast majority of our customers are on Amazon, you are also seeing more and more of them becoming not exclusively Amazon. They either are on Microsoft and on Google, or they are looking to get on. There are a lot of different reasons for it. If you spread your workloads between the clouds, then you are able to negotiate your contracts better.
One thing that opens up the story for data management companies like us is we have the ability to provide the user with a single namespace on-prem, and based on policy, we are able to move data from their on-prem environment into different clouds and between different clouds. So, you can start with Amazon, but over time, if it's become more expensive, you can move the data from Amazon, for example, into Microsoft.
If you have a single namespace, [it] looks like everything is on-prem. When you read the data, we know where it's actually being stored, whether it's on-prem, or in Amazon or Microsoft or Google. Then, we are able to move the data back and let the users read it.
Aren't the egress charges expensive for users to move data out of Amazon to another cloud?
Tso: You are exactly right. We often have these conversations with users who want to get out of one cloud and onto another. And they do the math and go, 'Oh, gosh. It's really expensive.' But just because it's expensive, it doesn't mean that people don't want to do it. If you are with a single vendor, [that] is a really difficult position to be in.
We've been on the sidelines hearing how some of these conversations go. When a large company says, 'I'm now planning to move my workload from your cloud into another cloud unless you give me a better deal,' that conversation is a very powerful one. We've had customers tell us, if they don't have that second option, the only conversation they can have with their cloud provider is, 'Well, if you commit to a certain [greater] amount of workload in the next 24 months, then I can give you 10% off.' So, [they're] not really meaningful discounts.
From my point of view, we are completely neutral. We are just providing this enabling capability to allow people to work with more than one cloud. And that by itself is extremely valuable.
Do you think the multicloud strategy trend will eventually fade away, given the expense associated with it?
Tso: I don't think that trend's going to go away. Cloud's becoming not just a nice-to-have project; it's now becoming a core strategy for a lot of companies in IT. No CIO is going to ever want to have single-source anything. At some point, when it becomes vital, you absolutely have to have more than one vendor. But I do agree that in only extreme cases would people look at moving data from one cloud into another. It's a lot of work, and it's a lot of cost.
How much data are enterprises keeping on premises versus the public cloud?
Tso: If you look at just a snapshot of where they are right now, enterprises probably have 80% to 90% on-prem and 10% or 20% in the cloud. But if you look at new data, we are seeing that data that's created in the cloud tends to stay in the cloud, and data that's created on-prem tends to stay on-prem, because moving data between the cloud and on-prem is still very difficult and slow. So, to the extent that an enterprise has moved its computing infrastructure into the cloud, then the data created by that infrastructure tends to stay in the cloud.
What are the biggest trends you've been seeing this year?
Tso: More and more companies that may not have the most sophisticated IT teams or are not bleeding-edge companies are starting to look at object storage. And the starting point that everybody's looking for, on average, is getting bigger.
A couple years ago, when a customer said, 'I'd like to have a petabyte or 2 PB of storage,' [that was] what we'd call 'the big user.' Now, petabyte levels are pretty normal. More than 10 PB starting points are becoming quite common. The amount and the appetite for data that people have is increasing rapidly. And people are looking for more ways to get value out of the data that they are keeping.
Are existing Cloudian object storage customers increasing the amount of data they store, or are new customers deploying multiple petabytes right off the bat?
Tso: It's both. What we have seen in the past is that people tend to want to start small and experiment, and then they go big. What is new is that we see a steady drumbeat of new customers starting with big deployments.
And something interesting that happened in this past quarter was it was the first quarter in our history where we actually got more from existing customer extensions than we got from new [customers]. Now, that's not to say that the new customers were growing slowly. We set records in terms of new customers. But the expansion from the existing base was also extremely robust.
There are customers that are expanding from, say, 700 TB to 2 PB to 4 PB in one go, and we are aware of a couple that are expanding from possibly 8 PB to higher than 10 [PB] in one chunk. So, it's just a very healthy trend that we are seeing throughout our base of customers.
What types of customers are using object storage, and what types of data are they storing?
Tso: We've got customers in media and entertainment, so a lot of film and media assets. We also have customers in the medical area and genomic research. Another area is retail, so images and digital assets for the online commerce, and a lot of IoT [internet of things].
What are the greatest areas of growth for Cloudian object storage?
Tso: Overall, both the enterprise and service-provider segments are growing strongly. Our enterprise segment is definitely growing faster. There are a lot of enterprises that are moving on from their traditional NAS-based storage toward scale-out object-based storage.
Frankly, the growth in the service-provider side has been kind of surprising to us, because we thought several years ago that it might be harder for the smaller service providers to continue to flourish. But our thinking on that was actually wrong.
We've seen continued strong growth in the midsize to small regional service-provider segment. I think the reason is that a lot of companies without cutting-edge IT teams have had long-term relationships with these service providers that help them with everything, from hosting to providing the latest technologies. They are being guided into the cloud age by these service providers. And it makes a lot of sense for them because they may not have the scale to have a team of experts who understand how they can best use the cloud.
Top 2017 enterprise data storage trends include object storage
Use cases for object storage technology