The growing adoption of cloud has brought with it the growing adoption of object storage, which Gary Ogasawara, CTO of Cloudian Inc., described as an evolution of block and file.
Founded in 2011 and headquartered in San Mateo, Calif., Cloudian is a privately held file and object storage company that claims to be the most widely deployed independent provider of object storage systems. Executives recently stated that Cloudian has come off its best year to date and now has around 700 customers.
On Feb. 28, the vendor introduced HyperCare, its new remote managed service of its object storage products. HyperCare provides "cloud-like" storage to customers so that they can consume products the way they do other cloud services, but in their own data center. As a managed service, HyperCare allows customers to focus on their business without managing the underlying storage.
In this Q&A, Michael Tso, co-founder and CEO of Cloudian, and Ogasawara talk about the reasons why object storage and the S3 API have become an attractive storage model, as well as the drawbacks to the technology.
How do you define the term object storage?
Gary Ogasawara: Object storage provides the API to get to data from software applications. ... It's also been the evolution of storage. Block came first, which uses things like iSCSI, and then file with POSIX. Object now allows a software developer to write one line of code to say, 'I want to get this object; I want to store this object; I want to list what's in this bucket' and so on.
Is object storage difficult to adopt?
Ogasawara: The S3 API has been the key to wide adoption. When you're a software developer, you want to make sure that when you write code, you don't have to change it for every vendor. Because the S3 API has gained this de facto status, it's been really good for the industry and really good for software developers building applications around it.
Michael Tso: The barrier to adoption is really that it's something new. It took a long time for the industry to move toward file. Now we need to get the industry to move away from file to get into these new APIs. And the first step is the industry has to agree on one set of APIs. It doesn't help if everybody's got their own flavor. You need to have one [set], and when you have the one [set], then you need to have the community build around that. That is where S3 is really key. Because you now have every application that anybody can think of -- be it warehousing, archive, AI -- all supporting object and supporting S3, specifically, and that means that platform now has a chance to succeed.
Is object a fundamentally different storage technology or did the cloud make it necessary?
Ogasawara: It really has been a paradigm shift, in the Kuhnian sense. There was a time about 10 years ago, about when Cloudian was started, where there was a lot of innovation in distributed systems and distributed systems theory -- how you could scale out a system and make it fault tolerant. That allowed all this NoSQL movement to gain speed, and it allowed us to build based on that distributed systems foundation. We've displaced these legacy vendors that are brittle because they can't scale out by adding nodes or are brittle because they can't tolerate a certain failure. Or you can't specify fine-grained policies on how you store data. Advances in distributed systems allowed these new classes of companies and products to appear.
What are the drawbacks to object storage?
Tso: The major drawback has been that it needs a different API. There are people out there that say, 'I can make NFS more scalable,' and profitable companies will come along and do this really well. But you can only solve the problem in a certain way and still feel confined by the box of what API you have been compatible to. The moment you break the compatibility and you get into S3, you can do all these things that you could not do before.
But it's new. ... To use the power of object, people have to rewrite their applications, and that is always something that takes time. It's basically a chicken or egg problem -- without a lot of people writing applications, you don't know what's going to be the de facto standard. But without a de facto standard, nobody's going to write any code. It took a while for S3 to emerge as a de facto standard. But now every vendor in storage -- even if they're doing file -- are saying, 'I'm going to be supporting S3.'
What is the attraction to object storage?
Tso: What we're seeing is a lot of people simply cannot store data on the traditional storage at the same rate, at the same cost and with the same way to manage. It just doesn't work. When you get to enough of a scale, object is really the only way.
Do you see the big vendors such as Dell, Hitachi and IBM as object storage competitors?
Tso: I think the way to speak of it is that, even within their own companies, their [own] object storage is competitive to their legacy business. On a per-unit basis, [object storage] is going to cost people less, so they have to sell a lot more of the new stuff to make up for the old stuff. It's a case of the Innovator's Dilemma where your legacy is so successful that you just can't possibly invest enough in the new stuff.
Editor's note: This interview has been edited for clarity and brevity.