When you think about the storage in your IT department, it's reasonable to compare it to an iceberg.
The peak of the iceberg is your primary storage for production apps, but below that is a vast bulk of secondary storage filled up with back-ups, archives, test and development, shared files, cloud, and more. These all form the vast bulk of the iceberg, lying hidden beneath the waterline.
Cohesity is a company that believes that vast bulk of secondary storage deserves as much attention as the primary. Why? ZDNet talked to Cohesity's CEO Mohit Aron to find out.
Do you see yourselves as an enterprise player? Gartner seems to think that the high-end of that market is still an area that eludes you.
It depends how you are going to define enterprise. The very high-end takes time, but we have customers there, which is very flattering because normally, with the high-end, it takes longer.
We are practically doubling our customers every quarter. The company is growing at a very fast pace and we are at some 300 people now. In February, we were less than 150. By the end of the fiscal year, which ends in July, I anticipate we may cross 500.
We are growing internationally too. We started off focusing on US sales, but now we are all across EMEA and we are looking to expand into APJ.
We have redefined the secondary storage market and now people align with the way we have aligned secondary storage. They recognise the problem that we recognised.
I draw the analogy to the smartphone. Before the smartphone, we would carry a phone, a GPS device, a camera, a light, and so on. When the smartphone came, it consolidated all of that onto one platform. We are kind of doing the same thing for secondary storage.
Before us there were any number of different devices that people could buy to do back-ups and analytics and the rest. Then we came up doing all of that on one platform.
Now, that vision is flying well and, because they have very complex datacentres, large enterprises really like us. They have products from practically everyone and that really hits their bottom line.
With our products, they can manage that complexity with something that is really simple and it's that simplicity that is really working with these datacentres.
We think we are really redefining the landscape of secondary storage.
Where did the idea for the product come from?
It came from a question: 'Why are back-ups an insurance policy?' That was the question that led me to form this company.
If you think about it, today when any enterprise company puts together a back-up solution, it costs. Worldwide, people spend billions of dollars buying back-ups, but they don't do anything with it.
It is just standing idle in the datacentre occupying shelf space. It's only used as an insurance policy and a very expensive insurance policy at that. It's only used when you have to retrieve data or when you lose data or for compliance. And it's complete hardware with memory, hard drives, CPUs and so on. Why don't we use it?
That question led me to think: 'what else can we do with these devices?' And then I thought, there is this thing called secondary storage that you can all consolidate over. You can leave your primary storage alone and everything else can be consolidated on this platform.
Now we have this logo: "Not just a back-up." The short-form of that is JAB and we all know that jabs are painful. We like to say, don't buy JABs.
Our device is not just a back-up. It can not only do back-up, it can also do file services, object storage, test and development, and more.
That's where the idea came from. Let's do more than your back-up can. That means that the requirement is to build up our back-up platform so that we can do more on it and we also can use it to fix back-up problems. Today, in the legacy world, you have to go and buy some back-up software from one vendor, a piece of hardware called a media server on which to run software, then you buy the storage from another vendor and then you put up a solution. Then that is your back-up solution and by the way, it's just an insurance policy.
Now, we come into the space with a platform, and on that one platform you can do everything that you need to do back-ups. That's how we start but you can use it to do more things.
Is that all it is?
No. The other part of the secret sauce is that it is one thing to talk about all that, but you don't just want all this data on one platform. You also need a scale-out the device so that its scales like Google. That's our secret sauce.
We built a device called SpanFS. The name comes because you can span multiple things. It spans storage tiers like SSDs and hard drives. It spans the public space and the private. It spans workloads like back-ups. That's the technology that brings all that onto one platform.
You say you have been picking up many customers, are there any areas that you are specialising in?
It's across the board. What tends to happen is that these customers act as references for others in that vertical. Financials is one, and some of our sales teams almost run a community by themselves. Then we have healthcare, technology, retail, government, etc. The product is very secular.
When you are showing your technology to a company do you demonstrate an application in use that might blow them away, or do you just ask them 'where do you need help?'
We try not to be too general, so we tend to look for a pain point they may have. Back-up tends to be a pretty good one to look at. But they usually have many different back-up solutions from different vendors, so we can offer a very simple solution. It also costs 50 to 80 percent less. We think it's very easy to upgrade and very easy to grow in a web-scale, Google-like fashion.
From that it's land-and-expand. Customers can actively web-scale their storage and as they fill it up, they can buy more from us and a lot of our business now comes from repeat customers.
Then secondly, the customer can look at adjacent, secondary storage workflows. In other words, that's a file server sitting out there that you are paying a lot for maintenance to some company. How about if you consolidate that, because we speak NFS, we speak SMB, we speak S3.
This thing doesn't only need to be a backup, you can do more with it.
Do you do this on their servers or do you do it on your severs as well?
The philosophy we have taken as a young company is that we don't want to class all sorts of hardware, because we don't want to be in the business of just solving hardware bugs.
We have a default white box and if a customer says that's fine for us, that's all we need. It's not clunky, it's fine.
But if a customer says 'we are only HP buyers, then, guess what? HP is our partner and an investor, so we have qualified a SKU for HP. And for Cisco so they can buy the particular SKU that we've qualified and then they can buy the software from us and the hardware from Cisco or HP. They get what they want, and we get what we want.
What new developments do you have coming up?
I am a big fan of saying that we don't want to be a jack of all trades. What we do want to be is the master of one trade at a time.
We started of being the master of data protection. Now we believe that we've done a great job in data protection, but it's time now to pull in other parts of the vision and so we've been focusing on file services.
Over the last quarter, we have had a fair amount of deals come in because we can do file services. People were using us for batches and file services. And you will see a lot more happening on file services. We made a couple of announcements in the cloud area and there will be ongoing announcements too.
We are friends with the cloud and we believe that we should offer the best of both worlds to our customers. We can support the storage box or the cloud or both. Cloud work is ongoing. We added the ability to run a virtualised cluster in Amazon and to do fail-over between them, do back-ups of VMs that might be running on Amazon and so forth. All of this is already announced or is coming up shortly.
Is one of the advantages of your approaches that users don't have to buy as much hardware?
In legacy systems, you buy hardware and when you fill it up you have to throw it away and get another system which is a bigger box.
With a scale-out approach -- a web scale approach -- you don't have to do that. You can continue using whatever it is that you're already using, and you just add more nodes and scale out. And you only need to get rid of old hardware when it's really end-of-life and so slow that it makes sense to replace it. In that way we help customers to maximise their existing assets.
What about the people? You are in a very specialised area requiring people with very specialised knowledge. How do you recruit them?
It's been a very rewarding experience to mentor some very bright minds, showing them what we have learnt over the years. The people who do well here are people that are knowledgeable about distributed systems design.
When you build a system that doesn't work at scale, you can assume that if something fails, within half an hour some admin will come running and fix the system. But when you are working with web-scale systems, which are very, very large, you can't make those assumptions. In fact, the assumption you can make is that something will always be down and that it will stay down for an extended period. And yet, the system must work, and work successfully.
These were the underlying tenets which I learnt when I worked building the Google File System which is a very scaled-up file system. That worked to a very large scale. A small system there could be anywhere from a couple of hundred nodes to 10,000 nodes and at that scale you cannot assume that if something goes down admin will come within half an hour to fix it. Things go down and stay down and maybe a couple of months later somebody will fix it.