Flash is reaching an inflection point: Q&A with Tegile
The storage capacity race continues: Tegile's latest offering is a half-a-terabyte flash drive that is packed into a standard 3U slot -- and there's a similarly packaged one petabyte drive to follow soon.
We last spoke to company CEO Rohit Kshetrapal last year but with a company moving as fast as Tegile, it's not too soon to have a catch-up.
Q: What is your take on the current storage market?
Tegil CEO Rohit Kshetrapal: We have been focused on the breadth of our platform. We separate our platform into whether it is hot data, cold data, and metadata, or whether it is indexing of the metadata. How does that make a difference to a platform? It starts with the media.
You have the hard drive -- and that is still very much part and parcel of it -- and then you have flash coming in and coming in multiple grades. So it is moving from eMLC to MLC to TLC (enterprise Multi-Level Cell to Multi-Level Cell to Triple Level Cell), with capacity going all the way from 2TB to 16TB and as density increases, flash is slowing down.
Kshetrapal: Last quarter we did 65 percent new business and 35 percent with existing customers.
And then you have got the fast portion of it as well -- NVME (Non-Volatile Memory Express) and so forth. Now these are all [coming in the next] six- to 12-month technologies we are talking about, so how are you going to adapt your architecture to deal with this?
That is what we are talking about when we talk about value. We started with multi-protocol from the very beginning. It is about the customer's workload and if they require extremely fast or low latency, then we go for all-flash systems.
On the other hand, we want to provide the customer with flexibility, so we can start with some flash and then you can move upwards to all flash or down with hard disk.
[With the alternative] model, the NetApp model, all their products are managed with the same operating environment. When you have others coming up the product lines, such as a Numble or similar, then you have a separate product line for all-flash, a separate product line for hybrid, separate product for fibre-channel, separate for iSCSI, and then it can be very difficult.
That platform approach allows us to access many layers of flash, moving up and down, moving from a hybrid with 10 percent flash, then up to 20 percent, 30 percent or all-flash or dense flash.
We are starting to see this in our sales growth. In the last quarter, we did 65 percent in new business and 35 percent in expansion. So for us, that is the part that is really moving for us.
Q: How do you compete against a company like NetApp which haa stacks of disk already installed with customers?
Kshetrapal: We compete where the customer has one of two problems: they either have a performance problem or the customer has a performance review coming up. That is generally the issue.
When they have bought an array [from NetApp] by year two or year three, the performance problems start coming up and that's when you start having a conversation with them. What they are looking for, generally, is the right performance and simplicity of management -- and we have a platform which provides both of those.
Q: Would it be fair to describe you as in consolidation mode currently?
Paul Silver, VP EMEA, Tegile: No. We are doing our 'land and expand' rather than being foolish and trying to change everything. That is not going to work with large corporations. We want to get in, choose a part of the business, and improve it.
A lot of flash: Half a petabyte in a single rack and Tegile plans to double that density.
Kshetrapal: We can give you many examples: hedge funds that have had performance issues, even high schools that have had performance issues with their arrays.
Q: Is the high-end your main stream of business?
I would say it is the mid-range and the high-end of that. What is really interesting with flash is that it is reaching its inflection point. So now we can really start to think about how we are going to start replacing the disk infrastructures.
And what our architecture is all about is that you can have a fast component and a slow component in the same array. So the question is: how do we take these large systems and migrate them into a single environment? We take this flash technology and often we back up onto disk and then we put it back into a carrier. But is that efficient? I think that we can break the back of that now.
Instead of taking a NAS environment and just putting it into disk, why not have an infrastructure area where you can put the cards directly in? We worked with our investors and our partners in components, including SanDisk, and came up with a platform.
In the system box, you are not putting hard drives, you are putting in cards. Today, each card is 8TB capacity so that means that in one rack you can put half a petabyte.
Now it has to be hot-swappable, so if a card fails you just swap it with another card. You can buy a rack and if you want 190 terabytes, you can put cards in for 200 terabytes. The idea is that in our new rack, there are enough slots for as much or as little as you want and you are free to populate it as you like.
Q: With all these slots, how do you handle the cooling?
Kshetrapal: It is all built-in. There is a fan in it to take care of the cooling for you and it is, literally, just like a standard expansion shelf.
Silver: When you think about it, you can have a lot of storage -- half a terabyte in a standard 3U system. A large storage environment -- with lots of NAS arrays, lots of slots -- can be replaced by one of our boxes: a single 3U system that doesn't use anywhere near the power and resources of conventional storage. It uses 35 watts of power in a single rack and 500 watts under full power which is a fraction of what you would expect from conventional storage.
Kshetrapal: We will soon have the next version of this array and it will double the capacity. That means a petabyte in one standard rack.