HP tries to make sense of the 'super-disrupted' storage market
Over the course of an hour Johnson spelled out HP's plan for storage, which could be summed up in one word: flash.
Q: The mass storage market appears to be very unsettled -- would that be a fair comment?
The storage market is super-disrupted currently, and that's driven by the ambitions of customers to drive down their storage bills. They see storage as a big part of their IT budget consuming a lot of their resources.
And the pressure has been from people wanting to get something cheaper than EMC's high-end storage and the like. The result of that is that a range of new storage models.
So currently we have software-defined storage (SDS), server-based storage (SBS), object storage and flash, and they have all come along with the aim of offering a lower cost model for consumers.
If you went back a couple of years, you would have seen HP, EMC, NetApp, Hitachi and so on. Some five or six vendors all dividing up the market share. Today we count maybe 20 vendors in the market.
So as these new storage models have come along, it has brought with it a tranche of new startups. Joining them are a bunch of new large players who you could split into categories like all-flash players, venture-capital-backed companies like Pure, SolidFire and Violin Memory. And then you have the software companies like Microsoft, VMware, Nutanix and Tintri coming into the software-defined space.
And you see some vendors like Hitachi, Fujitsu and Lenovo who have all been major vendors for a while, but all of whom are now doubling-down to focus on storage.
So the market is super-crowded and super-competitive, and so we have seen the price of storage drop -- and the revenues in the storage market are expected to drop too.
Storage market sales could fall by 12 percent
Q: How much are you expecting the market to fall by?
In the EMEA market we're expecting that the decline in revenues will be something like 12 percent, which bears out that there are new models driving cost reduction.
Now as a vendor that can be a difficult thing to manage, but we are also seeing better customer outcomes. From HP's point of view we have developed a simpler strategy.
We have doubled down on 3PAR as our primary storage architecture. We bought 3PAR five years ago, and since then we have been extending it upwards and downwards.
A few weeks ago we launched the 3PAR 20000 model. This can be an all-flash platform or it can be a hybrid, mixed disk and flash system. The idea is that we present a single system to the market with the same operating system that can be fast or slow depending on requirements, NAS or SAN, and so we think that for any kind of primary storage system 3PAR fits the bill. [With prices starting at $75,000, it is a very high-end flash storage server.]
To give you an idea of the scale that we get, you can have 5.5 petabytes in a single footprint and up to 15 petabytes in a single system instance, which means that the user could consolidate three of EMC's largest systems into a single 3PAR system.
What we are focused on is staying ahead in storage. So a few weeks ago, EMC refreshed their XtremIO flash arrays with a 1.6TB flash array and now we are launching 4TB.
This is driving the total cost of ownership of storage down. Using the 4TB array we can go to an EMC or other customer and say, we can reduce the cost of your legacy, high-end system by 85 percent.
And we have the one platform. EMC is looking to have flash variants of VMAX and VNX, so they have to re-engineer their platforms around flash drives. We are just focused on 3PAR and that is why we are focused on a single platform. We can innovate much more quickly.
Q: But one of the arguments for multiple architectures surely is that it gives you much more flexibility?
If you look at [EMC] VMAX -- traditional, high-end storage in this space -- it used to be the best SLA [Service Level Agreement], but it's not an all-flash platform so it's no-longer the fastest. It's certainly the most expensive and yes, it sort of defined the operational parameters for the data center.
If you look at XtremIO, yes they have a high-performance architecture. But all-flash architectures are high-performance, so it's no-longer the differentiator. That is why you will typically see EMC sell VMAX and XtremIO side by side. That means you have two platforms, two management training processes and so on.
So what we say with 3PAR is that it's the biggest all-flash platform and it's driving the biggest cost-of--ownership reductions. We have sold 10,000 3PAR units in EMEA before we even start discussing flash.
Customers want to focus on business outcomes, not architecture
We know that, going forward, customers won't want to mix and match because less and less do customers want to be systems integrators -- managing different technologies in a complex environment.
We expect to see the service providers take off [in storage] because more and more the customers want to focus on business outcomes rather than on IT integration.
The whole single-architecture approach is working very well for us.
Q: How about the flash competition with companies like Tegile?
There are something like 20 startup vendors out there, so I don't know the specifics about what they do individually.
I think the big issue is that the market is so crowded there may not be room for everybody. So can they be big on their own, or are they looking to be bought by someone else? We don't know, but I think this market is going to be defined very quickly.
Now we are growing at 660 percent, year-over-year and we will compete, but I think it's going to be a bun-fight to see who can take the market share. Given that the overall storage market is declining and flash is growing, this is clearly where the market is playing.
Q: What happens when you get into a fight on price?
We win. We win every time. We have just reduced the price of flash last week, and if you look at the 2GB format that is giving a 30 price reduction.
That is GB by GB, so we are looking at about 1.4 dollars per GB.
And it's not just about the price of the technology, it's also about how you use compaction and how you calculate the other data centre costs.
So we are looking at a scenario where if you replace legacy storage with an all-flash approach, you can get a return on investment within 12 months.
Of course the next question is, how does it affect the business? For a long time storage has been seen as a constraint on business -- a necessary evil. What people are going to see now, as the price of storage tumbles, is that they get the opportunity to do more big-data analytics.
To take an example, imagine a car manufacturer. There is a lot of data in a car, and a car can be connected. You can collect engine management system data and driving style information, radio stations, music that you have listened to, route information...
Now that can be hundreds of gigabytes of data, every day from every car. With that data the manufacturer can do all kinds of things such as fine-tuning better parts. When you combine that with the multi-trillion dollar insurance market, the car manufacturer at the point of sale will have the perfect opportunity to sell you can insurance policy.
And if on top of that they have all the data on how you drive they can then tailor the policy for you. So then they could offer you free insurance for two months while they collect data on how you drive. After that they could offer you a bespoke policy based on the results of that.
So at HP we can do a lot with all the business intelligence and the data.
SDS for cost-optimised storage
Looking at software-defined storage [SDS] side, our view is that there is going to be a place for hardware defined storage for three, four or five years and 3PAR is our hardware-defined architecture.
So we all know about security and that now it offers six or seven nines of availability and customers will continue to demand it and 3PAR is our answer.
Now there is another slice of the market that says I don't need that level of availability, but I need something that's more cost-optimised. This is where software-defined storage has a role, and we have something called StoreVirtual.
StoreVirtual was an acquisition we made about five years ago called LeftHand Systems [Actually LeftHand Networks was bought in 2008].
So we have five years plus of software-defined storage experience with about 150,000 customers using the system today. An important point about StoreVirtual is that it comes in about four different formats.
The first is already on a HP system when you buy it. It's already there and all you have to do is switch it on.
The second format is the StoreVirtual 4000 and we do configure a system using the same software but we will pre-package it with a chassis and some disks, wires and cables and you can buy it as a full system.
You can also buy the software in what we call a hyper-converged system -- think of Nutanix or TinTri -- with SSD disks, StoreVirtual software and so it is a ready-to-go system for small-to-medium-sized businesses.
Or you can buy it as part of our cloud offering, which we call Helion.
Q: How fast can you deliver one of these systems?
This is a channel play, so what we're talking about is distribution stocking and rapid delivery.
In the entry-level and up storage and server market we have traditionally been the number-one, but in competition with Dell, IBM and Lenovo. We aim to remain number one with a very price-driven strategy.
But we want to get out of that very price-driven approach to the entry-level market, and so we are going to go to software-defined storage with StoreVirtual very quickly to get out of that.
The price of storage is going down faster than the capacity is growing, but to my mind that is a temporary blip. That is why we are seeing the revenues in the market decline at the moment. But now we have technologies like flash coming along, and customers can see new opportunities and the chance to use storage to take advantage.
Backup and reliability
In this new world we are going to have to think very differently about storage, and the keys will be backup and reliability.
In the old world you bought a storage array and then you bought a second one to use as a backup for disaster recovery -- an expensive option.
What we think is going to happen is that you will get a much stronger classification of data going forward. We think that most companies will consider that about a quarter of their data is mission-critical and needs to be completely protected.
For the rest we will offer 3PAR and our backup and protection software StoreOnce. We have integrated StoreOnce to 3PAR directly, so you can take snapshot copies on 3PAR and move them to the low-cost platform quickly.
There is no host or application intervention at all. It's not as granular as a regular backup, but it gives you a copy if you ever need it.
Q: Do you see flash coming down in price dramatically?
Yes. So we moved from 900-gigabyte drives to two-terabyte drives in November. We have just moved again from 2TB to 4TB this month [June].
Also we invented the Memristor, which is non-volatile memory and the next generation of memory, and that gives us a fantastic opportunity to start talking about petabytes of scale.
One stat to think about is that if data centres were a country they would be the fifth largest consumer of power in the world, which shows the staggering growth in the need for power for the data centre.
You know this is necessary because some have computed that if the world continues to gather its data on spinning disk it will run out of power by 2020. Flash and whatever follows will be the way forward.
Further Reading:
Seagate opens $74M research facility in Singapore
Encrypting data at rest is vital, but it's just not happening
Storage company Tegile wants to make a splash with flash