You might have noticed that some storage industry talents are migrating from product to OEM vendors (component manufacturers). The latest one I’m aware of is Rob Peglar, Formerly at EMC Isilon, now at Micron.
If you look around, major storage OEMs, in both Flash and Disk markets, are playing very differently than in the past. Have you noticed?
They want to be more visible and to have direct contact with you. This leads to competition with their traditional customers, the array vendors. Here’s why.
The market is (quickly) changing
“Software is eating the world” someone once said, and I totally agree. Software-defined is quickly becoming mainstream and most vendors are quickly refocusing their product portfolio (sometimes just rebranding!?).
It has been happening to the entire datacenter: computing first of all, but now storage and networking are going towards the same direction.
End users are looking at new architecture designs where the value is in the software installed on top of white boxes. In fact, almost every next generation architecture out there is based on commodity hardware: scale-out file systems, object storage, Hadoop and Openstack clusters are just a few examples.
Furthermore, emerging startups are building their products on commodity hardware as well, sometimes giving you only the software and an HCL with a series of certified best practices/configurations.
It’s clearly visible in storage. Capacity (and performance) needs are growing exponentially, but some vendors (especially traditional ones) are not showing a similar increase in sales and margins… in fact, their revenues are flattening out. What is happening?
DIY storage!
Enterprises have accepted that in order to drive down infrastructure costs they have to rely on alternative architectures… and this is even more true when they target the Petabyte-scale. Traditional approaches are no longer efficient in terms of scalability, manageability and agility, and the success of hyper-converged solutions is the proof of that.
In the past we saw it in HPC and in all cost-sensitive organizations but now, because of the scale of the problem, it’s happening everywhere.
Many end users started building their own storage infrastructures a while back. At the beginning they were only PoCs in the labs, but now these infrastructures are ready to go into production, if they haven’t done so already!
Open source software (like Ceph, for example) has played a major role in this, and enterprises have finally caught on: for many applications, commodity hardware is the right choice (if the software is well written). Sometimes it’s only about building the cheapest storage system for less important data, in other cases it’s more than that.
One of the consequences is that now end users are starting to look around for components instead of finished products. And this is a huge opportunity for the OEMs. In the past they were hidden by storage vendors (typically you wouldn’t ask which type of hard disk your vendor was using in a particular array, would you?)… but now you want to know if a disk is better than another, how many IOPS can you get from a particular PCI card, failure rates and so on. Users are digging deeper to find out what is best for them, they are becoming the new manufacturers.
The ex-OEMs
I think OEM vendors recognized this trend some time ago, and now they are aggressively working to build reseller channels. For example, I had a direct experience with HGST which, in Italy, operates through a VAD and lately they have been signing contracts with a fair amount of resellers (as far as I know it’s the same for other EU countries, and other vendors). And they are also working on pre-sales to help these resellers build knowledge and solutions.
They have technology and components that are at the base of traditional products… and now they’ve also put other companies, like Datacore or Nexenta, in touch with their resellers. Not to mention their investments (and acquisitions) in startups that can help to drive up sales as well as broaden their product line-up.
In some cases, if you take a look at what they can offer, you’d be surprised.
Closing the circle
I’m not saying this model is good for everyone, in most cases, and for the average user, traditional approaches are still fine. But more and more organizations are building bigger storage infrastructures (e.g. data lakes) and they must shrink costs from $/GB down to cents/GB to make them sustainable. And I’m talking about both TCA and TCO!
For Disk and Flash manufacturers there is a huge opportunity here, but they have to prove that not all the components are born equal… or they’ll get screwed again on the price.
They’ll try to be even more visible and relevant in the future, but they’ll never admit to competing with their customers.
End users will greatly benefit from this situation; “unofficial” competition between traditional storage vendors and OEMs will allow users to have many more options and freedom of choice.
