Storage Switzerland just received confirmation from one of our more trusted sources that EMC is actively in discussions to acquire ScaleIO. This makes some sense; EMC has a tendency to buy early stage Israeli based companies (XtremeIO and Kashya) and they have a tendency to buy companies that we are getting ready to cover. As a result, I’m confident that something is in the works. The confusing part might be what does EMC want with ScaleIO? They already have a rich storage portfolio including the recently announced ViPR product that we covered a few weeks ago. What would ScaleIO bring to the EMC storage arsenal?
Who is ScaleIO?
To understand what EMC may do with ScaleIO first requires a basic understand of the ScaleIO solution. While a more detailed Storage Switzerland briefing note is forth coming, in short, it could be described as a software defined storage solution that abstracts the storage software services layer (volume management, RAID snapshots) from the physical storage system. In fact, ScaleIO takes this a step further due to their ability to aggregate storage within the compute infrastructure itself. As we discussed in our recent article, “What is a HyperScale Data Center?“,we describe these types of architectures as HyperScale Storage Infrastructures.
What is HyperScale Storage?
A HyperScale Storage software solution does more than just abstract the storage controller so that you can use disparate hardware, it also embeds a software defined controller architecture into the compute tier and aggregates the server’s direct attached storage into a shared pool of capacity. As a result, you can use any storage hardware as well as inexpensive direct attached storage, while still maintaining all the benefits of shared storage.
This is the key requirement, in our opinion, is this ability to aggregate storage in the compute tier and present it as a shared pool. There are of course differentiators within the HyperScale Storage market that these solutions use to compete against each other, like ability to provide QoS, ability to automatically support different types of tiers of storage (SSD and HDD) and perhaps most importantly, the software platform the hypervisor can run on. Other companies in this market include include HP (Lefthand VSA), VMware (VSA and vSAN) and OnApp Storage. For an in-depth product analysis on OnApp, please refer to the following article: OnApp Scale Out SAN For The Cloud.
Why ScaleIO for EMC?
With that as a background, what does EMC want to do with ScaleIO? As I mentioned in the opening, EMC already has a very robust storage portfolio including the recently announced software defined storage solution ViPR. The one element that the entire EMC storage portfolio lacks, however, is the ability to aggregate direct attached storage into a shared pool; ScaleIO would provide them that capability. One thing seemingly lacking from ScaleIO is a robust set of data services like tiering and replication. More importantly, it also lacks the ability to integrate with SAN attached external storage. In fact, most HyperScale solutions are sold to replace external SANs not compliment them. Through ViPR, EMC could integrate ScaleIO into its portfolio easily while adding robust data services to the ScaleIO product.
Other than the intended sweet spot of HyperScale storage, Web 2.0 Hyperscale data centers, EMC could modify this software to then compete with VMware’s and HP products while delivering a very inexpensive starter solution that combined compute and storage and eliminates SAN expense. Then as the end user needs grew, other components of the EMC portfolio could be layered into it.
ScaleIO would also give EMC a competitive offering in the managed service provider and cloud service provider spaces; a market that called, “Cloud Defined Storage” which is basically HyperScale storage with a provider focus. Finally, it would give EMC an interesting enterprise offering that could serve as a performance tier by aggregating PCIe based SSD into the server infrastructure.
The downside for EMC is that it brings yet another storage product into an already crowded portfolio which could lead to still further confusion for EMC customers and prospects. On the other hand, it seems that EMC has successfully navigated this problem in the past and I don’t see why ScaleIO would make things worse. Clearly there would be overlap in the offerings so EMC prospects and customers would have to examine the portfolio to determine which solution would make the most sense.