In June, SiliconAngle’s John Furrier published one of the first posts that IBM was going to acquire Storwize for $140M.  In late August, Dave Vellante reported on SiliconAngle that the deal had been closed and provided some details of the transaction from the perspective of the Storwize CEO Ed Walsh. Vellante’s story chronicled how Walsh and his team cleaned up the financials of the company; and packaged the technical nuggets into a marketing story which catapulted the valuation of Storwize leading to the acquisition by IBM.

What’s Next – Real Time Compression

Now that IBM owns Storwize what’s the play?

The word is that IBM has a big announcement coming in October and will be dumping the Storwize name and instead is going to market the compression technology under a new brand – Real Time Compression.  It’s too bad, as Storwize is a good name; but this is common when a large company makes an acquisition – they’ll try to take the new company and fit it into a brand strategy that’s more consistent with their agenda.  The key question with IBM: What is their brand strategy / agenda for storage?

“Real Time Compression” (as a name)  isn’t that sexy but it definitely says what the Storwize technology does.  The thing about the Storwize technology that not a lot of people understand is that it can actually improve performance because of it’s in-line implementation.  Many disk array architects looking at data reduction technologies don’t want to impact storage performance so they’re very careful about how they implement data reduction approaches. Most wouldn’t even consider doing it in real time.

In my data center I always choose the cautious approach to deployment, however, when I see great technology, I do proof of concepts, I look for a competitive edge.

Products like NetApp’s dedupe and EMC’s Celerra compression are overhead intensive which is why they’re used as a post process or batch operation – specifically so they don’t get in the way of performance. If these companies tried to use their technologies in real time they would kill storage performance.

Real time compression as implemented by the engineers at Storwize on the other hand can actually increase performance because the overhead of the technology is so miniscule that it is easily offset by the increased capacity of the cache that is achieved by compressing data before the array in an appliance. Real time compression also significantly decreases the elapsed time of down-stream off-line processes such as backup and de-duplication and is complementary to these functions.  This is one of the reasons IBM paid $140M for Storwize – a company with less than $5M in revenue. The technology is groundbreaking and solves a very difficult problem.

New Use Cases – Beyond File into Block and Big Data

It will be interesting to see what IBM does with this technology over time. Because this is IBM, don’t expect this to happen quickly, but the other reason why IBM paid up for Storwize is because it had an architecture that allows IBM to extract the IP from the compression engine and place it in a zillion use cases. Today, real time compression is only available as an appliance for NAS filers but in the future, the smart money says IBM will use this technology in file and block-based devices and will embed real time compression as a standard feature of arrays and other appliances (e.g. SVC). Politics inside IBM’s product groups and competing agendas will definitely slow this down but it will happen, eventually.

In addition, given all the talk of columnar compression these days it would make sense for IBM to put real time compression into database machines. IBM sells a nice data warehousing appliance based on DB2 and it just bought Netezza so it would make sense to see the technology find its way to that side of the house over time.

Data warehouses are performance hogs and columnar compression is a key technology being touted in that space. At Oracle OpenWorld last week, Larry Ellison emphasized columnar compression as an important innovation. The reason is that data warehouses are big giant matrixes. Traditionally, databases use a row as a record and the elements of that row are fields (i.e. data, zip code, part number, etc). But compressing that row won’t buy you much. With data warehouses, where you are analyzing tons of historical data, fields will be selected for deep analysis (e.g. price over time based on various factors). This data is columnar in nature which unlike the rows, contain tons of compressable data.

It’s highly likely that over time, the storage devices that support data warehouses are going to use real time compression to complement other techniques like columnar compression to increase compression ratios and speed up operations even further.

Not an Overnight Development

Don’t bet on all this happening fast. The extension of real time compression into all these other use cases will take a long time. IBM is a huge company with lots of development teams that aren’t likely to disrupt their roadmaps radically. In typical IBM fashion, they’ll methodically evaluate and test real time compression and then figure out where it fits. So expect this to play out over the next 18-24 months.  However, they still have a competitive advantage – no one has cracked the code on making compression real-time.

In the meantime, real time compression looks to be the area of emphasis and it seems as though IBM will try to differentiate from all the other data reduction technologies out there by emphasizing not only efficiency but performance. This should play well with data center managers, especially those that were fearful of buying from a small company prior to the IBM acquisition, it did for me.

It will be interesting to see if, for example, IBM is able to sell real time compression into its N-Series accounts – or will the IBM sales reps and channel not push the product because they’ll be afraid to sell less storage. That would be incredibly short-sighted because history has shown when storage costs less customers ALWAYS buy more but psychology and inertia in big companies is sometimes hard to overcome.

The bottom line is that since the XIV acquisition, IBM has been making big moves in storage, data and related businesses with acquisitions (Diligent, Storwize, Arsenal Digital, Netezza) and internal moves (e.g. SONAS). The company seems to be getting serious about competing in storage and playing the integration card. That’s good because it’s clear Oracle has IBM in its sights and IBM needs to move fast on elevating its portfolio and storage brand. Real time compression looks to be a key component.