IBM builds up its cloud with Netezza as a service and NoSQL as software

2 Comments

IBM announced a bunch of new cloud data services on Monday, including an intelligent data-preparation tool, an in-memory analytic database and even a local version of a historically cloud-based database. It’s an impressive and in some cases even unique set of capabilities that complements the work IBM has been pushing with its Bluemix platform since February.

Here’s what’s new:

  • DataWorks seems potentially very useful as a cloud-based take on the handful of new data-preparation technologies currently emerging from software startups.
  • Cloudant Local is, as far as I know, the only NoSQL database offered both as a cloud service and a local deployment by the same company that built it. (Cloudant launched as a cloud database startup in 2008 and IBM acquired the company in 2014.)
  • With the Netezza-powered dashDB, IBM joins Amazon Web Services, Google and Microsoft with a homegrown analytic service built atop columnar database technology.

Already on Bluemix, IBM has gone beyond traditional PaaS services (various managed database, big data and monitoring services, for example) and exposed various Watson and other machine learning capabilities as Bluemix services via REST APIs. On the application front, IBM even released a freemium analytics service that incorporates Watson’s natural-language prowess and (like all good data analysis software now does) targets the coveted “business user” with good visualizations and a slick UI.

A sample of what's available in the Bluemix platform.

A sample of what’s available in the Bluemix platform.

The loosened-up approach to building cloud services is personified when you walk around a conference like the IBM Insight big data conference taking place in Las Vegas this week. The business suits are still there, but the tie is often gone and the top button is undone. Some of the strongest sessions feature startups building cognitive computing applications on Watson rather than large companies doing business intelligence.

But the driving question for IBM across its entire cloud business still has to be who’s going to buy the services it’s selling. As good as some of it looks on paper, IBM is pretty late to the party when it comes to delivering any legitimate type of cloud computing platform. One could make the argument it’s showing up with a couple cases of Sam Adams after other guests already built their own brewpubs in the backyard.

During a press “roundtable” at IBM’s conference on Monday, Watson division VP Stephen Gold noted that the new Watson services on Bluemix have attracted 1,500 developers and powered 2,200 applications in the three weeks since they launched. Imagine what that number would be if AWS released something similar.

Microsoft EVP Scott Guthrie and CEO Satya Nadella. Credit: Jonathan Vanian / Gigaom

Microsoft EVP Scott Guthrie and CEO Satya Nadella at a recent cloud computing event. Credit: Jonathan Vanian / Gigaom

Competing in the long run against AWS, Google and Microsoft means competing against companies that have more servers, lower prices, multi-year headstarts and much more experience managing webscale applications. They’re spending billions of dollars per quarter building out their infrastructures. Like IBM with Watson, they all understand machine learning and artificial intelligence, too. Microsoft has a particularly compelling message and platform aimed at the enterprise developers and CIOs IBM is presumably targeting.

IBM has been talking the cloud computing talk since about 2007, and now it’s finally walking the walk. Its decisions to sell off once-valuable businesses such as servers and microprocessors will look prescient if it can sell enough enterprise developers and buyers on Bluemix and SoftLayer, IBM’s cloud infrastructure business. First things first, though, it needs to convince those folks that all the blue jeans and T-shirts in Silicon Valley and Seattle — regardless how persuasively they argue otherwise — don’t really understand what enterprises need.

2 Comments

Steve

Netezza isn’t columnar. It is a shared-nothing MPP architecture, but actual storage is row oriented.

Comments are closed.