You are currently on IBM Systems Media’s archival website. Click here to view our new website.

POWER > Business Strategy > BI and Analytics

Analyst Joe Clabby on Why Power Systems are Best For Business Analytics

illustration: Caroline Tomlinson/getty images

Business analytics is becoming an essential tool for the entire enterprise. For the past 20 years, it’s mainly been used by specific departments, but now many organizations expect every department to access and use the insight provided by business analytics. This new emphasis impacts IT, as an enterprisewide business analytics focus requires a more holistic view. To learn more about the challenges and solutions, IBM Systems Magazine, Power Systems edition spoke with Joe Clabby, analyst and president of Clabby Analytics.

Q: Clabby Analytics has three maxims for buying and deploying information systems running business analytics. What are they and why are they so important?
It’s a very fundamental belief at Clabby Analytics that no single processor can do all computing jobs optimally; that’s the first maxim. Computing jobs fall into three categories—serial computing, compute intensive and parallel computing. The mainframe is known for being the strongest serial or batch job processor in the market and is capable of running very large transaction-processing workloads. Due to recent improvements, mainframes can now run compute-intensive applications and are clearly moving into the business analytics sector.

Power Systems* servers are known for running compute-intensive and parallel-computing jobs and are usually associated with AIX* or UNIX* environments. But now Power Systems servers are being positioned as market-leading business analytics systems that can run AIX, IBM i and Linux* applications. Linux is associated with Apache Hadoop, a programming model for running queries against big data databases. Power* technology is excellent at running Hadoop and other business analytics applications.

Let’s also look at x86 versus Power servers. Although x86 is good at processing many fast threads, it can only execute two threads per cycle. Power Systems servers can execute four. What this means is right off the bat, Power technology is twice as powerful. Power servers are also known to do compute-intensive jobs more efficiently. Both systems perform well in processing parallel tasks, but to scale x86, you must throw more processing cores into the configuration—more cores burn more power and vendors often charge for applications based on the number of cores. So using x86 solutions may drive up license costs. My point is that, despite what many IT buyers think, x86 is not the answer to running the most optimized solutions, as it doesn’t do every job optimally.

Q: The second maxim is that buying integrated infrastructure solutions from one vendor is the most cost-effective move. Why?
If you buy your stack from one vendor, you get a better-integrated stack. Let’s compare Oracle and IBM. Oracle is in the business-application solution business as well as the middleware, operating system, database and hardware businesses. Oracle can build a completely integrated environment from top to bottom. This benefits customers because they don’t have to go through the expense of integrating a lot of software.

In contrast, IBM doesn’t sell applications per se, but works really well with the ISV community and builds the underpinning infrastructure, database and systems environment for the ISVs. IBM offers solutions for three distinct processors—POWER*, System z* and x86. IBM has an integrated database and infrastructure, and can deeply integrate these elements to get high performance out of the system. IBM offers both integrated infrastructure and three server environments on which that infrastructure can be deployed and optimized. This gives IBM buyers choice as well as better-optimized systems that can be tuned to execute specific workloads extremely efficiently.

Here’s an example: IBM PureSystems* technology. You can get up to 80 percent virtualization running PureSystems technology compared with traditional x86 blades that run in the 25 to 30 percent virtualization range. If you construct your own system from components, the best you’ll get is 40 percent integration. From the start, you get better performance from PureSystems. If you’re buying an integrated system from IBM and getting 80 percent virtualization, that’s like getting a free server.

Shirley S. Savage is a Maine-based freelance writer. Shirley can be reached at



2019 Solutions Edition

A Comprehensive Online Buyer's Guide to Solutions, Services and Education.

Picturing Success

Capture insight from real-time operational data with smart analytics systems

Unlocking the Secrets of Success

An introduction to business analytics

Layers of Integration

Components of a complete analytics solution

IBM Systems Magazine Subscribe Box Read Now Link Subscribe Now Link iPad App Google Play Store