Investing for the Long Run
Study shows AIX leads the pack with SAS in grid environments
As one of the leading business analysis software providers in North America, SAS helps companies assemble data pools that can be analyzed and collated to produce useful business intelligence. In recent years, companies have had the option to deploy the comprehensive SAS solution on grid architecture, allowing for increased flexibility in computing-resource allocation and utilization.
To evaluate the impact of the IBM platforms running the analytics suite from SAS in the grid environment, IBM engaged predictive performance expert Solitaire Interglobal Ltd. to analyze the potential benefits of its AIX* operating system against commonly available marketplace competitors. In a recent interview, Leamon Hall, manager of the IBM and SAS alliance, and Kat Lind, chief systems engineer at Solitaire, reviewed the study’s key objectives and outcomes.
IBM Systems Magazine: Why did IBM commission this research project?
Leamon Hall: Given the state of the economy in recent years, our SAS customers are clearly trying to get the most “bang for their buck” when it comes to investments in their processing environments. So when it’s time to select a hardware platform, they’re clearly interested in those with a strong, demonstrable return on investment. When we decided to take a closer look at this, we had seen market evidence that a Lintel [Linux* on Intel*] processing environment may show a lower cost of acquisition, but we were more interested in total cost of ownership. We suspected that deploying a complex business analytics environment, such as SAS, could quickly eat away at that original cost of acquisition advantage and damage the overall ROI. That was the theory we wanted to test.
ISM: In addition to IBM’s AIX operating system, what competitive systems were tested?
Kat Lind: We tested a variety of Lintel and Windows* platforms. We worked to ensure that the comparison platforms were either current or minus-one releases and that they were commonly deployed across almost every commercial vendor. This helped ensure that the data we generated would be relevant for apples-to-apples comparisons between platforms.
ISM: What is grid topology, and how is it distinct from other forms of IT architecture?
LH: In a traditional SAS processing environment, our customers typically have one large computer that processes massive volumes of SAS data residing on storage. As these volumes of data have grown exponentially, the data movement and I/O requirements become the bottleneck. For these types of workloads, the grid architecture helps alleviate the processing and I/O imbalances by enabling a massively parallel processing, or MPP, environment that allows individual jobs to be divided into subtasks and run in parallel on whatever hardware resources are available within the grid. While a number of platforms can run this complex SAS grid architecture, we believe that IBM has a technological advantage. IBM has invested a considerable amount of research and development into AIX and Power Systems* servers. For that reason, we asked Solitaire to provide us with a detailed analysis for platform selection. We wanted to know where Power* AIX stood compared to other alternatives.
“Executives were substantially more pleased with AIX
as deployments got more
volatile and complex.”
—Kat Lind, chief systems engineer for Solitaire Interglobal Ltd.
ISM: What metrics were used in the testing and how were those measures chosen?
KL: It’s important to remember that the human interpretation of a system’s performance is just as important as a totally objective set of measures. So, the main subjective findings we included in the whitepaper are those that provided clear differentiation. For example, our customer satisfaction findings were distilled from feedback gathered at multiple organizational levels. While system performance may be a key priority for a company’s IT group, it’s often a secondary or tertiary priority for those with line-management responsibilities. For executives, satisfaction issues generally circle around how much a technology investment costs and how many complaints they hear about its performance.
ISM: Let’s start with customer satisfaction. What were the key findings in this research category?
KL: The key finding here is that executives were substantially more pleased with AIX as deployments got more volatile and complex, factors that are increasingly present in larger IT environments. While the initial cost made executives in smaller organizations not as happy, that finding was mitigated a bit by the low number of complaints about the AIX platform in all situations, regardless of size. On the technical side, it was a bit surprising to find that satisfaction levels regarding AIX were high across the board—small, midsize and large environments. That’s a place where research normally uncovers some difference of opinion, because these people have day-to-day exposure to the platform’s performance.
ISM: Total cost of acquisition and total cost of ownership are important measures for cost-conscious IT leaders. What stood out in this area of the research?
KL: Most organizations tend to view total cost of ownership as simply the initial equipment costs, rather than the cost of the purchase over time. As an initial purchase, the AIX solution does not look good for small shops with low volatility, since the cost of the base equipment is higher. But when you add in the total costs of ownership, which include licensing, staffing and other long-term operational considerations, the picture significantly changes. At that point, the AIX solution becomes much more cost-efficient.