IT executives at Al Rajhi Bank fully understand the concept of “workload optimization,” the practice of examining workloads and assigning them to servers best suited to run those workloads. That’s good because enterprises that understand this concept stand to save a lot of money in operational costs and cost of acquisition (particularly the cost to acquire software licenses) compared to enterprises that run all workloads on a single architecture. Workload optimization is also important for Al Rajhi Bank because there’s a lot of work to manage. The bank is:
• The world's largest Islamic bank
• One of Saudi Arabia’s oldest banks, with operations dating back more than 50 years
• One of the largest joint stock companies in the Kingdom of Saudi Arabia.
Unlike banks around the world, Al Rajhi Bank performed spectacularly during the worldwide economic downturn. It stuck to its Islamic-focused lending policies, which require greater loan security than other banks, and avoided a large bad loan burden.
The bank’s IT philosophy has also yielded a competitive advantage. The bank is a strong believer in buying the best—not the cheapest—platform, operating environment, application, and database solutions to suit computing needs. The bank looks at the applications it’s trying to run, determines the level of service that must be met, and then chooses the best system environment for the job. In Al Rajhi Bank’s case, this means it operates several different types of computer systems, including System z’s, midrange UNIX servers, and x86-based servers (in towered and blade configurations).
At the high end of its information systems infrastructure, Al Rajhi Bank operates mainframes. In the midrange, it uses IBM POWER- and HP Itanium-based servers. In the data center and at the branch level, Al Rajhi Bank operates x86 servers. The bank’s IT environment is comprised of heterogeneous systems, each chosen on the basis of being the best type of system to handle a given job.
Workload-Based Server Selection
An example of how workload/server selection occurs at Al Rajhi Bank can be found in a particular payroll processing application the bank runs monthly. According to Ahmed Mohammed Anwar, Al Rajhi Bank’s head of Production, the bank has an application that’s so large—and that must be executed in such a short timeframe—that distributed servers wouldn’t be able to crunch the numbers and enable the bank to meet its processing goals in the timeframe necessary to meet government-mandated deadlines.
In Saudi Arabia, employees are paid once a month (on the 26th and 27th of the month). The government mandates that employees be paid within four hours of the receipt of payment of the payroll funds from the employer. In turn, Saudi Arabia’s banking watchdog organization, Saudi Arabian Monetary Agency, mandates that banks disperse the money they’ve received within four hours of having received it.
“We have to run a big payroll application that consumes a lot of systems resources on those two days,” Anwar says. “Payroll processing is data-intensive—so we need to do a lot of database processing. The transaction processing characteristics of this application are straightforward, but major batch updates need to be run as payroll claims are processed. And no other system in the world does batch processing as well as a mainframe.
“We need a system that can perform payroll processing for hundreds of thousands of customers—and we need extra computing capacity for those two days. So, we make use of IBM’s capacity-on-demand services to process our workloads in the timeframe required.”