If you’re buying a service, you need to know what the service provider is delivering for your money and feel confident you are being charged accurately. But mainframe IT outsourcing arrangements sometimes breed mistrust between supplier and customer because of a lack of clarity around workload measurement and billing.
Contracts tend to be billed on the basis of how much ‘work’ the outsourced mainframe performs, usually measured in MIPS (million instructions per second). The actual terms of the billing agreement can vary greatly; companies might agree to pay on the basis of the peak MIPS measurements during the month, or the top eight peaks in the month, the average daily MIPS consumed or a variant of these.
But while MIPS are a well known measure of mainframe activity and performance, there are different approaches for calculating MIPS consumption and this can complicate issues and cause confusion. I’ve heard a number of real-life stories where MIPs-based mainframe outsourcing contracts have gone awry.
One outsourcing customer had agreed to be billed according to the average MIPS consumption based on the daily weekday peaks during the month. They subsequently discovered that approximately 70% of all the peaks used in the calculation fell in the early hours of the morning – when all the batch jobs were scheduled to run at once by the outsourcer’s staff. While it was easy and convenient for the technical staff to round up all the jobs at the same time in the wee hours, the high MIPS peaks it generated contributed to higher bills for the customer.
The differences in measuring MIPS consumption led to problems for another outsourcing customer who discovered they had been charged using the IBM MIPS calculation instead of Gartner MIPS as specified in their outsourcing contract. The mix-up led to this customer paying 10% more than they should have for the outsourced service for close to 4 years.
It would be easy to surmise from these examples that outsourcers are the guilty party here. But let’s be clear: this is not a case of deliberate deception. Outsourcers are legitimate companies with high standards and reputations to uphold, and the last thing that they want to do is alienate their customers. Errors occur because the outsourcers are managing complex infrastructures with many staff and multiple customers, making it difficult to totally banish misinterpretation, misunderstandings and human error.
One of the problems is the information imbalance in mainframe outsourcing arrangements; all the information about how outsourced systems are being managed is available to the outsourcer, but it is difficult for customers themselves to assess if things are being run efficiently and to check the accuracy of their bills.
Now customers and outsourcers alike are starting to use IT business intelligence tools to monitor MIPS usage and verify that the outsourcing charges are correct. These tools analyse the SMF (system management facility) records that mainframes generate (providing data on system operations including I/O, network activity, software usage, error conditions, processor utilisation, etc.) and turn this into useful insights such as MIPS consumption patterns and forecasts or breakdowns of how disk storage is used by individual applications.
With these business intelligence tools it becomes easier to confirm the accuracy of outsourcing bills. And if there are any areas of uncertainty, the data they provide gives customers the back-up information to enter into discussions with their outsourcing suppliers. The tools can also provide a breakdown of MIPS consumption by different business departments, allowing companies to apportion IT costs fairly and accurately. At the same time the outsourcer benefits by being able to clearly and unambiguously demonstrate that their charges are fair, putting an end to billing disputes.
Sometimes mainframe outsourcing charges can be high if a customers’ applications are not using mainframe computing power efficiently (i.e. are using more MIPS than necessary). The tools we’re discussing can help identify where this might be happening and point to applications that can benefit from mainframe tuning to ensure they are using less CPU.
Both outsourcing customers and the outsourcers themselves should be happy that technology is helping to reduce the information imbalance within mainframe billing. It will mean customers can be more confident about the accuracy of their outsourcing bills and will help to cement stronger relationships based on a common understanding.