IT Service Level Management Solution

Realising value from a sustainable ICT strategy requires the right level of management information, available as accessible IT intelligence.

The need is three-fold, and resoundingly satisfied by SAS IT Intelligence and our skilled experienced services.

• Data centre managers and workload managers need this information for cost effective management of physical and virtual infrastructure, with a focus on advanced analytics to support optimisation for power-effective outcomes balanced with utilisation.
• Those managing the ICT infrastructure need information to create energy optimal architectures in response to business demand
• The business consumers of the ICT infrastructure need information to understand the ICT energy impact of their business activities, to influence demand for services
• Suppliers of ICT hardware want customers to have this information, in the hope that customers will obtain timely empirical data to support upgrade initiatives should newer models have new efficiencies that make it worth replacing old equipment.

The ability to provide such energy consumption information in a cost allocation model that relates to each specific business activity, and for each management information consumer, is a core competence of the approach that we take with our Sustainable ICT solutions.

The sustainable ICT issue is a deceptively complex problem. The combinations of different workloads, usage patterns over time, hardware configurations, and make and model of hardware all conspire to alter the power consumption profile. There is much more to consider besides this, from supporting data centre facilities to distributed desktops and systems.

Using our Capacity and Performance Optimisation models extended from the successful SAS IT Resource Management solution, we can help you by identifying the optimal balance between workload performance and efficient power consumption.

The optimal balance is not the same for each configuration. Not all workloads are the same, and two or more workloads may perform better or worse when combined on the same equipment. Different equipment, and combinations of equipment also affect the balance between power consumption and workload throughput.

Our goal is to ensure optimal power consumption profiles now, with optimal workload combinations, running on the equipment and architecture best suited for it, while providing predictive and modeling insight so customers can effectively manage and plan now based on the predicted future scenarios.

Data Centre Example

Lets take data centre servers as a case point example, though the viewpoint equally applies to supporting data centre infrastructures and distributed desktops and systems.

With our analytics and reporting we already perform Capacity and Performance analytics and modeling to determine optimal ways of managing ICT infrastructure. Traditionally we tune to performance of software workload, to keep business consumers of services happy with such measures as response time and availability. However now we also treat energy consumption as another metric for Capacity and Performance processes, and these processes then help us to obtain optimal energy usage.

For example, for a given combination of workloads we are able to analyse and report the comparative energy efficiencies between different makes and models of hardware in a data centre. We can provide these views across entire data centres, entire enterprises, across all businesses or customers, across all suppliers. With this we can, for each platform, provide baseline data on which to perform effective energy efficiency tuning exercises, taking into account unique combinations of workloads, usage patterns and many other factors.

This presents new opportunities for improvement. Sometimes there are replaceable components (e.g. Memory or other component devices) that can dramatically improve the power and performance efficiency of a machine. The performance tuning question becomes one of matching workloads to the configurations that have the best balance of efficiency and throughput from a power consumption perspective. Sometimes the information will influence the usage patterns of business consumers, or new purchasing decisions for services and equipment.

For example, with this type of analytic capability, data center architects can experiment with different configurations, modeling and tuning according to the relative energy consumption effectiveness of both the data centre architecture and the anticipated workloads to be performed. In some cases, it may be possible to identify component upgrades that can improve the outcome, in others it may be a case of matching workloads to a different piece of equipment. In other cases it may provide the commercial basis for recommending the replacement of specific equipment, or that specific tuning exercises are required.

It becomes an opportunity to KNOW you are getting the very best power to performance ratio out of your existing hardware investments, and a means by which to communicate effectively, with empirical data, to hardware and software suppliers, on what specific power performance features are being experienced for specific workload profiles.

Good news for suppliers

This is good news for suppliers too. It gives suppliers a great opportunity to then step in and assist customers with specific tuning services, because it is often the case that there are better tuning options available, both configuration, devices and components, for specific workloads, that vendors can help customers enable in order to improve power consumption efficiency.

Such numbers also support the business case for purchase of different equipment or devices that can target specific workloads to achieve better power management features. This is particularly the case with older equipment, because if it is otherwise performing well a customer may not consider its replacement a priority unless analysis showed that it was indeed commercially warranted.

Our reporting and analytics provide excellent insight into these issues. We can forecast power consumption according to workload and equipment mixes among many other variables, so that a customer can have plenty of lead time on predicted future problems. Such analytics can also reveal new workloads that may not yet be a problem, they might not be a top ten consumer yet, but are such a rapidly growing consumer of power that they will end up in the top 10. With this type of insight you can take action on such workloads before they become a more serious problem.

Attainable forecasting in dynamic environments

We also have a superb forecasting capability for analysing vast numbers of systems and workload combinations, and deliver new forecast results on a daily basis. To forecast a rapidly changing, dynamic environment was not something that was possible a few short years ago. It could take weeks or even months to produce a sound model on which to analyse and forecast. However with the tools that we have, we can produce first rate models and forecasts on large amounts of dynamic data in a very short time.

In our view, the question of managing ICT infrastructure from a energy/performance balance perspective is essentially a questions answered through our solutions that support the Capacity and Performance management process. It is a process well described by decades of ICT practice, most recently as an ITIL process. Energy consumption is an additional resource that needs to be analysed, modeled and forecast along with all the other traditional ICT infrastructure metrics, and compared and contrasted to these other resource metrics on the basis of workloads, applications and the software based business services that it all exists to.

Conclusion:

Our aim at Kohlbahdin is three-fold. On the one hand those managing the ICT infrastructure need this information to create energy optimal architectures. On the other, the business consumers of the ICT infrastructure need this information to understand the ICT energy impact of their business activities, and lastly suppliers need this information to help customers to understand the commercial advantages that can be obtained from investing in better technologies and services.

The ability to provide such energy consumption numbers in a cost allocation model that relates to any of these three specific business activities is a core competence of the approach that we take with our Sustainable ICT solutions.