Raising the bar on service level management

For some time, organisations have been under the impression that the negotiation of comprehensive service level agreements (SLAs) with third-party providers was all that was required for the successful optimisation of their information technology (IT) infrastructures.

In truth, it is the management of these support agreements that is key to the efficient running of an organisation`s IT resources and a significant return on investment (ROI) over time.

Usually, SLAs or even more loosely worded service contracts are entered into between service providers and customers in a bid to define the services provided and agree on the metrics associated with these services.

While an SLA is important in terms of defining customer expectations, it is almost worthless when it comes to the optimisation of resources - from both the service supplier and the customer - to maximise IT efficiencies.

Most SLAs are developed after first defining existing services and then by deciding what new services are required to meet operational objectives.

An SLA is nothing more than a tool that helps service providers to define benefits against costs - and deliver impressive communication services to demonstrate to the customer (sometimes falsely) that he is getting the best value for money.

It is the effective management of SLAs in tandem with business and operational objectives that will result in the rapid identification of what are acceptable and unacceptable service levels and give customers a handle on the risks associated with these levels.

The techniques, skills and technologies linked to the management of SLAs are grouped under the service level management (SLM) banner, and the correct application of the methodologies associated with SLM will result in a host of ROI benefits.

With this in mind, the challenge is to define SLM objectives not by traditional goals such as network performance and availability, but rather to measure success in terms of the appropriate utilisation of the corporate infrastructure.

SLM initiatives so constructed will be characterised by the inclusion of important elements, such as response times to failures or predicted failures and user network access levels.

In the past, the SLA boilerplate guaranteed certain levels of network and server availability. However, it often failed to specify that users should have access to business-critical applications with an appropriate level of performance, or response time.

What good is a network if you can`t access the applications running on it?

Companies need to raise the SLM bar by guaranteeing end-user quality of experience (QoE).

This is often anathema to service providers who, while motivated to provide SLM functions, offer only the bare minimum in terms of service levels necessary to support their sale.

It is therefore up to consumers to demand more from their service providers.

Traditional approaches to SLM involving event filtering, correlation, and even root-cause analysis, simply fall short of guaranteeing QoE.

These approaches fail to accommodate the dynamic nature of modern networks and are unable to automate the labour-intensive, time-consuming steps required to investigate and correct each potential problem.

In the past, SLAs included "nuts and bolts" guarantees mainly because the technology was not available to measure real end-user QoE.

Once providers realise they need to provide QoE-related service guarantees, they will hopefully take the appropriate steps to lift the burden that is still on the shoulders of the operators, who perform all tests, diagnose the cause, measure the business impact, and then decide on and execute an appropriate response.

The good news is that there are management products on offer that have a QoE emphasis - monitoring end-user applications independently and reporting problems back to a central console.

Once collected, the metrics can then be applied to SLM. This task should be even easier in a server-based computing environment (such as Citrix or Terminal Services), since the actual end-user experience is occurring on a monolithic server in a data centre.

It will be interesting to see if vendors, service providers and consumers consider these issues and provide - or demand - QoE-related service-level objectives and guarantees in future.