Service Level Management: Not Just for IT Anymore

by Rick Sturm

Recently, there has been conversation on this site's discussion list that can be distilled down to two basic questions. First, is it possible to implement an effective Service Level Management (SLM) process, including Service Level Agreements (SLAs), for non-IT services? And, second, if it is possible, how does one go about doing this?

Certainly it is possible to implement an SLM process for non-IT services. It is done everyday in a wide variety of industries. In some cases, the process is formal and structured, with a written SLA. In other cases, the commitment to providing an acceptable level of service is implicitly understood. Let's begin by looking at the question of what is meant by non-IT services.

One non-IT area that utilizes SLM is the telecommunications industry. Carriers commonly offer service level agreements to their business clients. Also, regulators impose minimum standards for performance (e.g., availability) and may impose penalties for failing to meet these requirements. While providing telecommunications is not an IT function, there is little difference between the challenges faced (other than cultural) in establishing an SLM program in an IT setting and in a telco. Both environments are built around computers and networks. That equipment is either well instrumented or readily lends itself to instrumentation and thus an automated set of processes is available to support SLM.

Switched-Off SLM

Now, let's consider a more challenging situation. Is it possible to implement an SLM process in an environment that is devoid of instrumentation and not suitable for instrumentation? In today's business world, it is difficult to conceive of a venue that cannot be instrumented. In the past we might have been able to agree that delivering letters and packages was such an instance. However, in recent years, companies like Fedex and even the U.S. Postal Service have demonstrated that even in these service areas, it is possible to instrument processes. However, it is conceivable that you could be faced with a situation in which instrumentation is not possible, perhaps due to prohibitive cost or security concerns.

Over the last few years, experience with SLM and SLAs in the IT environment has conditioned many people to think in terms of software tools when it comes to implementing SLM/SLA. Certainly, there are many products available in the IT and telcom environments to assist with the SLM and SLA processes. Companies have bought these tools because the tools add value, and have made it faster, cheaper, or easier to administer the processes. Tools can capture more data and sometimes different (more accurate) data than can be captured manually.

Yet, in spite of the value that tools add to the SLM processes, the fact remains that Service Level Management does not require automated processes. The same principals of SLM apply, even in a totally manual environment. The same six processes are required:

• Define
• Measure
• Assess
• Set objectives
• Monitor
• Improve

The difference between a manual environment and one that enjoys the benefit of automation is found in the way that data is captured and reports are generated. Let's take a quick look at those six steps.

Begin At the Beginning

If you want to establish an SLM program, you must first define the service that is to be addressed. There is no automation necessary at this stage, just a clear understanding of the service being delivered and its boundaries. (Interesting disparities in viewpoint sometimes emerge between the provider of a service and the client regarding the definition of the service.)

Before proceeding to the next step, "Measure," it is necessary to decide what factors will indicate the level of service being provided. Here, the manual and automated worlds diverge. Those indicators of level of service must be measurable and accurately estimated. For this step in the process, IT and telecom managers are faced with an embarrassment of riches. They have available more data than they can ever hope to employ. Their challenge is turning data into information and identifying those few key pieces to use in reporting service levels. Their counterparts in a manual environment, where automated data collection is not possible, are faced with a far different problem. Their choice of metrics to use in SLM is not a question of which is best, but rather, "Is anything available?" They must look for any relevant captured data or find a way to capture some. Regardless of what metrics are chosen and regardless of how the data will be captured, the key to the effectiveness of the SLM process is selecting metrics that both the service provider and the client agree are meaningful.

Where do you begin to find data, if automated sources aren't available? The accounting department is a good place to start the search. Accountants do have a penchant for collecting numbers and also for archiving them. Logs, receipts, timecards, shipping records, production data, and more are all possible sources of useful information. There are still companies that rely on a stopwatch to measure the response time of an online system!

It may already be apparent that another major difference between the automated and manual approach to SLM is the timeliness of the information. In a highly automated SLM process, it is possible to provide information on a real-time basis so managers can take corrective action, sometimes even before Service Level Guarantees have been violated. In the manual world, the time lag is more often measured in hours, weeks, or even months.

Brainpower is Key

The next step in the SLM process, "Assess," varies only minimally whether data collection is highly automated or totally manual. In the automated world, captured data is already in a file and can be made available for analysis with a variety of software tools. If the data has been collected manually, it subsequently must be converted into electronic form for analysis.

Setting objectives has always been a manual process, and is the same in the automated and manual worlds alike. Much like the "Measure" step, "Monitoring" brings the same challenges of capturing relevant data. The last two steps in SLM, refining and improving the service, are also largely manual.

I hope that by now you can see that Service Level Management in a highly automated environment (e.g., an IT shop) differs from SLM in a manual environment primarily in how data is captured. Obviously, the automated capture of data is preferable. If you are going to implement SLM in an environment that has little or no automated data collection, I recommend that before resorting to manual data collection, you give critical and creative consideration to possible automated sources or ways to create automated sources of data.

Now for your opportunity … if you have written, or know of an SLA for a non-traditional service (i.e., not covering computing or communications services), send me an e-mail describing it. The one that I consider to be the most interesting will receive a copy of "Foundations of Service Level Management" autographed by me.

Rick Sturm is the founder and president of Enterprise Management Associates, the first technology analyst firm to specialize exclusively in management software and services.

home