Enterprise Service-level Agreements
The popularity of service-level agreements (SLA) has been growing for some time. SLAs entered broad networking consciousness back in the days of the first data service offerings. Back then, many companies took a second look at their private line-based networks in light of new data services such a frame relay. In particular, data services offered companies the attractive option of pushing some of their costs -- such as network provisioning and maintenance -- off onto the carrier who offered the service.
As attractive as this sounded, however, there was one major stumbling block. Asked to abandon their private networks, planners knew that they were turning their data over to a public infrastructure in which there was a risk that capacity was oversubscribed and congestion was a possibility. Before committing their data (and their careers) to the service provider, they demanded some mechanism that would ensure that they were, in fact, going to get the service they were paying for or, at least, that they were going to be reimbursed when the service didn't live up to the provider's promises.
SLAs fit the bill for providing this mechanism. The SLA concept was revised and modified to accommodate the requirements of the new data service offerings. SLAs were structured and written around the metrics that defined the new data services, and as a result, corporate planners were ultimately satisfied that they had in hand an effective mechanism that amounted to a performance agreement between their company and the data service provider.
SLAs proved so successful in this market that the concept quickly became part of networking consciousness and is now crossing over into the enterprise. Enterprise SLAs provide the same sort of assurance as their older, wide-area networking brethren, with the major difference being that they amount to agreements between organizations within the same company, rather than agreements between companies. Otherwise, their purpose is the same, namely, to provide a mechanism in which explicit criteria are used to measure the quality of a service between an organization using it and the organization providing it.
Obviously, enterprise SLAs are a very attractive concept. Ultimately, however, their widespread adoption is dependent on the availability of tools that can be used to monitor and measure compliance with the explicit performance criteria written into the terms of the agreement. Fortunately, tools that meet this need are already appearing on the market. There are now quite a few companies that position themselves as providers of utility software that can be used to measure enterprise SLA compliance.
Readers will note that, to this point, the exact nature of the enterprise service under SLA management has been kept vague. This is because, at this stage of the market, different companies are pursuing different strategies with regard to the services they monitor. Some, following in the footsteps of their wide-area networking kin, have chosen to focus exclusively on the enterprise network. Their goal is to deliver the tools that will give IT departments the ability to establish network performance SLAs with their clients.
Others have taken a different path in that they focus entirely on the transaction performance of the application. Over time, companies in this category expect to take a more comprehensive approach to the problem by integrating application transaction performance with the performance of other enterprise services such as the network. With these comprehensive tools, IT departments will be able to write SLAs that cover a range of their services: application performance, network performance and whatever other services the tool manufacturers decide to monitor.
Companies also differ by the scope of the ancillary tools they provide. Some offer a limited toolset because they have decided to focus primarily on the monitoring aspect of SLA compliance. Their philosophy is to provide the best tool possible to do one job extremely well. Others offer a range of tools that treat SLA compliance as an end-to-end process. One tool is initially used to establish baseline performance for a particular service; the next is used to monitor the service on a day-to-day basis; and finally, simulation tools are used to spin what-if scenarios that calculate the impact on service performance if changes are made to the environment.
It will take some time to determine if enterprise SLAs catch on with the same enthusiasm as their data service kin. It will also take some time to determine the solution set most desired by IT departments: point service monitoring products or tools that take a comprehensive approach to SLA compliance. --Sam Alunni is vice president of networking at Sterling Research (Sterling, Mass.). Contact him at email@example.com.