Service Level Agreement

An agreement on a targeting goal of service level between service providers and service users is called a service level agreement. Service Level Agreement is also called SLA for short.

Overview
Services are unsubstantial and are difficult to grasp them in the first place. Especially, medium- to long-term services could cause a gap between service providers and users in understanding about the contents. Thus, clearly writing a statement about service level in a contract document or drawing a dedicated SLA document and then agreeing on the service level in advance in order to get rid of such a gap are called SLA. Originally, SLA was a network service provider's system to guarantee its level of services provided to users, and required that guarantee of service level, such as the minimum network speed of a service line provided, average delay in the network, and the upper bound of unavailable time, and penalties in case those guaranteed terms are not met (e.g. monetary penalties like a discount of the service charge and formulation and enforcement of an action plan) are included in the service contract. However, recently SLA is introduced to various fields besides network services.

Purpose of SLA

 * Maintaining high quality service level.
 * Improving user-friendliness by clearly stating service level.
 * Clarifying a range of responsibility of each of service providers and users.

Outcome of SLA

 * Clarification of a cost reasonable for service level.
 * Understanding about problems and issues of existing systems through agreement, accomplishment, reporting, and improvement.
 * Trust relationship established between service providers and users.

Background of Development of SLA
Originally, SLA was a system introduced by major network service providers in the late 1980s. In Japan, it was first introduced by IIJ (Internet Initiative Japan) in June, 1999 as "Service Level Guarantee System."

Nowadays, SLA is used in various services besides network services, such as hosting services, ASP, online application services, outsourcing services for operation and maintenance of corporate information systems, maintenance services of OA equipments etc.  In addition, expecting the outcome of SLA, these days it is regarded as an effective technique not only for a contract between companies but also for a system within a company.

General SLA Items
Examples of general SLA items include the following. Availability
 * Service Time: Service hours in which users can use a service.
 * Service Availability: Probability that users can use some specific service.
 * Time to Recovery: Time taken from detection of a trouble to recovery of a service, which allow users to use the service again. This is also called MTTR (mean time to recovery).
 * Mean Time Between Failures: Average length of time of uninterrupted system operation. It is also called MTBF.

In addition, availability can be calculated by using a formula, ("Up Time" / ("Up Time" + "Down Time" )) / 100. It is represented as a percentage of "Up Time" of the system. The larger this value is, the shorter is "Down Time." Furthermore, the probability of system failure is decreased, and the time to recovery becomes shorter. By the way, 99.999% availability is called five nines, and many service providers aim to achieve availability that exceeds five nines.

Performance
 * Response Time: Time taken until response is returned since an end-user executes some operation.
 * Turn-around Time: Time taken from the beginning of some operation to its end.
 * Throughput: The number of operations processed per unit of time.

Capacity and Data Maintainability
 * Disk Capacity: Guaranteeing disk capacity assigned to a user
 * Disk Utilization: Enforce a limit of disk utilization so that it does not exceed the pre-determined value.
 * Backup Frequency: Specify backup frequency of some data
 * Backup Retention Period: Specify how many generations of backup media are kept.

Others
 * Help Desk and Service Desk: Average number of calls, time to solve problems, ratio of cases in which problems are solved at the first call, and so on.
 * Security: Specify whether anti-virus software is used or not and the frequency of virus scans, enforcement of access control, acquisition and check of audit logs, installation of intrusion detection systems, application of security patches and its frequency, and so forth.

SLA of BPMS
Since recently more and more BPMS products are provided online, the level at which a service is provided is an important factor in BPMS as well. Moreover, SLA that exceeds five nines is required in some use cases. Therefore, it is desirable that agreeing on satisfactory SLA between service providers and users before providing a service.

Related Articlea

 * Workflow Engine