A good SLA has to list all key success features concerning your business. For example, it, for your business to remain viable, the database cannot be down for more than three hours at any given point in time, that fact be evidently documented. If your database wants to have the capability to accommodate at least 400 concurrent sessions, then it must be clearly stated.
All expectations related to each critical success factor must be documented in real expressions. In other words, the SLA must state actual numbers for each expectation. For example, if your system needs to have standby database located across town, that would cost X number of dollars and it expected to impose a performance strain of Y on the primary production database, due to replication, that should be mentioned. X and Y may not be real numbers, but may be percentage, based on educated guesses. It would help to indicate to the user community and the management what kind of tradeoffs they are looking at to achieve high availability. Merely indicating that there would be some performance tradeoff would not help to place their thoughts on the same page as yours. Actual concrete numbers of some kind will validate your expectations and make it real facts, rather than letting it remain just vague assumptions. Again, an SLA is a contract. So rather than just talk about “if we do this, then that happens” it should be clearly specified this; if we do that, then this happens, but the latter option costs twice as much.
An SLA has to be dynamic and up to date. Some organizations create an SLA and conveniently forget all about it after a few months. The primary reason this happens is because an SLA is rigid; so rigid that it is impractical to adopt. A variety of facts, applicable during the time the SLA was created, have been taken into account. As the organization undergoes changes, however, those facts themselves do not remain static. Accordingly, the SLA has to be revamped periodically to accommodate those changes. Especially in the internet era, where changes and usage patterns occur swiftly, your SLA should be flexible enough to expand and accommodate those changes. If it means having to rework your SLA every two months, so be it! For example, in the case of a startup e-commerce site, the customers the site will draw can only be guesstimated. The initial SLA would be created based on this requirement. After couple of months, however, more concrete figures indicating actual use will be available and can be used to recreate of SLA. The older SLA is, the more of a gap develops between the concerns it addresses and actual reality. The purchase of new hardware/software using the latest technology might make certain clauses in the service level agreement redundant. During such times, the Service Level Agreement has to be updated immediately.
Dbametrix is team of Oracle specialists to provide high availability of DBA services with low costing remote dba plans. For more details, check website http://www.dbametrix.com