It's very common that high availability means duplicate hardware so costs grows up.Nowadays, CIOs and DBAs has the main challenge of reduce the money spent increasing the performance and the availability. Since Grid Infrastructure 11gR2, there is a new feature that helps them to afford this challenge: Server PoolsNow, in Grid Infrastructure 11gR2, you can define server pools across the cluster setting up the minimum number of servers, the maximum and how important is the pool.For example:Consider that "Velasco, Boixeda & co" has 3 apps in a 6 servers cluster.First One is the main core business appSecond one is Mid RangeAnd third it's a database not very important.We Define the following resource requirements for expected workload:1- Main App 2 servers required2- Mid Range App requires 1 server3- Is not a required app in case of disasterThe we define 3 server pools across the cluster:1- Main pool min two servers, max three servers, importance four2- Mid pool, min one server max two servers, importance two3- test pool,min zero servers, max one server, importance oneSo the initial configuration is:-Main pool has three servers-Mid pool has two servers-Test pool has one serverLogically, we can see the cluster like this:If any server fails, the following algorithm will be applied:1.-The server pool of least importance2.-IF server pools are of the same importance, THEN then the Server Pool that has more than its defined minimum servers Is chosenHope it helps
