Many new Oracle features require new memory structures, which are commonly cached in the shared pool. This has created an incredibly difficult technical problem for Oracle software architects. Not only are the existing memory structures difficult to manage together, but with the constant addition of new memory structures, the requirements continue to change. This forces memory management optimization to remain flexible while at the same time being pushed, we hope, toward optimal performance. But regardless of the challenge and the claims, as they say, it is what it is, and that is what we, as performance analysts, have to deal with.

My objective in this chapter is to clearly present the relevant architectural components of the shared pool, the performance challenges we face, and how we can alter the situation in our favor to improve performance. Many of the architectural concepts-such as hashing, hash chains, latches, and mutexes-have already been covered in earlier chapters, so let's dive right into the shared pool.

There are many problems-some would say challenges-associated with the shared pool. The constant inclusion of new memory structures, new requirements of all kinds, and highly dynamic memory management contribute to the situation.