Tag Archives: InMemory

in memory option

Oracle 12cR1 patchset 1 is due this month and there is a new parameter that you can set to boost your performance. It is a bit of a SET "_FAST"=true parameter.

The in memory parameter is part of the sga. It is not mandatory to size it correctly, even if you do not have enough memory to hold your complete database, you can still play around with this parameter.

In a way, alter table t inmemory reminds me to the Oracle 8i alter table t cache and the Oracle 9i alter table t storage (buffer_pool keep).

But it is not free, I expect something close to the partitioning option, and it surely requires Enterprise Edition.

And also Oracle makes big noise about it, experts talk about a 1000x improvement, watch Database Industry Experts Discuss Oracle Database In-Memory.

The in memory cache is redundant with the database cache. It stores columns instead of blocks (or even results with the RESULT CACHE in 11g)

Don’t miss the Oracle Blog of @db_inmemory

Providing in-memory database is also positioning against HANA, a SAP in memory database. From OTN : Oracle Database In-Memory
Versus SAP HANA

A few years ago, Oracle acquired TimesTen. TimesTen is an in-memory database that works differently, where you can have fast response time (microseconds?) and could lose transactions (better faster than zero-data-loss). While TimesTen improves transaction speed, inMemory mostly improves queries (not writes).