| Some Oracle 11g Tidbits [message #214724] |
Wed, 17 January 2007 15:07  |
smartin
Messages: 1803
Registered: March 2005
Location: Jacksonville, Florida
|
Senior Member |
|
|
I haven't searched this site or googled to see if anything about this is already making its rounds, so apologies if this sort of thing has already been posted. And I haven't been able to stop by much at all lately. Hopefully 2007 will be better.
And I'm posting this here in this forum intentionally because it is more of a "hey, did you hear about this..." sort of thing. None of this is official, or final, or complete, etc.
But I went to one of the free Oracle tech days (here in Jax) today. Among other things, they talked about some upcomming 11gR1 features (no release date, I'm guessing in time for the next Oracle Open World, which is earlier this year than last, and it will be all about 11g).
Some features off the top of my head are:
- table compression for OLTP systems (I wonder how they will do that)
- standby physical database that can be opened read only at same time redo is applied
- new partitioning type called interval, which is same as range except it will auto create new partitions as needed instead of you having to manually allocate or prealloc them
- possibly some sort of new database buffer cache specifically for read mostly data, that would presumably be accessed with less overhead / enqueues / latches than otherwise
- online oracle patch upgrades of a single database (can already do with RAC and DG)
- moving away from sys as sysdba being all powerful single entity
- flashback transaction; like other flashbacks, except at the consistent transaction level
- possible support for terabyte SGA sizes and petabyte database sizes
- some kind of query plan optimizer than can do its work during maintanence windows
- some kind of ADR (like with AWR, but for audit data)
- typical incremental upgrades and improvements; increase supported datatypes in specialized situations, etc, etc.
- several composite partitioning types that I didn't memorize, like list list, list range (I think those were two...but figure a bunch of new ones that aren't there now)
- some kind of (quite interesting sounding) feature where when you create objects (tables, packages, etc) you assign an edition (version) to them. You can then have multiple editions existing in the db at the same time, which would allow you to upgrade an application online by letting user sessions roll from the old version to the new
- some kind of database replay feature whereby you could replay all or part of a databases activity in its entirety exactly as it occured the first time. Highly useful for a testing environment, where you would create a "snapshot standby database", upgrade that db to say a new oracle version, capture real live production activity, and play that against the standby snapshot database to test that everything works
- a hint at the possibility of virtual columns inside tables (cols based on a formula), which could at the very least be used in partitioning schemes
I took notes, and can try and check them in case there was anything else "biggish" that they covered. It was brief.
Linux and Solaris seem to be the two platforms that will release first.
Unfortunately, I did not win the drawing for the Apple IPOD Shuffle, and got the coffee mug instead.
Scot
[Edited for typos and to add virtual columns]
[Updated on: Thu, 18 January 2007 07:49] Report message to a moderator |
|
|
|
|
|
) 2 Votes
] 
Blog
Search
Help
Members
Register
Login
Home
