WOW I'm busy! Thats a good thing but it sure slows down my blog posting. I apologize for that. I've been working on my latest rant essay to post here, but it's not done yet, and I really should do something Oracle 11g related, so here it is!
As you may know, 11g now manages both the SGA and the PGA automatically though some new parameters, memory_size and memory_max_size.
Memory_size replaces sga_size and pga_aggregate_target, combining the two into one memory pool for Oracle to manage (well, not really one big memory pool, they are still separate, but Oracle deals with all the details under the covers).
Memory_max_size is roughly like sga_max_size + pga_aggregate_target, gobbling up memory for future use should you use it.
So, why the comment about /dev/shm? If you are going to move to 11g's flavor of memory management on Linux, then you might run into this little error:
ORA-00845: MEMORY_TARGET not supported on this system
and you might say to yourself, "Wow, that stinks, I can't use memory_target on Linux."
Well, thats not quite the case. In fact, the problem is likely that /dev/shm is not configured big enough. You will get this error if /dev/shm is not configured large enough to handle the memory you are allocating with memory_target. So, if you are planning on using these new features, be preparing now and increase the size if /dev/shm.
More on this and many other topics in my new Oracle Database 11g New Features book which is available for pre-sales now on Amazon.com. It will be out VERY SOON!!!
Tuesday, September 18, 2007
Wednesday, September 05, 2007
Greetings from Denver
I'm in Denver this week teaching a class. Wow, what a week. I flew to Denver in my airplane to teach. It was about a 4.5 hour flight and it was great! Nothing like flying over the rockies at 12,500 feet to get your blood pumping! Especially when the reported density altitude at the airports nearest to you (which are at about 7500 feet) is something on the order of 9,800 feet MSL (above sea level). You just hope your fuel calculations were right!!
Today has been the anti-technology day. Everything has broken and all at once. My wireless would not work, my cellphone would not dial out, we are using iSQLPlus here and that went south... UGH!
On a good note, I finished up the last chapter in Oracle Database 11g!! Yeahhhh.... So now it's a process of going through the edits and getting it ready to print. I love the writing but hate the editing process!!!
I'll try to post a new 11g feature shortly...
Today has been the anti-technology day. Everything has broken and all at once. My wireless would not work, my cellphone would not dial out, we are using iSQLPlus here and that went south... UGH!
On a good note, I finished up the last chapter in Oracle Database 11g!! Yeahhhh.... So now it's a process of going through the edits and getting it ready to print. I love the writing but hate the editing process!!!
I'll try to post a new 11g feature shortly...
Thursday, August 30, 2007
Oracle 11g - Statisticly speaking!
OK, it's time for another Oracle Database 11g New Feature! Today some new statistics related features.
In Oracle database 11g you now have two new kinds of statistics that you can collect. Collectively these are known as extended statistics. The two kinds of extended statistics you can collect are:
1. Multi-column statistics
2. Expression statistics
Prior to Oracle Database 11g Oracle had no way of understanding the relationship of data within multiple columns of a where clause. Oracle Database 11g adds multicolumn statistics to the mix to try to solve this problem. Now the optimizer can generate more intelligent cost based plans when you have multiple columns in your where clause, based on the combined selectivity of both columns. Multi-column statistics are not generated automatically, when you generate statistics. You have to define the columns you want to generate the statistics on when analyzing the table. Here is an example of generating multi-column statistics on the table DUDE for columns DUDENO and DUDES_JOB. Note the "for columns" syntax that defines the columns to build the multi-column statistics on:
exec dbms_stats.gather_table_stats(null,'DUDE',
method_opt=>'for all columns size skewonly
for columns (DUDENO,DUDES_JOB)');
Expression statistics allow Oracle to collect selectivity information based on the application of a function on a column. This has direct relationship to the use of function based indexes. Again, you collect expression statistics with dbms_stats when you collect table statistics as seen in this example:
begin
dbms_stats.gather_table_stats(null, 'DUDE',
method_opt=>'for all columns size skewonly
for columns (lower(dude_name))');
Now the optimizer can rationally make execution plan choices with regards to the selectivity of the data in the dude_name with the lower function applied.
There is even more in my new book, Oracle Database 11g New Features. It's available for pre-sales on Amazon and should be out in November!
http://www.amazon.com/Oracle-Database-11g-New-Features/dp/0071496610/ref=sr_1_2/102-1704151-7971347?ie=UTF8&s=books&qid=1188535426&sr=1-2
There are more new statistics related features in 11g including publish/subscribe and restore of old statistics!! I'll talk about that in another post.
In Oracle database 11g you now have two new kinds of statistics that you can collect. Collectively these are known as extended statistics. The two kinds of extended statistics you can collect are:
1. Multi-column statistics
2. Expression statistics
Prior to Oracle Database 11g Oracle had no way of understanding the relationship of data within multiple columns of a where clause. Oracle Database 11g adds multicolumn statistics to the mix to try to solve this problem. Now the optimizer can generate more intelligent cost based plans when you have multiple columns in your where clause, based on the combined selectivity of both columns. Multi-column statistics are not generated automatically, when you generate statistics. You have to define the columns you want to generate the statistics on when analyzing the table. Here is an example of generating multi-column statistics on the table DUDE for columns DUDENO and DUDES_JOB. Note the "for columns" syntax that defines the columns to build the multi-column statistics on:
exec dbms_stats.gather_table_stats(null,'DUDE',
method_opt=>'for all columns size skewonly
for columns (DUDENO,DUDES_JOB)');
Expression statistics allow Oracle to collect selectivity information based on the application of a function on a column. This has direct relationship to the use of function based indexes. Again, you collect expression statistics with dbms_stats when you collect table statistics as seen in this example:
begin
dbms_stats.gather_table_stats(null, 'DUDE',
method_opt=>'for all columns size skewonly
for columns (lower(dude_name))');
Now the optimizer can rationally make execution plan choices with regards to the selectivity of the data in the dude_name with the lower function applied.
There is even more in my new book, Oracle Database 11g New Features. It's available for pre-sales on Amazon and should be out in November!
http://www.amazon.com/Oracle-Database-11g-New-Features/dp/0071496610/ref=sr_1_2/102-1704151-7971347?ie=UTF8&s=books&qid=1188535426&sr=1-2
There are more new statistics related features in 11g including publish/subscribe and restore of old statistics!! I'll talk about that in another post.
Monday, August 27, 2007
Oracle 11g Oops...
Been quite busy of late, so my blog has not had quite the number of updates that I'd like. One thing I thought I'd share with you is that I found my first Oracle 11g bug last week. Apparently when you are using the FRA with 11g it decides to archive redo logs to the FRA (as one would expect) AND to the default archivelog destination directory (which one does not expect). I've opened an SR with Oracle on this and we will see how it goes. I've heard a story or two of other bugs that have been discovered, but I've not heard details yet.
So, the moral of the story with 11g is, be careful out there!
I'm finishing up another chapter of my new book, Oracle Database 11g New Features. I'll have some more Oracle features to share with you in the next couple of days, so please standby and be patient.
So, the moral of the story with 11g is, be careful out there!
I'm finishing up another chapter of my new book, Oracle Database 11g New Features. I'll have some more Oracle features to share with you in the next couple of days, so please standby and be patient.
Thursday, August 23, 2007
Landing on Insturments...
So, more Oracle Database 11g in a couple of days. I'm wrapping up another chapter right now and I'll post a bit or two soon.
For this post though, I'd like to share with you that I'm back working on my instrument rating. For me, the best part of flying is cross-country trips. Some people like to get out and just putter around, but I love the experience of actually flying somewhere. Seeing new things, new airports and so on. Because of this, weather is a bit more of an issue for me.
I started my instrument training back in Chicago after I got N7598U. I logged about 15 hours of training or so. Now, I'm back at it! I took my written a few months ago, so thats out of the way. I've just started the flying back up this week and I've logged about 4 hours. Here is one of the approach plates that I use for doing an ILS into Ogden...

I find this stuff pretty cool. This is a precision approach which means it pretty much takes you down to the runway both vertically and horizontally.
I flew this approach and a couple of others on Tuesday, and flew down to Provo on Wed. Saturday I'm back up again and I hope to be done by October if I'm lucky.
More on this later, and more Oracle Database 11g!!!!
For this post though, I'd like to share with you that I'm back working on my instrument rating. For me, the best part of flying is cross-country trips. Some people like to get out and just putter around, but I love the experience of actually flying somewhere. Seeing new things, new airports and so on. Because of this, weather is a bit more of an issue for me.
I started my instrument training back in Chicago after I got N7598U. I logged about 15 hours of training or so. Now, I'm back at it! I took my written a few months ago, so thats out of the way. I've just started the flying back up this week and I've logged about 4 hours. Here is one of the approach plates that I use for doing an ILS into Ogden...

I find this stuff pretty cool. This is a precision approach which means it pretty much takes you down to the runway both vertically and horizontally.
I flew this approach and a couple of others on Tuesday, and flew down to Provo on Wed. Saturday I'm back up again and I hope to be done by October if I'm lucky.
More on this later, and more Oracle Database 11g!!!!
Subscribe to:
Posts (Atom)