Wednesday, October 24, 2007

Celebration Time....

I never thought that I would write a blog myself ever. To be really honest, I really didn't know what is a blog actually till so long. Still even after reading others blogs for some time, I never could convince myself that I can write too ( and I really am sure about this that my writing is no where to be called good now too) so I was always a little reluctant about having my own blog. Than way back in September 2005, my first blog post on Blogger saw the sunshine. Again the color, theme all on the blog were picked coz of a reason and for a person. Than lot of things, people changed and I was like shaken from head to toe for some time( I am still not stable but trying my best to be) and there was a quietness over this space for nearly half a year till the time when I posted that I am still alive. A lot was going on at that time which was enough to make me distracted and disturbed over lots of things and people. At that time Wordpress also was picking up very fast so my Blogger's blog got a new twin at Wordpress and said hello to the world. As Tim said in one of the comments on Howard's post that writing is not some thing which should be taken as a motive to change this world.  Its you who is writing what you feel. I feel the same. Every one is having its own way to reflect his/her feelings.  Some paint their feelings, some give them a tune and some give their feelings a face by words. I am not at all a good writer but I hope some day I shall improve some what and write some good things  which still may not be able to change the world but will be enjoyed by most.  So even I am late in mentioning ( a month late) , still I guess I can say that its anniversary of this blog. These are my random thoughts reflecting my feelings in my words. I hope if you are reading this than you like it(if not much than a little is okay for me :). I have got some very nice comments so far. I thank all of you who said those nice words for the blog. I just hope you will keep coming back here!


Anonymous Anonymous said...


I'm posting here because I can't post on hjr's site.

There is a question there asking if changing the SGA size can change the plans.

Of course it can. Note this from the performance manual:


Short table scans are scans performed on tables under a certain size threshold. The definition of a small table is the maximum of 2% of the buffer cache and 20, whichever is bigger."

So, you could have small tables that are no longer "small."

Many other things can happen too, such as slower aging of blocks.

October 25, 2007 4:21 AM  
Blogger Aman.... said...

Hi there,
If you would had left your name,I would had been got a chance to say thanks (your name):).But not an issue.Thanks a ton for the wishes.
About the SGA sizing and change of plans,thanks for the input.I am having a small doubt though.I am just taking this as an example.Isnt the manuals trying to say that if the cache is smaller then some tables which are lesser than the 2% of the cache will be called "smaller" tables and we will be able to see in v$*stat views , small scans going on increasing?Isnt it?My point was that ORacle wont pick Index Range SCan over Full table Scan if the cache is smaller or the shared pool is bigger.I can be definitely wrong.Your comment is much awaited and welcome.
Thanks alot for coming and stopping by here and definitely a big thanks for the input about sga sizing.Learning never stops and thats what I learn everyday :).

October 25, 2007 8:38 AM  
Anonymous Anonymous said...

Your welcome!

I'm not sure I'm following your comment, but that's ok.

The point is, the optimizer just follows rules, and decides which rules based on the inputs from statistics and whatever else the developers have seen fit to let it know about. One thing it must know about is whether the table is considered small. It is possible that some tables are near the boundary of "small" and become "small" when the SGA is enlarged. So then the things that influence which set of rules to use have changed. If you were an optimizer and knew you could scan something in memory cheaper than asking for I/O and traversing trees, wouldn't you? Would you rather take a bus or have someone drive you to the airport? The decision changes based on what else you know! :-)

I suggest you google about for Jonathan Lewis' comments on these matters, I'm sure he has posted in several places about it, perhaps on his blog or in his book, but don't have time to search just now.

October 26, 2007 2:21 AM  
Blogger Aman.... said...

Hi there,
I understood your point.Well I shall like to drive from my rooftop if I can rather than travelling all the way to airport first.No doubt about that :).
Your point is based on the size of the caches and based on that stats Optimizer calculates.Please dont think that I want to confront you at any point.I am sure you wont take it like that way.Let me ask you this.Lets us suppose we have a table of about 1million rows having index on one column of it.Now I have a buffer cache of 10m and oracle is picking up a particular plan eg FTS for it.Now I changed my buffer cache to 300m.Will oracle change the plan?

I guess I shall say 2 points.
1)your point is valid and it has a logic behind it.Execution plan "may" change but its not certain will.
2)We should test it out.I shall try and try to post the results.
Thanks a ton for very nice explanation :).

October 26, 2007 8:52 AM  

Post a Comment

<< Home