Comments for Martin Widlake's Yet Another Oracle Blog

Comment on Hey Mum, I’m Famous!!! by khurrampc

khurrampc — Sun, 28 Apr 2013 22:36:35 +0000

Congratz Martin , you really deserve for it.

Comment on Hey Mum, I’m Famous!!! by mwidlake

mwidlake — Sun, 28 Apr 2013 16:28:37 +0000

Thank you Balaji – I hope to be doing more technical blogs soon

Comment on Hey Mum, I’m Famous!!! by Neil Chandler

Neil Chandler — Sun, 28 Apr 2013 14:12:16 +0000

lol. My career has been going downhill since I was 18.

Comment on Hey Mum, I’m Famous!!! by mwidlake

mwidlake — Sun, 28 Apr 2013 13:37:48 +0000

HAHA!!! Indeed, Neil. It’s difficult to comprehend how impressed they will be, exactly… And thus the decline of my career begins

Comment on Hey Mum, I’m Famous!!! by Neil Chandler

Neil Chandler — Sun, 28 Apr 2013 13:02:03 +0000

Well Done Martin. I’m sure your current clients will be very impressed

Comment on Hey Mum, I’m Famous!!! by balajiintech

balajiintech — Sun, 28 Apr 2013 11:08:48 +0000

To me. Ur already famous sir. Your My guru on IOTs and many more topics.

Congrats sir.

Cheers, Balaji Tr.

Sent from iPhone

Comment on The Three Tenners- OUGN 2013 by David Harper

David Harper — Sun, 28 Apr 2013 05:40:08 +0000

“If only the idiot on the right had held his the same way around as the other two…”

Nonsense, you’re holding it the right way round — Charles Darwin proudly to the front, which is the correct way for a scientist to display a tenner in public!

Comment on Index Organized Tables – the Basics. by khurrampc

khurrampc — Sat, 27 Apr 2013 22:25:24 +0000

Thanks Martin for yours clarification , regarding the IOT block consumption and heap table, it make sense that IOT has got further column other then Keys columns which cause more blocks in IOT , i have doubled check it by creating heap table only with PK columns and IOT only with PK columns which cause lesser block in IOT to heap.Thanks

regarding the second question i am still not clear from this statement of yours “Primary Key Issues” like “Partitioning is important in this technique” , what technique you are talking about? you mean to say whenver need to create unique key somehow go for IOT partition?

Khurram

Comment on Index Organized Tables – the Basics. by mwidlake

mwidlake — Fri, 26 Apr 2013 14:30:19 +0000

An IOT is in effect a unique index {the Primary Key is a special type of unique index} with the rest of the table data stored with the index value. If you partition a table, any unique keys must contain the partition key to be local indexes, ie to be partitioned in line with the table partitions. {This is because it would be a massive overhead for oracle to have to check all partitions to ensure uniqueness of a new record and that would be the only way it could be done if the partition key was not provided to identify the correct single partition} And if you are partitioning an IOT then you are by definition partitioning that unique index. So it has to contain the partition key.

Comment on Index Organized Tables – the Basics. by mwidlake

mwidlake — Fri, 26 Apr 2013 14:20:40 +0000

Hi Khurram,

The answer is that the PK for the heap table only holds the primary key columns. The IOT index holds the rest of the columns also, so it takes up more space. In the example in post 2 the IOT is in fact larger than the traditional heap table AND Primary Key. I now realise I did not come back to this point as I intended, so maybe I need to do a new post.
I am not exactly sure why the size of just the freshly created IOT is greater than the Heap-table-and-Primary-Key {and it not always the case} but if you SHRINK the IOT table it ends up smaller than the heap-table-and-Primary-key. My belief is that as the data is created, the block splits to populate the IOT cause the blocks to be partially empty, anything up to 50%. This will also be true of the PK of the heap table, but as the PK is smaller, the impact is less.The heap table should be very efficiently packed.