Wednesday, January 2, 2013

Teradata 13.10 new features (part 3)


Td13.10 not only enhanced existing Multi valued compression but also introduced two new methods

3. Algorithmic compression
Users can now write their own compression and decompression algorithms for characters. 
These will have to be implemented in UDF and needs to be tested thoroughly before implementing it in full scale.
There are some samples functions Teradata has provide and can be found in SYSLIB.
TransUnicodeToUTF8, TransUTF8ToUnicode, LZCOMP, LZDECOMP, LZCOMP_L, LZDECOMP_L, CAMSET, DECAMSET, CAMSET_L,DECAMSET_L.



4. Block level compression (applicable to DBA’s, Capacity planners)
BLC is applied at data blocks and allows row data to be stored in datablocks* in a new compressed format.
To enable BLC on tables, new command "COMPRESS" is available in FERRET utility. After BLC is enabled on table, all tows added to table are compressed automatically. Similarly "UNCOMPRESS" command can be used to uncompress packed Data.

BLC is one of most straight forward methods where one doesn’t have to worry about type of data across various columns in table. Row header and internal Teradata structures are also compressed using BLC.

BLC plays major role in saving lot of I/O and Space but at same time it can have some performance implications as more CPU can be utilized to compress and uncompress data dynamically. 

datablocks*
(Data blocks contains data rows .Block size can vary between 64kb to 128kb in latest version)

To be Continued...

No comments:

Post a Comment