cancel
Showing results for 
Search instead for 
Did you mean: 

Data Partitioning Strategy

0 Kudos

Hi

I have scenario here , i have 3 tables and i am anticipating they are going to grow in large size possibly over  2 billion records in future.

The application using these tables is yet to be deployed , i am looking here strategy for partitioning the table.

Do i create partition for these tables after the application is  deployed for production and seeing the growth rate of table data ( monitoring some threshold limit of no of records which i am not aware of now) or do i design partitions for  tables in schema before i deploy this into production?

If i have to partition can i partition on Range partitioning on timestamp column, ? on month or on year vale of timestamp?

many thanks

Accepted Solutions (1)

Accepted Solutions (1)

Former Member
0 Kudos

Hi Aniruddha

You can partition on date range. See the link below

Using Date Functions to Partition - SAP HANA Administration Guide - SAP Library

As to when you create the partition, I would think it will be far easier to create the partitions now rather than wait for the tables to fill up

lbreddemann
Active Contributor
0 Kudos

The later is not necessary a good advise as the sheer existence of additional partitions (used or not) impacts resource management and query execution.

Generally speaking, table partitioning should be seen as an ongoing, dynamic activity and not as a once off decision.

0 Kudos

thanks Lars for your suggestion,

I have One more query .

say i have  table with date column and i want to partition the data on this column based on month, if i query this table on non-partition column then is the performance going to be slower in comparison to query on partition  column table

thanks in advance

0 Kudos

thanks Howard for your advise.

lbreddemann
Active Contributor
0 Kudos

As usual: that depends.

But, very likely, the query on the partitioned table without partition pruning, will end up using more system resources. In effect you're querying as many table data structures as you have partitions.

0 Kudos

thanks Lars for your valuable guidnace

Answers (0)