Skip to Content
0
Sep 14, 2019 at 01:15 PM

SAP HANA: Are Read Ops on Delta Storage Optimized?

242 Views Last edit Sep 14, 2019 at 01:16 PM 2 rev

Hello,

I'm reading through some documentation on HANA and am a bit confused as to what format the data has when it's inside the Delta Storage. I understand that to optimize write ops, the delta storage must have a row-based table. These row-based records are then converted to column-based format in order to be stored in the Main Storage when a Delta Merge is performed. Now the question is, what happens when you have to read data from the row-based Delta Storage, which in the general case is more expensive?! Do you just bite the bullet, and lose a bit of performance by reading a row-based table? If not,then do you have structures like L1 and L2, where L2 will be an uncompressed, column-based representation of L1 which will then be used by the read op instead of a row-based table?

Thanks,

Dimitar