Row and Column store hybrid tables

[Former Member]

Dear all,

is it possible to have a hybrid table, where a subset of fields is column based whereas the other subset is row based. At the moment you can only set it globally per table. This would make sense for line-item tables where a lot of write-to-db cycles are being executed. In this situation you can set the "identifyer fields" as column based and fields like amount and text fields (since they are most diverted) as row based.

Best regards

Mark

[lbreddemann]

Ning already answered the question if it is currently possible (no), I'd like to comment on the assumptions you made.

Line item data is not necessarily more diverse than the header data. Depending on the set of possible services/product items and potential business partners, the ratio between both can be very different from business to business.

In either case, even the line item information is not that often updated, e.g. the line item entries are not updated with status or timestamp information every few seconds.

On the other hand, the line items are crucial for many reporting and analytics scenarios. This is where you want the ability of massive parallel aggregation, scale out and partitioning.

These features are only available for column store tables.

For the ID fields in turn: these are the type of columns that benefit the least from a column store. They cannot be compressed well and usually are not part of analytic scenarios

[former_member182967]

Hi Mark,

As per note 2222277 - FAQ: SAP HANA Column Store and Row Store

4. Can I assign a table to both column store and row store?

No, at any time a table can only be assigned to one of the stores, either column or row store.

Regards,

Ning