MDC use case

[Former Member]

Hello,

We are using our HANA machine for 3 purposes:

1. Loading data\ETL with BW powered by HANA

2. HANA modeling and BO reports

We don’t want that:

1. BO users will be influenced by the BW loads

2. BW loads will get slower (or get OOM)  because BO query

3. SLT loads (or HANA delta merge) will influence BW or BO

Does it make sense to use Multitenant Database Containers (MDC) for these proposes.

How flexible is it to move memory between the tenants?

Is it possible to control it by time – meaning at night to have more memory for the BW and in the morning to have more memory for BO reports?

Thanks,

Amir

[lbreddemann]

Looks like you want to eat the cake and have it too.

Flexible resource assignment between single the MDC database and total isolation from workload effects...

What you can do with SPS 10 (haven't checked SPS 11 so far) is to define memory allocation and cpu core usage limitations for each DB. This isolates the workload of the single DBs and you shouldn't see any OOMs or slowed down queries because of whatever happens in another DB.

That said, the machine has to be (of course) sized appropriately to support multiple DBs at once. The current sizing guideline is to add up the hardware requirements for each DB to yield the overall sizing.

In case you want to change the resource limitation, you will have to manually change the configuration of one or more of the DBs - there is (currently) no automatic 'make-use-of-all-resources-if-any-are-available-but-guarantee-at-least-X-CPU/memory' function.