Sizing specifically for the live part of SAP HANA live

[Former Member]

Hi SCN,

I have been looking around for a while now and I couldn't get any information regarding the sizing for the reporting side of SAP HANA. I am aware that this is quite a specific question, still, here's my bottle in the ocean !

For more context, a customer of mine is implementing FICO on HANA, and they have bought HANA live on top of that. They already had a reporting solution they wanted to keep so we have plugged Cognos on HANA live's views via JDBC. Usage wise, some FICO users, not all of them, will be using the views provided in the live package. Some extra users will be using the reporting even though they don't have any access to ECC. And yes, that comes with licences stuff to do.

I would like to be able to have an idea of the potential impact a heavy use of the database, for reporting purpose, might have on the ECC layer in order to recommend (or not) some extra power for the appliance. The testing we have done so far shows that surpringly enough the bottleneck is the network, and that the appliance is not suffering at all from a few users asking for data, even when the dataset that has to be fetched is large-ish (around 1millions line per users). Figures are as follow : 200 users for FICO, 80 for the reporting (including let's say 60 FICO users) on a 512 Go appliance from IBM (that can be extended to 1To).

Any input is more than welcome for helping figuring out a sizing method for the reporting side alone so that several scenarios (ECC+reporting 1, ECC+reporting2, etc...) can be evaluated.

Many thanks for your help,

Kind regards,

François

[former_member184458 · Accepted solution]

Francois,

all relevant notes and white papers w reagrds to sizing are maintained in this link:

scn.sap.cpm/docs/DOC-59920 as part of the Suite on HANA FAQ.

In General, you can run a sizing script on an existing data set in order to determine the memory size. In an OLAP heavy-use case it is also important to consider the ratio of number of cores to the memory size to provide enough computing power.

thank you

erich

[Former Member · Accepted solution]

This message was moderated.

[former_member184458 · Accepted solution]

Sorry William,

i am not sure I get the issue? You need to perform the ECC sizing to get the table size which provides you the data volume. Then of course you need to perform the data mart sizing as you might run other data in the side car as well.

thanks

erich

[Former Member · Accepted solution]

Hi Erich,

many thanks again for your help as well as my apologies for a little mix-up : I have been inadvertedly using William's account for the previous answer.

Scipting does make sense, and I have would probably have gone with it in the first place.

I had been trying to find in the quick sizer tool some kind of options that would display the extra power needed for reporting only on an already existing applicance - and I will admit this is an odd request.

The link you provided (scn.sap.com/docs/DOC-59920) contains a lot of info, that's great and much apreciated.

Thanks & regards,

François

[Private_Member_9643]

Hi François, we are in process of connecting HANA live contents to cognos for dashboarding purpose. Do you have any document or instructions which will be useful for us?

[Former Member]

Hi Kamaljeet,

not sure I can send a document (customer property) but I can try and help tho.Could you let me know a bit more about the configuration : connectivity, security, etc... ?

On my end, we had a working environnement with Cognos on top of SAP HANA, pulling up Hana live views or modelling done in Cognos, with a JDBC connection. All the data came from an ECC on HANA (SOH) freshly done (ongoing at the time). Security wise we we had to go through a Windows active directory since the users would connect from an IBM portal - this meant at the time that we had to sync several security repo (win ad, ecc, hana, cognos)...At the end of the day, we had a working solution for authentification (the next step was sso with Kerberos), authorisations (a bit of a headache to keep all sync, there are solutions tho) with good performances.

KR

François

[Private_Member_9643]

This is helpful. I needed the confirmation that we can hook HANA views on Cognos.

Do we have any limitation on Cognos side when connecting to HANA? Like we currently have BW system connected to cognos and it has lot of limitation due to BW being OLAP system but assuming JDBC is OLTP based connection so cognos should behave same way for HANA as it behaves for oracle or any traditional DB's?