cancel
Showing results for 
Search instead for 
Did you mean: 

Sizing in PI

Former Member
0 Kudos

Hi

I am trying to do a SAP sizing requirement for PI. I have 12,15,277 messages perhour. The senior consultant told me we need 32 CPU's for this. I saw in documents that only two lakh messages can be processed in an hour with 32 CPU. My peak hour requirement is more than that.

What are the SAPS, RAM and GB required for me if we are using 32 CPUS? I was asked to calculate these three factors for 32 CPUs. can you provide me inputs? Is there a connection between CPUs and RAM?

Regards

Accepted Solutions (0)

Answers (4)

Answers (4)

Former Member
0 Kudos

HI,

Some recomendation you can use.

You may have a look at the sizing document.

http://scn.sap.com/docs/DOC-16231

Also High Availabily should be set up for multiple cluster node scenario.

http://scn.sap.com/people/benny.schaich-lebek/blog/2009/11/23/high-availability-with-sap

Creating Integrated Configurations.

Thanks,

Sudhansu

rajasekhar_reddy14
Active Contributor
0 Kudos

Hi Madhav,

1)Check size of the messages are going to process per day/ hours, if your project in intial stage then talk with bussiness team (client) to get the appropriate details.

2)Check what kind of adapters are going to involve in integration, SAP to SAP integration works smooth way when you deal with high volume data but integration with third party / FT\P/JDBC/SOAP  required more cluster nodes to process  12,15,277 messages.

3) 12,15,277 is not a small number so you need to come up with innovation/good design to minimize the number of messages in to one messages (pakcaging concept).

4)Use quick sizer tool to calculate Hard ware detials and check with SAP.

Regards,

Raj

iaki_vila
Active Contributor
0 Kudos

Hi Midhum,

You should read carefully the SAP recommendations for the hardware requeriments and to use its sizing decision tool. The documentation is in the SAP service Marketplace in this link https://websmp205.sap-ag.de/sizing (For SAP PI 7.1)

Regards.

MichalKrawczyk
Active Contributor
0 Kudos

Hi,

1. you cannot determine this on the basis of number of messages only

a) if depends if you use both stacks or not

b) you can easily group messages and process a few milions via PI easily within an hour

(grouped nicely)

2. are you using PI sizing guide to estimate it ? if not try using the pi sizing guide

Regards,

Michal Krawczyk