06.01.2013 Views

Download PDF - IBM Redbooks

Download PDF - IBM Redbooks

Download PDF - IBM Redbooks

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

Table B-3 Recovery objective<br />

Type of outage Recovery time<br />

objective (downtime)<br />

Component failure (hardware, software,<br />

and network<br />

2 business hours 30 minutes<br />

Total outage is more than 1 day 48 hours 8 hours<br />

Recovery period<br />

objective (data at risk)<br />

Solution<br />

By applying Quick Sizer results and the sizing estimate from <strong>IBM</strong> Techline, we determined<br />

that we need a z196-702 with two central processors (CPs), one zIIP for the database server<br />

load, and 11 IFLs for the SAP NetWeaver AS load.<br />

Additional information: You can find the <strong>IBM</strong> Sizing and Planning Questionnaire for SAP<br />

Solutions at:<br />

http://www.ibm.com/support/techdocs/atsmastr.nsf/PubAllNum/PRS261<br />

For more information about using the SAP Quick Sizer, see “Questions for users of the<br />

SAP QuickSizer” at:<br />

http://www.ibm.com/support/techdocs/atsmastr.nsf/WebIndex/PRS2294<br />

To size the LPARs, give the production instances more memory than the minimum<br />

recommended by the Quick Sizer, because clients have experienced that memory proposed<br />

by Quick Sizer is conservative. However, development and QA instances can use the Quick<br />

Sizer recommendations.<br />

Because this system is new, we cannot estimate how much the processor is being used<br />

during production. Also the development systems might be more active during the SAP<br />

application customization and introduction phase. Therefore, we decided to share the<br />

processors between the production, development, and quality assurance instances for now.<br />

We might have to adjust the LPAR resources after the system workload settles, which is a<br />

good practice, because every client will have an individual SAP workload pattern. In<br />

particular, if SAP transactions or jobs were created or changed by LOB consultants, they<br />

often affect workload performance.<br />

For system recovery, we implement SCS in the z/OS LPAR to decouple it from the application<br />

server host. Also, this approach prepares the system if the client implements Parallel Sysplex<br />

data sharing and HA later. For data loss protection, use Metro Mirror to back up the DB2 log<br />

and data and to enable storage FlashCopy to use the DB2 BACKUP/RESTORE DATABASE utility,<br />

point-in-time recovery, and federated synchpointing of the applications.<br />

We designed four hosting LPARs: two for the database servers and two for the application<br />

servers. One pair of database servers and application servers is dedicated to production. The<br />

other pair hosts the development and QA instances. The database server partitions run each<br />

DB2 subsystem and one database instance per SAP application instance. Also one SCS<br />

instance is available per SAP database server instance in these LPARs. The application<br />

server LPARS are managed by z/VM as the host operating system. We have two application<br />

servers per production application, and one application per development and per QA<br />

application and instance. The exception is the Solution Manager, which has one application<br />

server central instance for production and one for QA and development, with no separate<br />

SCS servers and no separate QA or development instances. Also the SAProuter is available<br />

once per productions and QA or development.<br />

Appendix B. Planning, preparing, and implementing the SAP landscape 159

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!