07.02.2013 Views

Best Practices for SAP BI using DB2 9 for z/OS - IBM Redbooks

Best Practices for SAP BI using DB2 9 for z/OS - IBM Redbooks

Best Practices for SAP BI using DB2 9 for z/OS - IBM Redbooks

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

This very powerful combination of technologies provides an integrated<br />

environment capable of handling business intelligence today, and far into the<br />

future. This server infrastructure, coupled with DS8300 storage systems, proved<br />

to be a very resilient and highly available configuration over the course of the<br />

project. There were no hardware failures during the six months of this project.<br />

During this project large volumes of customer and benchmark InfoCube data<br />

were loaded into <strong>DB2</strong>. The InfoCube data in <strong>DB2</strong> was then indexed and stored in<br />

the <strong>BI</strong> Accelerator <strong>for</strong> fast query processing. All queries were directed to the <strong>BI</strong><br />

Accelerator since the focus of the project was to test the scalability of the <strong>BI</strong><br />

Accelerator. As with other database servers, some complex ad hoc queries<br />

directed to <strong>DB2</strong> would require creation of aggregates <strong>for</strong> fast query response<br />

time. The beauty of <strong>BI</strong> Accelerator is that aggregates are not required.<br />

The scale of our test environment was several magnitudes larger than the largest<br />

<strong>BI</strong> Accelerator blade configuration size (28 blades) previously tested. We chose<br />

to test three database sizes: 5 terabytes (TB), 15 TB, and 25 TB. Our project<br />

demonstrated that the <strong>SAP</strong> <strong>BI</strong> Accelerator on <strong>IBM</strong> blades can deliver very fast<br />

execution <strong>for</strong> a wide variety of complex queries with consistent response times.<br />

This was achieved <strong>using</strong> a highly compressed <strong>BI</strong> Accelerator index data<br />

structure and efficient parallel in-memory query processing.<br />

11.2 Jupiter environment infrastructure<br />

This section describes the Jupiter environment infrastructure that we built at the<br />

<strong>IBM</strong> System z Benchmark Center at Poughkeepsie, New York. It includes our:<br />

► Hardware and software configuration<br />

► Network infrastructure<br />

By design, some hardware components in this environment were oversized,<br />

because we did not have all the in<strong>for</strong>mation required <strong>for</strong> capacity planning at the<br />

beginning of the project. This large configuration was designed to ensure that<br />

there would be no resource constraints during any of the scalability tests.<br />

Chapter 11. Project Jupiter: large <strong>BI</strong> Accelerator scalability evaluation 229

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

Saved successfully!

Ooh no, something went wrong!