Articles & Content

Archives

Need to Cut Costs? Time to Go Parallel on a zIIP? (A12)

Topic: DB2 for z/OS

Subtopic: 2009



DATE: 2009-10-7 (14:15 - 15:15)
SPEAKERS: Adrian Collett (Expertise4IT s.r.l.)

zIIP processors have provided great benefit for distributed processing; however, most people seem to overlook the fact that the biggest beneficiary of zIIP offload is actually parallel processing.So, with cost-cutting all the rage in today’s financial crisis, is there a case for turning on parallelism in our main non-BI production systems to benefit from zIIP offload? And if there is, how can we identify the processes and queries where parallel processing will benefit without creating overhead and how much offload can we expect?This presentation examines 3 different non-BI environments where we have tried to increase zIIP usage by introducing Parallel Processing. Not only will it provide real figures on the actual zIIP offload obtained for parallel processing in non-BI environments, but it will also examine some of the difficulties involved in identifying candidate workloads and provide some real-life horror stories of what happens when parallelism is used inappropriately!

EXP. LEVEL: Beginner,Intermediate

OBJECTIVES:

Brief overview of the basics of zIIP processors and the types of workload that can exploit them together with real figures on the actual zIIP offload obtained in three different environments(V8).

Understand how parallel processing can exploit the zIIP processor by providing examples of both complex and non-complex queries in a non-BI(Data Warehouse) production environment(V8) and highlighting the impact of the various thresholds involved.

Understand how to identify candidate queries and workloads that will benefit from parallel processing and zIIP offload, highlighting the difficulties involved and the importance of partitioning and the choice of an appropriate partitioning criteria.

Understand the overheads and pitfalls involved in activating parallel processing in non-BI(Data Warehouse) production environments, especially when used for an inappropriate workload.

Provide real figures on zIIP offload actually obtained due to the use of parallel processing in three separate Version 8 non-BI(Data Warehouse) production environments together with some hopes and wishes for Version 9.



Click Here to Download

NOTE: These are only open to members of IDUG. If you are not a member, please CLICK HERE for more information.