Guidelines for Configuring Virtual Partitions on Cellular Platforms

The benchmark produces two metrics, the maximum number of SD users a system can

handle (while keeping average response time below two seconds), and SAPS. The SAPS

(SAP Application Performance Standard) is a metric used to measure the throughput of a

SAP application server. For the SAP SD benchmark, 100 SAPS equal 2000 fully processed

order line items per hour. A higher SAPS rating indicates better system performance and

thus an ability to accommodate more users.

For this experiment, rather than measuring the maximum users for a particular configuration,

we choose to keep the number of users constant for all configurations while measuring the

SAPS. The faster the transactions complete, the higher the SAPS are. So, for the same

number of users the SAPS number is higher for an optimal configuration than for less optimal

configurations. We choose the number of users for this experiment in such a way that the

system is under-utilized while in a similarly sized nPartition configuration

. This leaves

enough room for less optimal vPars configurations. We use SAPS as a comparison

measurement.

SPECjbb

SPECjbb is a standard benchmark developed by SPEC [7]. It evaluates the performance of

servers running typical Java-based business applications.

SPECjbb models a wholesale supplier with a variable number of warehouses and represents

an order processing application for the supplier. It emulates a three-tier client/server system,

with middle-tier business logic connecting front-end thin clients to a back-end data store. The

emphasis of the measurement is on the middle tier, the business logic, and object

manipulation. The benchmark combines all three tiers into a single Java application for

portability and ease of use.

The study in this paper compares results from a single cell nPartition configuration with results form different vPars configurations all having the

same amount of resources as a single cell nPartition