SAS Enterprise BI Server with HP VSE study
This white paper demonstrates how the IT organization at a SAS customer site can benefit from Hewlett-Packard’s Virtual Server Environment (VSE) technology to achieve optimization on server utilization. Through the VSE technology, resources can automatically grow and shrink based on the service-level objectives set for each application. The flexibility offers real time capacity planning based on the workload change to meet the business demands.
The demonstration in this white paper involved the successful integration of HP Integrity servers, HP-UX 11i v2, Virtual Partitions (vPars), HP-UX Workload Manager (WLM), Instant Capacity (iCAP/TiCAP), and the SAS Enterprise BI Server.
The testing effort was carried out in three stages. The first stage involved using separate entry-class HP Integrity rx4640 servers for the SAS Enterprise BI Server implementation. The second stage involved implementing the SAS Enterprise BI solution on a single Integrity rx8620 server with 16 CPUs in a static virtual partition configuration. The transition of the implementation from the first stage to the second implementation can be viewed as the first step to consolidate multiple physical servers to a virtualized server environment. The third stage involved adding WLM and iCAP to the partition configuration in the previous stage and turning it into a dynamic partition configuration with a flexible CPU count.
In the static partition configuration in the second stage, the Integrity rx8620 server with 16 CPUs was configured to 5 virtual partitions with fixed CPU counts. Each partition was installed with one of the SAS Enterprise BI Server components. Dynamic partition configuration was built on top of the static partition configuration with flexible CPU counts in each partition. HP-UX WLM was utilized to dynamically and automatically manage CPU migrations among all virtual partitions based on the workload.
The drawback of the static partition configuration is that the CPU count needs to always size for peak workload. This presents a big challenge for resource utilization because even the most optimal sizing in the static configuration can not predict and match the changes in the dynamic business world. Under an average workload situation, the CPU utilization is usually low around 20-40% according to industry studies. To solve the low CPU resource utilization issue, iCAP CPUs were applied in the dynamic partition configuration. Thus, instead of using all 16 CPUs all the time, iCAP will allow customers to use only the CPU resource required to handle the workload most of the time with an additional CPU resource pool on the side to satisfy the peak workload.
Get pdf SAS Enterprise BI Server with HP VSE study
