Browse Prior Art Database

Dynamically determining the best suited configuration for virtual partitions Disclosure Number: IPCOM000243561D
Publication Date: 2015-Oct-01
Document File: 3 page(s) / 43K

Publishing Venue

The Prior Art Database


This idea proposes a method for dynamically determining the best suited configuration for virtual partitions. The suggested method allows the users to set multiple criticality levels at virtual partition level and allows power, resource and performance management in an efficient way based on the application criticality.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 41% of the total text.

Page 01 of 3

Dynamically determining the best suited configuration for virtual partitions

In a virtualized server environment, there exists multiple virtual partitions on the same server with each partition running different applications. The server might have many different types of hardware resources such as processors, memory, IO adapters, Disk drives etc. Each IO resource will have different bandwidth, speed, and capabilities. It is not always possible to distribute these resources evenly across the virtual partitions. Furthermore, it is not possible to predict the computational and resource requirements of each partition at the time of configuration as the demands of application can change dynamically.

As a result, some partition might get more processors, memory and hardware components with better capabilities than other partitions in the same machine. Sometimes, it is possible that virtual

partitions configured with lesser number of processors and slower IO resource may be running highly critical applications and virtual partitions having fastest hardware may be running less critical applications. Similarly, when the server is required to run at static power save mode, each

partition may be forced to run at certain frequency. Some virtual partitions running highly critical applications may have their cores running at same frequency as other virtual partitions running less critical application.

The current mechanism to deploy a workload on a partition is based on the criticality of the workload and the resources available on a virtual machine. Higher the criticality of a workload, it is deployed on a machine with higher resources.

The problem with this approach is, criticality of workload may not always demand higher computational needs to warrant more resources such as processors or memory. There exists many critical applications which are more I/O intensive and those applications do not need higher processor or memory. Similarly, some less critical applications may be CPU intensive which needs more processors and memory.

Though the system administrators may try their best to understand the application requirements and deploy them on most suited virtual partitions but still application may itself relocate or fail over under changing circumstances. This problem is even more significant in cloud environment where the hardware resources are transparent to deployment of workloads.

So there exists a need for a solution which always ensures efficient distribution of the hardware resources among the logical partitions considering the criticality of the application, type of resource requirements, other workloads running on the machine etc.

The core idea is, define the nature of application running on the partition that is CPU intensive or Memory intensive workload by analyzing its hardware utilization patterns, determine its

power and resource requirements and then use these information to build an Application Profile. Using the application pr...