Enerprise AIX/System P Virtualization
Business Challenge:
In the 1990s and early 2000s a large regional financial institution purchased and deployed UNIX servers using a traditional method based on vendor tools, benchmarks, or on references and experiences of other companies. The servers were sized for peak usage and typically sized not to exceed 65% CPU utilization in order to cover any unexpected spikes in usage and ensure predictable performance. As a result this company, like many others, experienced under-utilization of many of their servers. At the same time in other instances, they have found that they had severe performance problems when server CPU utilization topped at 100%. Consequently they would have to upgrade or replace these servers to handle a short peak during a month or suffer performance problems.
Solution presented:
In an executive meeting several years ago, Texcel Inc partnered with IBM and discussed these issues with the top level management. The consensus was that in order for this company to save money on recurring hardware purchases to meet load demands, they need to better utilize their resources. With better CPU and server utilization using the AIX/System p Virtualization technology from IBM they would realize that the need to purchase additional hardware for every project was no longer a neccesity. The smaller the infrastructure, the smaller the cost of supporting these projects including maintenance and depreciation of the project. As a result, implementing virtualization would create signifcant front-end and back-end savings.This solution was further elaborated upon using various technical training sessions and demos. The existing IT staff were invited to these trainings and demo sessions to make them comfortable with this technology.
Business Results:
Working together with this company's Engineering team and partnering with IBM resources, Texcel, Inc was instrumental in implementing technologies that made this virtualized environment more robust, including LPARs, MicroPartioning, High Availabilty, Capacity on Demand and Live Partition Mobility. This company evolved from the traditional sizing and server architecture of multiple, under-utilized servers to fewer, highly virtualized and thus highly utilized servers.