Cloud services provision optimization

Client: A Multinational firm

The Parallel Architecture Group of the Universitat Politècnica de València (GAP) worked with a multinational firm wanting to improve the provision of its services through the cloud.

The unpredictable nature of the time needed to provide services via the cloud is one of the elements that worsens the quality of service offered to customers. The most immediate response to address the problem is to oversize the system. Nevertheless, this solution is not cost-effective, as it increases the activity’s costs.

To respond to this challenge, the GAP Group has developed systems software applying state-of-the-art technologies in the different parts of the system (processor, last level cache, memory bandwidth and others), such as CAT (Cache Allocation Technology) to reduce interference between applications running on the last level cache.

The study consisted on identification of the problem and the reduction of the interference, which allows to increase the utilization of the system components. As result the system can support more requests per second for a given quality of service. Therefore, it is possible to offer the service to more customers without increasing the cost derived from the incorporation of higher capacities.