Influence of adaptive data layouts on performance in dynamically changing storage environments

For most of today's IT environments, the tremendous need for storage capacity in combination with a required minimum I/O performance has become highly critical. In dynamically growing environments, a storage management solution's underlying data distribution scheme has great impact to the overall system I/O performance. The evaluation of a number of open system storage visualization solutions and volume managers has shown that all of them lack the ability to automatically adapt to changing access patterns and storage infrastructures; many of them require an error prone manual re-layout of the data blocks, or rely on a very time consuming re-striping of all available data. This paper evaluates the performance of conventional data distribution approaches compared to the adaptive virtualization solution V:DRIVE in dynamically changing storage environments. Changes of the storage infrastructure are normally not considered in benchmark results, but can have a significant impact on storage performance. Using synthetic benchmarks, V:DRIVE is compared in such changing environments with the non-adaptive Linux logical volume manager (LVM). The performance results of our tests clearly outline the necessity of adaptive data distribution schemes.