Disk performance for random access fares significantly worse compared to sequential access. Time required to transfer random blocks to or from disk is dominated by seeking and rotational delay. To improve the throughput and reduce the latency, one can apply techniques to increase the sequentiality of disk accesses, such as block rearrangement and replication. We introduce an approach to improve read performance by replicating blocks into file system free space at the block level. This makes the replication module independent of the file system and therefore easier to implement and verify. A solution that requires no changes to the file system is also easier to adopt. Supporting a new file system is a matter of writing a user-space component that understands its free block data structures. We implemented a prototype as a stacked device driver for Linux and evaluated its performance on a number of workloads.

Disk performance for random access fares significantly worse compared to sequential access. Time required to transfer random blocks to or from disk is dominated by seeking and rotational delay. To improve the throughput and reduce the latency, one can apply techniques to increase the sequentiality of disk accesses, such as block rearrangement and replication. We introduce an approach to improve read performance by replicating blocks into file system free space at the block level. This makes the replication module independent of the file system and therefore easier to implement and verify. A solution that requires no changes to the file system is also easier to adopt. Supporting a new file system is a matter of writing a user-space component that understands its free block data structures. We implemented a prototype as a stacked device driver for Linux and evaluated its performance on a number of workloads.

Disk performance for random access fares significantly worse compared to sequential access. Time required to transfer random blocks to or from disk is dominated by seeking and rotational delay. To improve the throughput and reduce the latency, one can apply techniques to increase the sequentiality of disk accesses, such as block rearrangement and replication. We introduce an approach to improve read performance by replicating blocks into file system free space at the block level. This makes the replication module independent of the file system and therefore easier to implement and verify. A solution that requires no changes to the file system is also easier to adopt. Supporting a new file system is a matter of writing a user-space component that understands its free block data structures. We implemented a prototype as a stacked device driver for Linux and evaluated its performance on a number of workloads. / Science, Faculty of / Computer Science, Department of / Graduate

Disk drives are the bottleneck in the processing of large amounts of data used in almost all common applications. File systems attempt to reduce this by storing data sequentially on the disk drives, thereby reducing the access latencies. Although this strategy is useful when data is retrieved sequentially, the access patterns in real world workloads is not necessarily sequential and this mismatch results in storage I/O performance degradation. This thesis demonstrates that one way to improve the storage performance is to reorganize data on disk drives in the same way in which it is mostly accessed. We identify two classes of accesses: static, where access patterns do not change over the lifetime of the data and dynamic, where access patterns frequently change over short durations of time, and propose, implement and evaluate layout strategies for each of these. Our strategies are implemented in a way that they can be seamlessly integrated or removed from the system as desired. We evaluate our layout strategies for static policies using tree-structured XML data where accesses to the storage device are mostly of two kinds - parent-tochild or child-to-sibling. Our results show that for a specific class of deep-focused queries, the existing file system layout policy performs better by 5-54X. For the non-deep-focused queries, our native layout mechanism shows an improvement of 3-127X. To improve performance of the dynamic access patterns, we implement a self-optimizing storage system that performs rearranges popular block accesses on a dedicated partition based on the observed workload characteristics. Our evaluation shows an improvement of over 80% in the disk busy times over a range of workloads. These results show that applying the knowledge of data access patterns for allocation decisions can substantially improve the I/O performance.