Home > Storage > PowerScale (Isilon) > Product Documentation > Management and Migration > Storage Tiering with Dell PowerScale SmartPools > Example D: Metadata performance for seismic interpretation
The seismic data required for energy exploration is typically vast, and growing, with many organizations keeping petabytes of data online for faster analysis and more accurate drilling predictions. This data might be arranged in tens or hundreds of thousands of files across thousands of directories. Often less than 30 percent of those files might be used in any given month. However, administrators constantly must locate files for pre- and post-stack analysis, interpretation, and project planning. The latencies involved with traversing enormous directory structures and file quantities, listing directory contents, and other metadata operations, are often measured in minutes. This latency is clearly unacceptable for operations that are repeated multiple times per day. Seismic data is the core of exploration and is, therefore, critical, and while for rarely accessed files the data might not be timely, the metadata is. In summary, seismic data is critical. Its metadata is timely. Its data might be timely or not timely. In this case, metadata read acceleration, as opposed to L3 cache, is the preferred approach.
A mirror of all metadata is stored on SSDs, but data might be relegated to faster or slower disk based on its relative timeliness. The result is the ability to instantly locate any data, while being able to still cost-reduce most of the data on the system. Because metadata is accelerated, many other activities, including backups, migration, and replication, realize performance benefits as well.