Overall application performance, CPU and I/O utilization and the size of disk files are good indicators of how effective compression is for your application. This section builds on the performance tuning advice from Section 17.9.1.3, “Tuning Compression for InnoDB Tables”, and shows how to find problems that might not turn up during initial testing.

To dig deeper into performance considerations for compressed tables, you can monitor compression performance at runtime using the Information Schema tables described in Example 17.1, “Using the Compression Information Schema Tables”. These tables reflect the internal use of memory and the rates of compression used overall.

The INNODB_CMP table reports information about compression activity for each compressed page size (KEY_BLOCK_SIZE) in use. The information in these tables is system-wide: it summarizes the compression statistics across all compressed tables in your database. You can use this data to help decide whether or not to compress a table by examining these tables when no other compressed tables are being accessed. It involves relatively low overhead on the server, so you might query it periodically on a production server to check the overall efficiency of the compression feature.

The INNODB_CMP_PER_INDEX table reports information about compression activity for individual tables and indexes. This information is more targeted and more useful for evaluating compression efficiency and diagnosing performance issues one table or index at a time. (Because that each InnoDB table is represented as a clustered index, MySQL does not make a big distinction between tables and indexes in this context.) The INNODB_CMP_PER_INDEX table does involve substantial overhead, so it is more suitable for development servers, where you can compare the effects of different workloads, data, and compression settings in isolation. To guard against imposing this monitoring overhead by accident, you must enable the innodb_cmp_per_index_enabled configuration option before you can query the INNODB_CMP_PER_INDEX table.

The key statistics to consider are the number of, and amount of time spent performing, compression and uncompression operations. Since MySQL splits B-tree nodes when they are too full to contain the compressed data following a modification, compare the number of “successful” compression operations with the number of such operations overall. Based on the information in the INNODB_CMP and INNODB_CMP_PER_INDEX tables and overall application performance and hardware resource utilization, you might make changes in your hardware configuration, adjust the size of the buffer pool, choose a different page size, or select a different set of tables to compress.

If the amount of CPU time required for compressing and uncompressing is high, changing to faster or multi-core CPUs can help improve performance with the same data, application workload and set of compressed tables. Increasing the size of the buffer pool might also help performance, so that more uncompressed pages can stay in memory, reducing the need to uncompress pages that exist in memory only in compressed form.

A large number of compression operations overall (compared to the number of INSERT, UPDATE and DELETE operations in your application and the size of the database) could indicate that some of your compressed tables are being updated too heavily for effective compression. If so, choose a larger page size, or be more selective about which tables you compress.

If the number of “successful” compression operations (COMPRESS_OPS_OK) is a high percentage of the total number of compression operations (COMPRESS_OPS), then the system is likely performing well. If the ratio is low, then MySQL is reorganizing, recompressing, and splitting B-tree nodes more often than is desirable. In this case, avoid compressing some tables, or increase KEY_BLOCK_SIZE for some of the compressed tables. You might turn off compression for tables that cause the number of “compression failures” in your application to be more than 1% or 2% of the total. (Such a failure ratio might be acceptable during a temporary operation such as a data load).