Decompression

Compression saves storage and reduces I/O operations, but the data still needs to be decompressed before it can be used in result sets.

Although all compression is performed by the CPUs in the database nodes, EHCC data can be decompressed by the CPUs in the Exadata Storage Server or those in the database nodes. The choice is made based on the type of query processing performed. A Smart Scan query returns result sets directly to the PGA, so the CPUs on the Exadata Storage Server perform the decompression. Queries that cannot use Smart Scan return standard data blocks to the database node—if these blocks are compressed with EHCC, the database server CPUs must decompress them, which requires cycles and memory space to store the decompressed data.