Dynamic compression of an I/O data block

摘要

In one aspect, a method of compressing data includes splitting an I/O into smaller I/Os based on an I/O throughput. The size of the smaller I/Os are equal or less than a set block size. The method also includes asynchronously compressing the smaller I/Os. In another aspect, an article includes a non-transitory machine-readable medium that stores executable instructions. The instructions cause a machine to split an I/O into smaller I/Os based on an I/O throughput and asynchronously compress the smaller I/Os. The size of the smaller I/Os being equal or less than a set block size. In a further aspect, a system includes circuitry configured to split an I/O into smaller I/Os based on an I/O throughput and asynchronously compress the smaller I/Os. The size of the smaller I/Os being equal or less than a set block size.

主权项

1. A method of compressing data comprising:
obtaining an input/output request (I/O) from an upper layer I/O queue; determine if compression is enabled; putting the I/O in a lower layer I/O queue if compression is not enabled; splitting the I/O into smaller I/Os if compression is enabled, the compression being enabled based on an I/O throughput, the size of the smaller I/Os being equal or less than a set block size; asynchronously compressing the smaller I/Os if compression is enabled; if compression is enabled:
marking the smaller I/Os as requiring reassembly; anddetermining if the I/O is aligned; obtaining the I/O from the beginning of the lower layer I/O queue; determining if the I/O is a write I/O; encrypting the I/O data if the I/O is a write I/O; updating C-metadata if the I/O is a write I/O; and sending the I/O down an I/O stack.