User's Manual

Data compression
45
Hardware Compression
If data compression is used in software on the host computer rather than in the
hardware of the drive, you can slow down the transfer rate of the host because it
must perform compression computations in addition to its regular computations.
Also, any other host that wants to retrieve (decompress) the data must have the
same software.
Hardware data compression (HDC) refers to the implementation of the DCLZ
algorithm in the data compression engine, with the compression processing activity
transparent to the host computer and the user.
Seagate’s data compression engine is designed to provide a complete data
compression system using the DCLZ algorithm. This IC provides support circuitry as
well as the core DCLZ compression machine.
A more detailed description of the data compression engine is given later in this
chapter.
Data Integrity
There are various types of data-compression algorithms, but in this document they
are divided into two basic types:
lossless
algorithms, such as DCLZ or ALDC, and
lossy
algorithms, such as those used in some consumer audio products.
Lossy algorithms drop out or lose some portion of repetitious data during the
compression process to reduce the actual data bytes that are recorded to tape. The
data lost during this process is lost forever and cannot be recovered. In consumer
audio, this is not a problem because this method reduces required storage space
and still provides better-than-analog recording and playback quality.
As you would expect, lossy algorithms are inappropriate for computer data storage of
any type; hence the choice of lossless algorithms for computer data storage use.
Lossless algorithms are designed to compress data using a complex algorithm,
ensuring that all data is compressed and recorded to tape and that all data can be
decompressed and returned in the identical format as before. No bits are lost, and no
data is compromised.
The DDS standards specify the use of the DCLZ algorithm, a lossless algorithm for
data compression.