Quantum FIREBALL PLUS AS 10.2 Product Manual page 56

Basic Principles of Operation
DisCache works by continuing to fill its cache memory with adjacent data after
transferring data requested by the host. Unlike a noncaching controller, Quantum's
disk controller continues a read operation after the requested data has been
transferred to the host system. This read operation terminates after a programmed
amount of subsequent data has been read into the cache segment.
The cache memory consists of a 1.9 MB DRAM buffer allocated to hold the data,
which can be directly accessed by the host by means of the READ and WRITE
commands. The memory functions as a group of segments with rollover points at
the end of cache memory. The unit of data stored is the logical block (that is, a
multiple of the 512 byte sector). Therefore, all accesses to the cache memory must
be in multiples of the sector size. Almost all non-read/write commands force
emptying of the cache:
5.3.1.3 5.3.1.3 Write Cache Write Cache
5.3.1.3 5.3.1.3 Write Cache Write Cache
When a write command is executed with write caching enabled, the drive stores the
data to be written in a DRAM cache buffer, and immediately sends a GOOD
STATUS message to the host before the data is actually written to the disk. The
host is then free to move on to other tasks, such as preparing data for the next data
transfer, without having to wait for the drive to seek to the appropriate track, or
rotate to the specified sector.
While the host is preparing data for the next transfer, the drive immediately writes
the cached data to the disk.
WriteCache allows data to be transferred in a continuous flow to the drive, rather
than as individual blocks of data separated by disk access delays. This is achieved
by taking advantage of the ability to write blocks of data sequentially on a disk that
is formatted with a 1:1 interleave. This means that as the last byte of data is
transferred out of the write cache and the head passes over the next sector of the
disk, the first byte of the of the next block of data is ready to be transferred, thus
there is no interruption or delay in the data transfer process.
The WriteCache algorithm fills the cache buffer with new data from the host while
simultaneously transferring data to the disk that the host previously stored in the
cache.
5.3.1.4 5.3.1.4 5.3.1.4 5.3.1.4 Performance Benefits Performance Benefits Performance Benefits Performance Benefits
In a drive without DisCache, there is a delay during sequential reads because of the
rotational latency, even if the disk actuator already is positioned at the desired
cylinder. DisCache eliminates this rotational latency time (4.17 ms on average)
when requested data resides in the cache.
Moreover, the disk must often service requests from multiple processes in a
multitasking or multiuser environment. In these instances, while each process
might request data sequentially, the disk drive must share time among all these
processes. In most disk drives, the heads must move from one location to another.
With DisCache, even if another process interrupts, the drive continues to access the
data sequentially from its high-speed memory. In handling multiple processes,
DisCache achieves its most impressive performance gains, saving both seek and
latency time when desired data resides in the cache.
The cache can be flexibly divided into several segments under program control.
Each segment contains one cache entry. A cache entry consists of the requested
read data plus its corresponding prefetch data.
5-11 Quantum Fireball Plus AS 10.2/20.5/30.0/40.0/60.0 GB AT