Larry Baddock wrote:

>>I'm losing my enthusiasm for compressed blobs. I'm not convinced the
>>big win is there.
>>
>>
>I agree with you here.
>
>Simple analysis of the comparisons for compression speeds, etc for different
>algorithms found from google seem to indicate that compressing on reasonably
>modern machines is around the 1-3 MB/sec mark. On tests that I have run with
>a large disk seek component on a standard serial ATA disk reveal that the
>disk is capable of about 4MB/s sustained transfer rate. (Of course, linear
>reads off the disk are round about the 55MB/s mark, but that doesn't tell
>you anything useful for real life usage patterns).
>
>
>

For blob compression to be worth implementing, the decompression cost
should be reasonable close to the CPU cycles to read the number of
blocks equivalent to the pages not written. The cost of kicking off an
I/O operation is reasonably high -- cross the protection ring, probe
addresses, queue the operation, put the process/thread to sleep, thread
or context switch, service an interrupt, mark the thread runnable, and
context or thread switch back. Ifthe CPU time is equivalent to the
elapsed time, (as appears to be the case), we're dead.

I think the answer is either more appropriate compression schemes or
forgoing blob compression.

--

Jim Starkey
Netfrastructure, Inc.
978 526-1376