Do you have a question? Post it now! No Registration Necessary. Now with pictures!
- Claude Hopper
December 19, 2008, 3:15 pm
rate this thread
Re: Hard drives
The thruput can vary greatly. I have one 7200RPM drive in my computer
that gives 60MB/sec. And a new one I just bought, that benchmarks
at 90MB/sec. So even at a constant speed (7200RPM), there can be
Drives come in speeds of 5400, 7200, 10000, 15000 RPM and so on.
Part of the time finding the data, includes on average, half the rotational
latency. So that component of the time is reduced, when you buy a higher RPM
When comparing transfer rates, you have to compare drives from the
same generation. Comparing a modern 7200 with an old 15K drive,
is going to make the 15K drive look like a loser. A fair comparison,
is when they're using similar technologies.
Caches on drives are now available up to 32MB. What we don't know
(what is unstated), is how that cache memory is managed. On the old
SCSI drives, you had the option to "tune" the cache management,
for either desktop or server usage. In a server situation, there
may be support for multiple threads, meaning the cache is
chopped up into chunks (my old SCSI drive supported 8 independent
transactions). In a desktop scenario, which is closer
to a single thread, then chopping the cache up into small chunks
doesn't make sense. I'm unaware of any tool that will tell you
anything about IDE cache, and as a result, I have to ignore the
cache spec when I buy a hard drive. If it is 32MB, then I want
proof it is being used to help me, and I see no such proof.
On IBM drives, part of the cache is stolen to hold a relocation
table. Perhaps other brands do something similar. You may find,
for example, that 15% of the cache is being used for sundry
purposes, and not for caching.
There is a performance database here. Site uses popups. When you
try the "average random access time (read) in milliseconds" table,
the top of the chart is actually an MTron Flash drive (0.1 milliseconds).
Underneath that, is the Seagate Savvio 15K.1 at 5.4 milliseconds.
The WD Velociraptor (a 10K drive) is 6.8 milliseconds.
Next, you can try "Maximum Transfer Rate (Read)", which is a
measure of how fast you could sustain the transfer of a large
file. The Velociraptor wins this one, at 127MB/sec. The
Savvio 15K.1 is only 108MB/sec. But for a person running
a database, they'd probably prefer the 5.4 millisecond
access time, over setting a transfer rate record. So not
every user, looks for the same characteristics in their
drives. Some need transfer rate, others, lower access time.
If I had a choice of those two drives, I'd have to take the
Savvio, as I'd rather have lower access time. My typical
computer usage pattern, makes better usage of access time
than of maximum transfer rate.
- » ssh on command line: force using a group size (prime size) of 1024 (and no...
- — The site's Newest Thread. Posted in » Secure Shell Forum