Access times / Drive rotation speed affects in RAID 5 array

[Bowser17]

I am wonding what type of increase i will see in a RAID 5 setup using
SATA 150 7200 RPM drives vs SATA 150 10K drives (WD Raptors). The
raptors are much more expensive than some of the 7200 counterparts,
but the access times are much faster. I would be using these in a
standard workstation environment, used for Digital photo editing,
programming / development, DVD authoring. I cant see to find a
comparison or benchmark specific enough to compare different drives in
an array, but only how many drives, or different RAID controllers. If
you can hit me on my email as well as posting to the group, this is my
first post, so im not sure if i will get notified.

 

[Arno Wagner]

I am wonding what type of increase i will see in a RAID 5 setup using
SATA 150 7200 RPM drives vs SATA 150 10K drives (WD Raptors). The
raptors are much more expensive than some of the 7200 counterparts,
but the access times are much faster.

Not really. WD is of course pretending a huge difference.
First, RAID5 does nothing for access times.

A typical 7200 rpm drive has 4.1ms access latency. This time
is allways present and is needed after the head seek is complete.
A 10.000 rom drive has 3ms latency.

As for seek time, it is interesting to see, that sime manufacturers
(Seagate, e.g. ) do not list it in their drive datasheets anymore.
I would call that lying. Still, from the product manual, average
seek for a 7200.10 is 8.5ms reading and 10ms writing. For a
10.000 rpm raptor-X, this is 5.2ms and 4.6ms. Alltogether
12.6ms/14.1ms vs. 8.2ms/7.6ms. So the access time is much faster.
Allmost twice as fast.

As for throughput, the c't magazines "Platten-Karusell" benchmarks
are very good. RAID with a good and fast controller will multiply
trhoughput by the number of drives minus one withing limits.
A 7200.10 500GB drive gets about 60MB/s reading and writing on
average. A 150GB Raptor-X gets 73MB/s reading and 72MB/s writing.
This is barely 20% percent more and borderline noticeable.

I would be using these in a
standard workstation environment, used for Digital photo editing,
programming / development, DVD authoring. I cant see to find a
comparison or benchmark specific enough to compare different drives in
an array, but only how many drives, or different RAID controllers. If
you can hit me on my email as well as posting to the group, this is my
first post, so im not sure if i will get notified.

For any large-file application (photo editing should qualify, DVD
authoring does qualify) the Raptors are a waste of money. They
are best in small-access applications, like database access.
For software development you may experience a moderate speed
improvement, but I doubt enough to justify the cost. Of course
some people that allways have to have the fastest solution will
disagree with this. Personally, I would advise you to go with
standard drives and invest the money saved in more RAM and
backup.

Arno

 

[Folkert Rienstra]

Not really.

Uhuh: "So the access time is much faster. Allmost twice as fast."
Babblebot so obsessed with himself.

WD is of course pretending a huge difference.

But not you, Babblebot: "Allmost twice as fast".

First, RAID5 does nothing for access times.

Not for the individual drives in it, no.

A typical 7200 rpm drive has 4.1ms access latency.

No babblebot, *all* 7200 rpm drives have an average 4.2 ms latency.

This time is allways present and is needed after the head seek is complete.

Nonsense. It's an average: based on half a rotation.
And drives can read a number of sectors in the sequence it encounters them first, they don't have to spin to the target sector first
before
they can start reading.

A 10.000 rom drive has 3ms latency.

As for seek time, it is interesting to see, that sime manufacturers
(Seagate, e.g. ) do not list it in their drive datasheets anymore.

I would call that lying.

Corse you would, you are Babblebot.

Still, from the product manual, average seek for a 7200.10 is 8.5ms
reading and 10ms writing. For a 10.000 rpm raptor-X, this is 5.2ms
and 4.6ms. Alltogether 12.6ms/14.1ms vs. 8.2ms/7.6ms.

So the access time is much faster. Allmost twice as fast.

As for throughput, the c't magazines "Platten-Karusell" benchmarks
are very good. RAID with a good and fast controller will multiply
trhoughput by the number of drives minus one withing limits.

In (simplyfied) theory.

A 7200.10 500GB drive gets about 60MB/s reading and writing

on average.

Nope. Thats outerband. The average is (outerband + innerband)/2
For that 7200 innerband is only half of outerband.

A 150GB Raptor-X gets 73MB/s reading and 72MB/s writing.

Again, outerband. And Storagereview says 88MB/s, 60MB/s innerband.

This is barely 20% percent more and borderline noticeable.

It will be different for the average: 100.9 48MB/s, Raptor 74MB/s.

And so what. The topic was accesstime.
Random access MB/s will be much lower for the 7200 than for the 10k.