Disk to disk through network

[kothdvideo]

A little off topic, but basically file transfers
Posted before about transferring lots of big files (gigs) between a
P3 workhorse & a P4 (with main storage). Both machines are 100Mbps
networked and on the same desk, using a kvm switch to manage.
So got gig ethernet cards.
Thought I'd do some "before" timings using a 1,048 meg file &
Explorer. D-Link router in the middle right now.
P3 send to P4 2:32
P3 sucks from P4 2:32
P4 sends to P3 6:43
P4 sucks from P3 5:46
Any thoughts?

 

[Peter]

A little off topic, but basically file transfers

Posted before about transferring lots of big files (gigs) between a
P3 workhorse & a P4 (with main storage). Both machines are 100Mbps
networked and on the same desk, using a kvm switch to manage.
So got gig ethernet cards.
Thought I'd do some "before" timings using a 1,048 meg file &
Explorer. D-Link router in the middle right now.
P3 send to P4 2:32
P3 sucks from P4 2:32
P4 sends to P3 6:43
P4 sucks from P3 5:46
Any thoughts?

Yes. Why do you post those results?

 

[Arno Wagner]

A little off topic, but basically file transfers
Posted before about transferring lots of big files (gigs) between a
P3 workhorse & a P4 (with main storage). Both machines are 100Mbps
networked and on the same desk, using a kvm switch to manage.
So got gig ethernet cards.
Thought I'd do some "before" timings using a 1,048 meg file &
Explorer. D-Link router in the middle right now.
P3 send to P4 2:32
P3 sucks from P4 2:32
P4 sends to P3 6:43
P4 sucks from P3 5:46
Any thoughts?

Remote operations are more costly than local ones?

Arno

 

[kothdvideo]

Yes. Why do you post those results?

Because I think it provides an interesting benchmark to toss things
at. Can they stick to the wall?
And when we have a problem we invite problem solvers.
And who are you? The monitor? Protector of the Disk?

 

[kothdvideo]

Remote operations are more costly than local ones?

Arno

Don't know! This seems an interesting problem. Could be the OS.
Can we rule out the network cards? Anybody know about different flow
rates based on input or output? And hopefully eliminate drivers &
default settings.
Then have some fun debugging. Does it stick? Like I'd like to know.
Seems like something is fundamentally wrong. And it may be cockpit,
but lets find out.
Else plug in the gig cards and say fi. Like you recommended.
Couldn't get the SURECOM but got a couple of sabrent for $35 CDN the
pair.
Will post new figures.
cheers
mrc

 

[Bob Willard]

A little off topic, but basically file transfers
Posted before about transferring lots of big files (gigs) between a
P3 workhorse & a P4 (with main storage). Both machines are 100Mbps
networked and on the same desk, using a kvm switch to manage.
So got gig ethernet cards.
Thought I'd do some "before" timings using a 1,048 meg file &
Explorer. D-Link router in the middle right now.
P3 send to P4 2:32
P3 sucks from P4 2:32
P4 sends to P3 6:43
P4 sucks from P3 5:46
Any thoughts?

Just as a WAG, is the P3 running W9x and the P4 running XP?

I saw similar results with a mixed LAN of W9x & XP. With all-XP, the
STRs were nearly symmetric -- and 70-80% of the 100 Mb/s peak rate.

 

[Peter]

A little off topic, but basically file transfers

Because I think it provides an interesting benchmark to toss things
at. Can they stick to the wall?

Not that interesting. I had a bunch of older PC's with tons of software,
slow, fragmented hard drives, memory overloaded with too many programs,
which exhibited the same performance problems. And that depends on OS too.

And when we have a problem we invite problem solvers.

And should provide as much of technical info as possible to diagnose
problems faster.

And who are you? The monitor? Protector of the Disk?

You want to know? Send me an e-mail.

 

[Arno Wagner]

Don't know! This seems an interesting problem. Could be the OS.
Can we rule out the network cards? Anybody know about different flow
rates based on input or output? And hopefully eliminate drivers &
default settings.
Then have some fun debugging. Does it stick? Like I'd like to know.
Seems like something is fundamentally wrong. And it may be cockpit,
but lets find out.
Else plug in the gig cards and say fi. Like you recommended.
Couldn't get the SURECOM but got a couple of sabrent for $35 CDN the
pair.
Will post new figures.

One thing you should do (unlrelated to your speed observation) is run
checksums or do a compare. If there is any instability you could get
bit-errors otherwise.

Arno

 

[J. Clarke]

A little off topic, but basically file transfers
Posted before about transferring lots of big files (gigs) between a
P3 workhorse & a P4 (with main storage). Both machines are 100Mbps
networked and on the same desk, using a kvm switch to manage.
So got gig ethernet cards.
Thought I'd do some "before" timings using a 1,048 meg file &
Explorer. D-Link router in the middle right now.
P3 send to P4 2:32
P3 sucks from P4 2:32
P4 sends to P3 6:43
P4 sucks from P3 5:46
Any thoughts?

Is either machine running RAID? If so what level and implemented how?

 

[kothdvideo]

Just as a WAG, is the P3 running W9x and the P4 running XP?

I saw similar results with a mixed LAN of W9x & XP. With all-XP, the
STRs were nearly symmetric -- and 70-80% of the 100 Mb/s peak rate.

Good one. Yes P3 is W98SE & P4 is XP Pro.
So if just change NICs should see the same skew.
thanks

 

[kothdvideo]

Is either machine running RAID? If so what level and implemented how?

No RAID.
P3 disk is primary ide master - 80 gig samsung (W98se)(only drive on
machine).
P4 source/target disk is primary ide slave - 80 gig maxtor (XP Pro,
but running off of a 200 gig Samsung, on a Promise controller).