Re: ENC: Performance

I wholeheartedly agree with all your comments.
I believe the person only had one drive and my point was
that the single IDE drive was probably not his bottleneck.
But, he didn't state quantitatively what was 'poor performance'.

Our experience with news machines running INN is that
performance is limited by transfers/second, not by
raw disk throughput. I expect squid is the same.

I'd be interested to see what performance levels other people have
achieved on various off-the-shelf platforms. In terms of
requests/second, bandwidth, and latency, how far can one push
the intel platforms? Is 100 requests/second achievable?

> On Wed, 3 Feb 1999, Jon Fraser wrote:
>
> > I know someone will bring up using SCSI vs IDE drives. As far
> > as SCSI vs UIDE goes, I've seen sustained data rates of
> > 8mbytes/second and up to 90 transfers/second from our UIDE
> > drives. Interrupt utilization seems to be a bit higher than
> > scsi, as our IDE drives/controllers can only tranfser 16 blocks
> > at a time. So, I don't think that the IDe drives are your
> > bottleneck.
>
> SCSI vs IDE isn't a question of throughput, its a question of concurrency.
> IDE can only support 1 device in operation per channel at any given time.
> So, in a system with 4 IDE drives, only 2 can be doing anything useful at
> any given time. On a SCSI system with 4 drives, all 4 drives can be busy
> at the same time. And even better, you can go beyond 4 drives.
>
> So, obviously, SCSI shows its true advantages when you get a system that
> has more than 2 drives (but, if you have 2 ide drives, make sure one is on
> the primary channel and the other is on the secondary channel!).
>
> Also, in terms of performance, more drives is better. For a 16 gig cache,
> you are better off with 4 4gig drives than 2 8gig drives, and MUCH better
> than 1 16gig drive.
>
> /-----------------------------------------------------------------------/
> / Eric Stern - PacketStorm Technologies - (519) 826-9395 /
> / http://www.packetstorm.on.ca /
> / WebSpeed - a transparent web caching server - available now! /
> /-----------------------------------------------------------------------/
>
>
>
Received on Thu Feb 04 1999 - 10:33:04 MST