[Samba] What project should I use ?

[Phil Brutsche]

John H Terpstra wrote:

 >> Oooh, I think I should add some of my know-how...
 > Indeed! And I hope you don't mind my reply - no criticism intended.

Not at all.  Criticism is good.  It can point out mistakes one may not
realize.

 >>> I'd build a box using 3WARE IDE RAID - it's faster and more cost
 >>> effective than SCSI is today.
 >>
 >> I dunno about the performance, but the price is certianly right.
 >
 >
 > From my evaluations on SuSE Linux 8.1 and Caldera OpenLinux 3.11
 > (with 2.4.18 kernel) using 3 x 60GB WD 7200 IDE drives on a 7500-4
 > controller I could get peak I/O of 452  MBytes/sec, and a sustainable
 > I/O rate of over 100 MBytes/sec. That is not exactly a 'dunno'
 > performance situation. These tests were done using dbench and RAID5.

I was using bonnie and RAID5.  There is a difference in drives - I was
using Seagate Barracuda IVs; you were using the WD Special Edition
drives with 8MB cache (the Barracudas have a 2MB cache).

 >> For the boot drive I would use a 40GB or 60GB RAID1 using a 2-port
 >> 3ware card.  There's nothing quite like the hell of that lone hard
 >> drive going bad at an inopprotune time.
 >
 >
 > Granted that redundancy on the boot drive is nice, you could run it
 > off a RAID controller, but then it you ever update the OS, and happen
 > to lose your RAID driver you are kind of HOSED!

In this case the RAID driver is the same as for the RAID5 data store.
If you lose that driver you're hosed period :)

 >> Watch out for the hardware RAID5 on those 3ware cards.  The
 >> performance on those is... disappointing to say the least.
 >
 >
 > Really? What was your setup? I too found them disappointing in 32-bit
 > slots. But in the Tyan K7 64-bit 66MHz PCI slots they simply roar!!
 > Oh, I did have to mess around with the driver.

I've used 6410, 7410, and 7500-4 cards in Supermicro P3TDL3 and 370DLE
motherboards - they have 2 64/66 PCI slots on a dedicated bus.

The difference between the motherboards:

The 370DLE supports up to 2 1GHz Coppermine PIIIs and lacks a U160 SCSI 
controller
The P3TDL3 supports up to 2 1GHz Coppermine or 1.4GHz Tualatin PIIIs and 
sports a U160 SCSI controller

These motherboards have ServerWorks ServerSet III LE chipsets.

Each system had either 1GHz Coppermine or 1.13GHz Tualatin CPUs and 1GB
ECC RAM from Crucial.

The OS is Debian 3.0 with kernel 2.4.19.

 > I found the driver provided by 3Ware out-performed the standard Linux
 > kernel one significantly in 2.4.18 kernel.

That could well cause the difference in performance numbers we see.

Hrm... the driver from 3ware's website is the same as the driver in 
2.4.20, and both are radically different from 2.4.19.  Something to play 
with!

 > Your 75 MB/sec seems VERY poor if it was in a 64-bit slot, but is
 > approx. what I got in a 32-bit PCI slot. The Tyan K7 motherboard I
 > suggested has 64-bit 66MHz PCI slots.

The saturated PCI bus was with a 6410 - a 32/33 4-port PCI card, as well
as a 7500-4 in a 32/33 PCI slot.  Like I said, I expect to go faster
once I put it in a 64-bit PCI slot.

AFAIK the 3ware cards are all 33MHz.

 > FYI: The theoretical 0 latency I/O limit of a 32-bit 33MHz PCI slot
 > is: 32 * 33 /8 = 132 Mbytes/sec. If you got 75MBytes/sec out of it
 > that is NOT bad. 32-Bit PCI is NOT a good solution for file and print
 > sharing for 50+ users, considering how cheap a 64-Bit PCI solution is
 > now.

Agreed.  75MB/sec is pretty good considering that it involves (I think) 
a minimum of 4 transactions to get a single stripe of data (1 PCI write, 
1 PCI read for each drive carrying half of the stripe) (drawn out for 
the benefit of the people on the list).

The cheaper of the Supermicro boards (and it's relatives) I mentioned go 
for under $100 USD on eBay and are great deals if you're not opposed to 
used/refurbished hardware (again drawn out for the benefit of those on 
the list :).

 >> Gigabit could be considered overkill.
 >
 >
 > I hope I have answered this point very clearly above. Using
 > 100-Base-T this will be the system I/O bottleneck.

Once you pointed out what we're doing differently, I have to agree.

 > You mileage may vary! ;)

Indeed!

-- 

Phil Brutsche
phil at optimumdata.com