[AUUG-Talk]: 1988: Specs and costs of mini-computers

Peter Yardley petery at eng.uts.edu.au
Mon Nov 21 15:35:11 EST 2011


Take a look at ...

steve jenkin wrote:
> I was hoping some of you on-list might remember the specs and costs of
> mini-computers circa 1988.
> A 2009 piece on ACM Queue, "Triple-Parity RAID and Beyond" by Leventhal
> sent me back to re-reading the seminal 1988 paper by the UCB group
> (Patterson et al):
>    "A Case for Redundant Arrays of Inexpensive Disks (RAID)".
> It wasn't until I read the '88 paper I put the BSD / CSRG at UCB connection
> together... So this group would've had a lot of VAX knowledge.
> I've been trying to recalculate the various cost/MB given in the paper.
> The RRP's given in the press (Computerworld) and IBM Archives for
> 3380 disks (7.5Gb, 14" platter, 6.5kW) and their controllers suggest to
> me $63/Mb for 'SLED' (Single Large Expensive Disk) rather than the
> "$18-10" in the Patterson paper. [Educational pricing?]
> I haven't tracked down the RRP's of the Conner CP-3100 (3.5", 100MB
> SCSI) drives mentioned in the paper for "$10-7/MB", but have found an
> on-line collection of disk prices suggesting the drives would've cost
> around $1,000 ea. (Dahlin's collection below)
> Have no real idea about SCSI controllers at the time, costs or models.
> Back then I worked on a new PDP-11/23 system with dual SCSI disks. I
> believe we used Emulex SCSI controllers.
> Can't remember price of system or disks. $30-$50,000 probable.

We had some Emulex controllers (don't think they were SCSI) on our VAX 
750 and think they cost about $10k.

> What I'm writing to you about is how, in 1988, a system mentioned at the
> end of the paper could've been built and a guess at their cost:
>    comparing an IBM 3380 disk
>    with 100 * 100Mb CP-3100 disks,
>    costing "disks and controller" $11-8/MB.
> I think there are 3 reasonable configurations to consider:
>   - small scale. eg. PDP-11 with 2-4 SCSI controllers and some uplink.
>   - high-end device: VAX with 16 SCSI controllers + IBM channel(s).
>   - high-end system: VAX with 16 SCSI controllers, part of a VAX cluster.
> In their follow-up paper in 1994, Chen et al, "RAID: High-Performance,
> Reliable Secondary Storage" use two widely sold commercial system as
> case studies:
>    NCR 6298 and
>   StorageTek's Iceberg 9200 Disk Array.
> The (low-end) NCR device was more what we'd call a 'hardware RAID
> controller' now, ranging from 5 to 25 disks. Pricing $22-102,000.
> It provided a SCSI interface and didn't buffer.
> The StorageTek's Iceberg was high-end.
> Starting at 100GB (32 drives) for $1.3M, up to 400Gb for $3.6M.
> Not sure the interface, probably multiple IBM 'channels' (ESCON by then?)
> I've been able to find technical data on VAX systems:
> HP's page for VAX 6000/300 tantalisingly says:
> <http://www.compaq.com/alphaserver/vax/archive/vax6000_300.html>
> 'VAXcluster I/O Servers:	Up to 15'
> with
> 'Max I/O Throughput:	60 MB/s'
> Whilst NetBSD provides a very rich catalogue of the VAX series, it's too
> much volume for me, cryptic and without pricing.
> <http://netbsd.org/docs/Hardware/Machines/DEC/vax/vax8000.html>
> <http://www.netbsd.org/docs/Hardware/Machines/DEC/vax/vax6000.html>
> Through this I have learnt that the original SCSI bus was 8-bits wide,
> single-ended and did 5MB/sec (uni-directionally?).

My memory is that SCSI-I was 1 MByte / sec if you had TTL transcievers 
and 500Kbytes / sec if you only had OC ones (like the early Apples). 
IPI-2 did about 6MBytes / second and ESDI (which was pretty common) did 
about 2 Mbytes / sec.

> The UNIBUS, the original PDP-11 backplace, did 5MB/sec, whilst the
> cheaper Q-BUS did 3.3MB/sec.

Think Q-Bus was on the LSI-11, but it could have been on the lower end PDPs.

> There was also SBI, VAXBI, XMI and CI (Cluster Interconnect), with
> speeds of 13.3MB/sec to the "60MB/sec" above:  maybe for 6*VAXBI.

I think the VAX had Unibus and Massbus as an IO Bus. The main bus on a 
VAX 780 was 10 MWords / sec. And I think the bus was 64 bits wide even 
tho the machine was 32 bit. That was Synchronous Backplane Interconnect 
(SBI) by memory.

> If you can point me at any resources or recount any of your
> memories/experiences at the time, especially with rough pricing and
> configs for my 3 'reasonable configs', I'd much appreciate it.
> If you actually built a somewhat comparable system at the time, VAX or
> not, that'd be even better :-)
> I already have one recollection from the time:
>   At Sydney Uni, they paid $30k for a 1MB RAM upgrade for a VAX (11/780?)

Sydney Uni Electrical Eng had 2 x VAX 780 and they home brewed their own 
clustering. The memory boards were 1 Mbyte a piece and you could have 8 
(I think) if you had twin memory bus controllers.

> regards
> steve jenkin
> -----------------------------------------------------------------
> IBM Archives: '20th Century disk storage chronology'
> <http://www-03.ibm.com/ibm/history/exhibits/storage/storage_chrono20.html>
> Dahlin's Disk price data: RRP's of HDD's
> <http://www.cs.utexas.edu/~dahlin/techTrends/data/diskPrices/data>

._--_|\   Peter Yardley                    |
/      \  Senior Network Administrator     | Peter.Yardley at uts.edu.au
\_.--._*  Information Technology Division, | Ph:  +61 2 9514-2358
.     v   University of Technology, Sydney.| Fax: +61 2 9514-2435

UTS CRICOS Provider Code:  00099F
DISCLAIMER: This email message and any accompanying attachments may contain
confidential information.  If you are not the intended recipient, do not
read, use, disseminate, distribute or copy this message or attachments.  If
you have received this message in error, please notify the sender
immediately and delete this message. Any views expressed in this message
are those of the individual sender, except where the sender expressly, and
with authority, states them to be the views the University of Technology,
Sydney. Before opening any attachments, please check them for viruses and

More information about the Talk mailing list