Re: Itanium Madison blasts Sun, IBM in encryption specs!
From: Rick Jones (foo_at_bar.baz.invalid)
Date: 04/30/03
- Next message: John Johnstone: "Improvements in TCP/IP Services anti-spam features"
- Previous message: Tom Linden: "ADDENDUM: Keyboard ?"
- In reply to: Andrew Harrison SUNUK Consultancy: "Re: Itanium Madison blasts Sun, IBM in encryption specs!"
- Next in thread: Andrew Harrison SUNUK Consultancy: "Re: Itanium Madison blasts Sun, IBM in encryption specs!"
- Reply: Andrew Harrison SUNUK Consultancy: "Re: Itanium Madison blasts Sun, IBM in encryption specs!"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]
Date: Tue, 29 Apr 2003 23:09:04 GMT
Andrew Harrison SUNUK Consultancy <Andrew_No.Harrison_No@nospamn.sun.com> wrote:
>> Further, HP has not (to my knowledge) announced prices for the 1.5 GHz
>> systems.
> 1.5 MB cache not 1.5 GHZ systems.
That comment was refering to the systems with 1.5 GHz Itanium 2 6M
CPUs (aka Madison), not the older McKinleys.
> And of course its a very horizontally scalable service
Indeed, web services can be quite horizontally scalable. However,
there have been no clustered SPECweb99_SSL results published to date.
That would seem to suggest that either the workload characterized by
the benchmark is either not horizontally scalable, or isn't easily
horizontally scalable. (SPECweb99_SSL does not prohibit clustered
results)
> box that costs 34K with 2 CPU's hardly makes much sense when you can
> buy a 5 x 2 way V210's with cryptos for the same price.
Do you mean to imply that it would take 5 of the two-way V210's to
match the SPECweb99_SSL performance of a single two-CPU non-SPARC
systems?
> The V210/V240 are new low cost low end 1/2U 1 GHz UltraIIIi servers
> BTW, similar capacity to a Sun V280 but well under half the price.
Sparring with you in Usenet is always fun because you are so fast on
your keyboard :) You were saying how much more expensive than the
Opterons the HP IPF systems were. I then said SPARC systems aren't
that much better compared to the Opterons (in otherwords, the pot
calling the kettle). You've come-back instead with a comparison
between the Sun systems and the IPF boxes instead of the Opterons.
So, lets then look at those V240s you brought-up (there being no
SPECweb99_SSL figures published for the V210, and I doubt you really
meant to imply that it would take 5 hardware-accelerated V210's to
match a two-CPU non-SPARC system at SPECweb99_SSL) and compare them to
those Opterons.
The SPECweb99_SSL result for a two-CPU V240 was 833 [1]. So, to match the
two-CPU Opteron results one would need two of those (actually, two and
a fraction - 1783/833 = 2.14) and to match the four-CPU result would
require 4 (actually 4.2 - ie 5).
Base price per Sun's online pricing pages for the V240 appears to be
$6,495. The SCA 500 adds another $695, so box cost - without the full
8GB of RAM - is $7190. Two of those to keep pace with the two-CPU
Opteron would be $14380, IIRC, you have asserted previously (opening
this pricing can) that the two-CPU Opteron systems (unspecified RAM)
were going for 5K? Four, dual-CPU V240s (should that really be five?)
to keep-up with the single, four-CPU Opteron system would be $28760.
Bill has taken the TPC-C pricing from the RackSaver and concluded that
the Opteron SPECweb99_SSL config would be around $49000. I do not
know if that price is truly "real" or not (seems everything is bundled
into one price, no per-line item pricing, so it is hard to check), but
for now lets go with that and then add the 8GB of RAM to those V240s
since that $49000 figure includes the RAM. That is it seems $1795 for
2, 1GB DIMMs, so 8GB of RAM would be $7180. I would trust that there
is a return credit for the four 512MB boards included in the base V240
config, so credit back that price (795x2) and so the add-on to each
system is 7180 - 1590 or $5590. So, each box is then 7190+5590 or
12,780, four of those would then be $51,120, five would be $63,900.
Heck, if we really want to have fun, then we compare the number of
Gigabit ports used by the V240-based solution - 8 to 10 - to that used
for the single-system four-CPU systems - 4. The V240's are 2U, so
four of them is 8U of rack space to the four-CPU system's 4U. Etc etc
etc...
>> http://store.sun.com/catalog/doc/BrowsePage.jhtml?catid=55844
>>
>> Of course, pricing is always a fun game isn't it - what is in the
>> base config, how much to get that to the tested config, how much of
>> the tested config was required to get the result. For example, was
>> the entire 12x36GB StorEdge 3310 SCSI array and dual Ultra3 SCSI
>> HBA required required to hit that 1008 number, what is the pricing
>> of ZWS compared to Sun ONE Web Server (although it was probably the
>> SNCA doing all the real work as an in-kernel accelerator rather
>> than the web server...), how much does that SCA 1000 or the
>> GigaSwifts cost. All that fun stuff.
> What has this got to do with the pricing of the server itself. You
> seem to have dissapeared off on an alarming Kerry Main type tangent
> which is very unlike your normal posting style.
I'm afraid I don't understand the Kerry Main reference.
If we really want to get into the ever so fun and twisty discussion of
pricing which (iric) you brought-up, and if the _solution_ put-forth
by a vendor to achieve a given SPECweb99_SSL result really did need
all those things from the SPECweb99_SSL disclosure (and there are
indeed times when a SPECweb99_SSL result had stuff it didn't really
need but just happened to be there) then comparing pricing should be
comparing _solution_ pricing not just box pricing. And after that, we
could go futher into the pricing hole and start arguing about whether
it costs more to admin four, two-CPU V240s or one, four-CPU system,
support prices etc etc etc... Probably one of the reasons why SPEC
has never been especially keen on that maze of twisty little
passages...
rick jones
[1] SPECweb99_SSL figures per http://www.spec.org/ as of April 26,
2003, SPECweb being a trademark of SPEC etc etc etc
-- Wisdom Teeth are impacted, people are affected by the effects of events. these opinions are mine, all mine; HP might not want them anyway... :) feel free to post, OR email to raj in cup.hp.com but NOT BOTH...
- Next message: John Johnstone: "Improvements in TCP/IP Services anti-spam features"
- Previous message: Tom Linden: "ADDENDUM: Keyboard ?"
- In reply to: Andrew Harrison SUNUK Consultancy: "Re: Itanium Madison blasts Sun, IBM in encryption specs!"
- Next in thread: Andrew Harrison SUNUK Consultancy: "Re: Itanium Madison blasts Sun, IBM in encryption specs!"
- Reply: Andrew Harrison SUNUK Consultancy: "Re: Itanium Madison blasts Sun, IBM in encryption specs!"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]
Relevant Pages
|
