cancel
Showing results for 
Search instead for 
Did you mean: 

Big Blue versus HP

SOLVED
Go to solution
Mike Fisher_5
Trusted Contributor

Big Blue versus HP

Hi happy Friday forum folk

I'm trying to switch a client from buying these:
IBM p630 servers with 4 x 1.45GHz CPU's
AIX OS

To buying these:
L or N-Class servers with 4 x 875MHz CPU's
HP-UX OS

Client needs to minimise CPU's because of licensing per CPU cost

QUERY:
What's my ammunition for persuading them to go HP?
[Other than IBM's awful performance at the Atlanta Olympics]
For example... I think I recently read a post that said the PA-RISC chip was 'better' than whatever the IBM chip is - but I could be mistaken

Mike
Don't get mad - get naked
Don't get mad - get naked
20 REPLIES
Pete Randall
Outstanding Contributor
Solution

Re: Big Blue versus HP

Mike,

This is almost completely off topic but the tiny little bit of relevance compells me to share:

At HPWorld in Atlanta last week, they had what they call a vendor expo Wednesday evening. That's where they provide beer, wine and food right on the expo floor so you can get mildly sloshed and commit to buying something you never needed (at last I think that's the point!). All during the day Wednesday, IBM had a small herd of Ford Mustang convertibles circling the convention center and the hotel sites. Each of these cars had a sign on the side saying "Win Me". They also had several flat bed trucks with IBM servers on the back circling as well. All this to get you to go to IBM's presentation/reception that same evening. Out of curiosity, we did go (actually John Poff talked me into it - I hate IBM). There couldn't have been more than 75 people there. We stayed for the presentation (they actually had a really good speaker), the drawing for the Mustang (even with 2 out of 75 odds - 3 out of 75 if you count JP, Jil and I - somebody else won the damn car), wolfed down some decent chow, slammed down a couple of adult beverages, and beat feet back to the convention center to catch the tail end of the expo reception.

My point here is that it seems to me that IBM went to one hell of a lot of expense and effort to draw HP folk away from an HP-centric event. That in itself tells me a lot about which is the better company and which one is trying to catch up!

I certainly hope that someone can come up with some hard facts for you, but, as I said, I hate IBM so I don't compare and am not familiar with their offerings.


Pete


Pete
Robert Gamble
Respected Contributor

Re: Big Blue versus HP

Mike,

However this goes for your client, IBM or HP, make sure you weigh all the facts.

1. try finding specs for 4 cpu IBM vs 4 cpu HP's.

2. price of hardware

3. price of vendor support for same amount of time

4. local support personnel training costs

5. im sure i missed something, so that goes here ;)

Hope this helps!
Geoff Wild
Honored Contributor

Re: Big Blue versus HP

Here's a couple of links for you - the same - one on HP's site and the other on a third party. It's older (2001) but the main point for you is the comment "This new single-system 1,000 GB TPC-H benchmark beats the IBM RS6000 SP550 clustered system -- while using half the number of processors" speaks for itself.


http://www.hp.com/hpinfo/newsroom/press/2001/010806a.html

http://www.tdworld.com/ar/power_hp_superdome_server/index.htm

Rgds...Geoff
Proverbs 3:5,6 Trust in the Lord with all your heart and lean not on your own understanding; in all your ways acknowledge him, and he will make all your paths straight.
Stefan Farrelly
Honored Contributor

Re: Big Blue versus HP

In my opinion the PA-RISC ships are definitely faster. Ive got a slightly old HP 9000 & Competitor spreadsheet which lists all HP servers Vs Sun/IBM etc. I dont have specs for the latest cpu's youve listed, but pretty close;

4 way L-class/rp5410 (750Mhz) 50,000 TPM
4 way N-class/rp7410 (750Mhz) 50,000 TPM

IBM p690 4x1300 Mhz 60,000 TPM

So you can see an HP server with cpu's quite a bit slower (Mhz wise) than the equivalent IBM cpu's is almost as quick, TPM wise.

Hopefully someone from HP reads this and has access to a later version of this guide and can give you exact figures for the 875Mhz HP and 1.45Ghz IBM cpu's.

Im from Palmerston North, New Zealand, but somehow ended up in London...
Mike Fisher_5
Trusted Contributor

Re: Big Blue versus HP

Pete:
Illuminating story - could make it part of my presentation
Smacks of desperation
I'm sure that had you won the draw you would have refused on a point of principle
[HP might have rewarded you with a mousemat & pencil for the publicity]

Forum all-star 2000:
Thank you

Geoff:
Useful links - still thinking about them

Stefan:
Good !!

All:
Stefan's info - just out of curiousity
Is that only to do with the reduced insruction set?
or are there other factors such as...
Other aspects of the system architecture?
HP-UX OS less bloated?

Mike 'now working from my laptop at the bar' Fisher
Soon be my turn to chalk up for the darts
Don't get mad - get naked
Pete Randall
Outstanding Contributor

Re: Big Blue versus HP

Mike,

I found a comparison chart:

http://www.dl.ac.uk/CFS/benchmarks/compchem.html

Unless you want to wade through that heavyweight technical stuff, just skip down to "Table 1. SPEC CPU2000". HP beats everybody!!!


Pete


Pete
Stefan Farrelly
Honored Contributor

Re: Big Blue versus HP

Pete,

that benchmark site is awesome. There are a lot of stats there though. In straight out cpu performance the IBM 1.4Ghz jobs beat the PA-RISC 875's hands down - only the HP Itanium2 1.5Ghz's are faster. But - the DL_POLY behcmark which is a 'server' test in that in performs calculations of a specific nature under the OS and times them shows the PA-RISC 875 at 109%, and an IBM 1.3Ghz at 100% - no stat for the IBM 1.4Ghz but I would expect it to be not far short of HP's 109% - but still same comparison I made earlier - that the IBM chip running at 1.4Ghz on a server is comparable to an HP server running only an 875Mhz PA-RISC chip.
Im from Palmerston North, New Zealand, but somehow ended up in London...
Pete Randall
Outstanding Contributor

Re: Big Blue versus HP

Stefan,

I was getting a headache just trying to sort through all the stats. You're right, it's an awesome site - and current, too! Most of the tests I looked at were run in June, 2003!


Pete "Even a Blind Squirrel Finds a Nut Once in a While" Randall




Pete
Stefan Farrelly
Honored Contributor

Re: Big Blue versus HP

Pete - have you seen any stats though for an Apple Xserve or G4/G5 running Mac OS X (Free BSD) ?? Apple keep boasting but ive yet to see any standard benchmark comparisons with real Unix servers. Still trying to find some. Apple have a 2Ghz RISC chip with L3 cache which looks pretty damn quick.
Im from Palmerston North, New Zealand, but somehow ended up in London...
Mike Fisher_5
Trusted Contributor

Re: Big Blue versus HP

Any views on my last query about the cause?
[see my last post above]

I'm doing a hutch clearance on bunnies
All gotta go by Labor Day

Mike '3rd Guinness' Fisher
Don't get mad - get naked
Bernhard Mueller
Honored Contributor

Re: Big Blue versus HP

Mike,

probably your customers primary goal is to reduce total annual costs (licenses and maintainance) and therefore I believe you might have better chances if you offered Itanium CPUs.

Regards,
Bernhard
Mike Fisher_5
Trusted Contributor

Re: Big Blue versus HP

Bernhard:
Why?
Don't get mad - get naked
Stefan Farrelly
Honored Contributor

Re: Big Blue versus HP

Mike - the IBM and HP RISC chips some completely differently designed, but there are some comparisons which you can make;

IBM POWER4+
Instruction cache 64KB
Data cache 32KB

HP PA-RISC 8700+
I-cache 750KB
D-cache 1.5MB

So you can see from this the HP chips have much larger cache sizes. IBM seem to be focusing on Mhz speed (as most manufacturers do nowadays) whereas HP seem to concentrate on throughput - they dont mind their Mhz ratings are lower than competitors. Thats why the SpecInt and fp benchmarks are so strange - IBM are miles ahead of PA-RISC - but in the real world, on a server running an OS, the HP is ahead despite its much lower Mhz speed.

On a side note, ive just run a string with 1000000 iterations (nice c program provided by someone on the itrc) on;
D-class 120Mhz 23s
rp7410 750Mhz 4.7s
Apple Xserve 1Ghz 2.2s (running OS X - Freebsd)

Cor - see what I mean ? the Apple looks quick - and thats only the 1Ghz version, not the 2Ghz version!

Im from Palmerston North, New Zealand, but somehow ended up in London...
Mike Fisher_5
Trusted Contributor

Re: Big Blue versus HP

Stefan:
Bl**dy hell - puts Jonah Lomu to shame
Some goods news about him
http://thestar.com.my/news/story.asp?file=/2003/8/22/sports/6111270&sec=sports

Pete:
You might want to check out the AVI file download near the bottom of this page
http://www.nzallblacks.net/downloads.asp
The Haka - Be afraid - very afraid
Don't get mad - get naked
Bernhard Mueller
Honored Contributor

Re: Big Blue versus HP

Mike,

with the same # of CPUs the Itanium with 1.5 GHz should have a performance gain over IBMs 1.45GHz. So the probability the customer soon needs an upgrade is lower.

Moreover, design cycles for new, faster processors will probably extend even more for PA-RISC, so the perfomance gain curve is expected to fall behind Itanium over time.

That's what I heard, and sorry for not stating clearly what I meant in the previous post.

Regards,
Bernhard
Dave Wherry
Esteemed Contributor

Re: Big Blue versus HP

One other thing about the cache, and I hope I'm remembering this correctly. IBM puts 2 processors on a board. Some level of the cache, it might be L3, is shared by the 2 processors. Where the cache on the HP systems is dedicated to a processor.

I had an IBM Partitioning class last year. While he did not mention the p630 the instructor told us something I did not like about the p690. He had been told if you need to replace or add a processor to a p690 there are several components in front of the processor slots that need to be removed. He was told to allow 8 hours to replace or add a processor. It makes me wonder if there are any gotcha's like that with the p630.
Rory R Hammond
Trusted Contributor

Re: Big Blue versus HP

We have older HP's and IBM P630's and P650's. Performance has been very good on the AIX box's but we have not yet load them up.

The smit interface logs all commands to asmit.script file that make automation easy.

AIX is NOT unix, I dislike like the "look and feel" of AIX. IBM'S out of box Vertas does allow you to shrink and expand files.

It is combersome to "define, undefine and working with devices" (lan cards, scsi cards fibre cards).
AIX volume group and LVOL mplementation is not as well done as HPUX. but I am not sure if it just my "relgious convictions".

I do not like the FSTAB implementation.

You can use "rc.xx" startups but you have to configure it manually. The syslogd facility was added as an after thought (my op). They use errpt which has some very good information but it is a mainframe throw back, which make you look in even more places for error messages and information.


The third party "monitor" is top like. I have not found an IBM alternative to HP Glance. It is cheaper to buy a New IBM box than it is to add a extra CPU.

Software costs are expensive. Though GNU type products are supported well on the IBM box.

IBM doc is volumious yet obscure, frustrating and hard to pick through.

HPUX software, doc and support is worth many thousand of dollars more than IBM's.

Retraining is also a cost that needs to be considered

If I had more time to think about it, I could give you more reason's NOT to BUY IBM.

Rory
There are a 100 ways to do things and 97 of them are right
Veronica_6
Occasional Visitor

Re: Big Blue versus HP

Pete, I was also at the IBM event. The speaker, Tascott(sp?) was indeed pretty good (though a little eccentric at times). The mustang would have been nice. After the event I shook the guys hand that won it (dennis vanamburgh sp?). I felt like punching him out right there, he was too young to win that car. IBM did indeed put on quite a show. If it weren't for my morbid curiousity to the people on stilt's I would probally have never gone.
Mike Fisher_5
Trusted Contributor

Re: Big Blue versus HP

Veronica - Welcome
Don't get mad - get naked
Pete Randall
Outstanding Contributor

Re: Big Blue versus HP

Veronica,

Thanks for the name of the speaker. I keep coming up with Don Tabstock but I don't mention it because I know it's wrong. Anyway, we just went for the free food, free booze, the chance at the Mustang (have you ever seen a less enthusiastic winner?) and had every hope of being back at the vendor expo much sooner than we actually were able to get back.

Hope you enjoyed Atlanta!


Pete


Pete