Post #154,943
5/13/04 1:38:58 AM
|
Went to an HP strategic briefing today. Good except for
one session I opted for run by Microsoft (I thought they would cover .NET strategy stuff).
The guy speaking proceeded in a schoolboyish fashion, to tell the tiny audience (room was the largest at the event, holds 300 people but only 32 turned up to this break-out), all about Microsoft and security.
After 15 mins of his drivel I decided I had better things to do (read the dailies).
The best way to describe his pitch was "We at Microsoft are wailing & gnashing our teeth at how pained we are about security breeches damaging your servers - believe us when we say we are doing *everything* concievable to teach our programmers to be security minded yadda yadda yadda"
Refined further, "Microsoft feels for you - deal with it" :-)
Couple of side notes ...
- Bill Gates was positioned as Microsoft's 'Chief Software Architect' and Ballmer the CEO.
- HP strategies were interesting - on demand computing, unified architecture (centered on Itanium & 32bit Intel), common memory, rack-mount expansion, max 128 MP configs. Growing support for Linux (HP & IBM are going to do a world tour re Linux support).
Doug Marker
|
Post #154,956
5/13/04 8:25:28 AM
|
Re: Microsoft's 'Chief Software Architect'
That happened [link|http://news.com.com/2100-1001-235639.html?legacy=cnet|over 4 years ago] when the price of a Microsoft share was (before 2 for 1 split) 107+ USD. That was not that long after his embarassing videotaped performance (vs. DoJ) in court.
Alex
There are 10 kinds of people in the world: those who understand binary, and those who don't.
|
Post #154,962
5/13/04 9:10:41 AM
|
We had IBM come in yesterday
Dealer along with IBM corp.
Got a presentation on the Fast 600/700/900 storage, the San Volume Controller, their blade servers, and their P-Series.
The lead sales manager was an old Unix admin who used to device drivers. I told my sysadmin (sitting next to me) that his future was in sales. They also had a couple of domain experts. One of which lived 1/2 hour from the office an would be our principle hands on support if we brought in the blade server.
I'll probably end up with a blade server very soon to play with.
|
Post #154,964
5/13/04 10:58:34 AM
|
Watch out to make sure...
That SAN stuff they are pushing is a REAL SAN solution Barry.
If it is the ESS, shitcan it. The ESS, though reliable has severe Bandwidth issues. Severe!
The best part is the the Fibre Connection HUB (yes HUB not switch) is/was connect via multiple 40MB/sec (upto 4 I think) SCSI cable to the ESS.
And it had to chained to add more connections for it. That ESS is a big expensive hole to dump money on. 1TB Shelf was on the order of $85K. The chassis could only hold 2 shelves before you had to get the "expansion cabinet" which only held 4.
Expensive. Now, if they have a better solution(read as true SAN) with SSA disk as the backend... Grab it bucky.
I have zero experience with the Blade stuff from them. I have heard very good things... but then I have heard very good things about Monarch too.
Luck!
--
[link|mailto:greg@gregfolkert.net|greg],
[link|http://www.iwethey.org/ed_curry|REMEMBER ED CURRY!] @ iwethey
Give a man a match, he'll be warm for a minute.
Set him on fire, he'll be warm for the rest of his life!
|
Post #154,968
5/13/04 11:28:42 AM
|
Not ESS
We already have an EMC Sym. Slow piece of crap. We know
what is in that market.
Fast 600/700/900. The difference is in the # of trays and the software.
I'll probably end up with a couple of 900s front ended by a SVC.
You can mix FC and SATA drives in the 900. The techie tried to explain that
the SATA is much slower. We'll see.
He then said it had a much lower duty cycle, ie: 10% vs 50%, and would tend
to fail sooner in a high duty environment.
I asked if the disk warranty / svc price was the same. Yup, it is.
Who cares if they fail twice a month rather than twice a year? 4 time the storage
at 1/2 the cost, fast "enough". Rebuild to a hotspare, maybe use 2 parity disks,
it will be safe.
He realized at that point I knew enough and it was time to stop bsing.
|
Post #154,978
5/13/04 11:58:09 AM
5/13/04 6:04:07 PM
|
This is exactly the points I have been (edited)
trying to make for quite a long time.
Sure SCSI is faster... but faster by how much, and reliability? Is the cost worth it?
If you set up the proper redundancy... who cares.
Exactly why I am using SATA and 3ware cards.
For instance: Gimme a machine with 15 PCI-X slots... a fist-full of 12 Channel 3Ware Cards... and a chassis that can hold 144 Drives. (144 drives, -12 for Parity, -24 for hot spares = 108 Drives capacity) x 350GB (341.179 GiB the right way)) == 37.8 TB. (MB == 1000 x 1000 x 1000 Bytes) or 36.9TiB the right way (MB being 1024 x 1024 x 1024 Bytes)
There are machines on the horizon that have expandable PCI-X Busses. They just add another peer to the bus.
--
[link|mailto:greg@gregfolkert.net|greg],
[link|http://www.iwethey.org/ed_curry|REMEMBER ED CURRY!] @ iwethey
Give a man a match, he'll be warm for a minute.
Set him on fire, he'll be warm for the rest of his life!
Edited by folkert
May 13, 2004, 06:04:07 PM EDT
|
Post #155,053
5/13/04 6:01:18 PM
|
Are you telling me you have that machine right now?
|
Post #155,055
5/13/04 6:02:00 PM
|
NO.
--
[link|mailto:greg@gregfolkert.net|greg],
[link|http://www.iwethey.org/ed_curry|REMEMBER ED CURRY!] @ iwethey
Give a man a match, he'll be warm for a minute.
Set him on fire, he'll be warm for the rest of his life!
|
Post #155,054
5/13/04 6:01:22 PM
|
Are you telling me you have that machine right now?
|
Post #155,056
5/13/04 6:02:05 PM
|
No.
--
[link|mailto:greg@gregfolkert.net|greg],
[link|http://www.iwethey.org/ed_curry|REMEMBER ED CURRY!] @ iwethey
Give a man a match, he'll be warm for a minute.
Set him on fire, he'll be warm for the rest of his life!
|
Post #155,072
5/13/04 7:47:30 PM
|
Here is a brief summary of the HP event just written
The event yesterday was very good value in learning HP's strategies & directions. HP's catchword was 'Adaptive Enterprise'
Themes:
-- Adaptive Enterprise (flexibility & adapdability being the drivers)
-- 'Virtualisation' of resources
-- 'On Demand' ability to instantly expand capacity if needed
-- Platform consolidation strategy. Reduce # of HP chipsets & OSes
-- Configuration flexibility
************************************************************************
Some key points learned from the event include ...
- That HP are consolidating their multiple platforms down to two hardware chips & 3 systems platforms
o Current hardware architectures include MIPS (NonStop), Alpha (exDEC), 32-bit Intel, Itanium, PA-Risc.
o Future chipsets will be 32-bit Intel & Itanium and OS will be Windows, Linux, Non-Stop.
o The 3 standardised platforms will be Windows (32-bit) Pro-Liant, Integrity (Itanium), Non-Stop (Itanium)
PA-Risc will go thru a few more iterations but will be phased out in favour of Itanium. HPUX appears destined to be replaced by Linux (same fate as IBM's AIX). The Linux message was strong but they didn't make a big deal of phasing out HPUX but that came across as a 'work-it-out-for-yourself' thing. (HP & IBM are embarking on a world tour to promote their support for Linux systems).
HP said many of the same things IBM is saying in regard to 'on-demand' computing. On-demand is the ability to expand systems capacity as needed to hande peaks in demand. HP have a similar 'on-demand' strategy for pay-per-use for processors & memory. In this scheme (kind of a lease) you rent/lease a machine with 16 processors but only use 12. The others can be activated 'on-demand' but when they are you are billed for the additional use. HP (like IBM) will only sell this type of config to a company that is likely to use it (peak loads & Fail-over use). The price for this type of facility is higher than for a single machine with max processors. They are also moving to common rack-mount components & expandable computing using blades (again same as IBM announced with Power5). They are deploying a new processor module that allows two Itaniums where there used to be one. This allows HP to scale from 2 to 128 processor MP configs (as does IBM's Power5 strategy).
HP talked a lot about 'virtualisation' of all resources. They say that virtualisation is intended to manage & eliminate excess capacity. (comms, processors, OSes). They mention nPARs (which I assume is the same as IBMs LPAR logical partition technology). The say their management software is heading toward a single management module (as distinct from seperate management modules for comms, storage, processors etc:).
HP also said they with be expanding their SAN capability as well as introducing a SATA cheap storage facility that has dual fibre channel access but is low performance (72,000 rpm cheap ATA disks vs 15000 rpm high-reliability disks etc: etc:).
HP mentioned the move toward GRID computing as an outcome of their virtual resource management.
A lot of the presentations came across as futures & it is up to us customers to accept that HP have the ability to execute the strategies.
In hindsight, the themes are very similar to IBM's recent Power5 announcements but only the chipsets & the actual implementaion software differ. The themes are then, the common Vendor view of IT as it is evolving.
Doug Marker
|
