page 1 of 2
Comments: 29 | Views: 132488
scobleizer
scobleizer
I'm the video guy
8754d9d5-9263-42db-b65b-a0e0f8c6d96c.jpg
Maurits
Maurits
AKA Matthew van Eerde
sweet... a terabyte of RAM...
How long until I can get a cell phone with a terabyte of RAM? (just kidding Wink
Maurits wrote:
sweet... a terabyte of RAM...
How long until I can get a cell phone with a terabyte of RAM? (just kidding Wink


Four or five years I would imagine.

In addition to it's 1tb RAM, it also has 64 Itanium2 processors and nearly 200(!) PCI slots and the accompanying bandwidth to handle that kind of network or fileI/O throughput...  An awesome machine, really.

-Bryce

rhm
rhm
What do you put in 200 PCI slots?
Lwatson
Lwatson
One ugly mug...
200 Sound Cards.....

200 IEE1398/USB 2.0 Cards

200 Scsi Cards

hope you have 200 Visa and/or Mastercards


Jeremy W
Jeremy W
that blogging guy
No PCI-X? I'm disappointed Wink

Seriously though, Dell's giving away 2GB of free RAM this week.

We're unlikely to see 1TB of actual RAM on desktops, as solid state disks are likely to come down in price to the point where you won't have RAM or HDD's, you'll merely have "memory" on the solid state drive.

Japan Atomic Energy Research Institute will Introduce 2,048 Processor Linux Supercomputer

MOUNTAIN VIEW, Calif. (November 3, 2004)—Silicon Graphics (NYSE: SGI) announced that the Japan Atomic Energy Research Institute (JAERI) will introduce an SGI® Altix® supercomputer as its new core computation system to assist in developing leading energy systems.

As a result of a competitive bidding process, Fujitsu Limited in cooperation with SGI Japan will deliver to JAERI the new SGI® Altix® 3700 Bx2 model which is based on 2,048 Intel® Itanium® 2 processors, the Linux® operating environment and over 13 terabytes of memory - the world's largest memory capacity. The system is scheduled to fully operate at the end of March 2005.

http://www.sgi.com/company_info/newsroom/press_releases/2004/november/jaeri.html

When I worked for HP Consulting we were using one of these with some lighter hardware to run the following:

- Full Lotus Notes/Domino server
- 50 Virtual PC images of W2003 Server and VS.NET
- A full SAP implementation
- Active Directory,etc.
- Websphere
- and the above environment mirrored for QA

As I said the spec's were lighter:
- 64 gigs of RAM
- 8 Procs
- a couple of Terabytes of storage


It's a great way to run a project.  Got a new developer on the team?  Just create a new Virtual image and remote desktop via your laptop and you're working.

AcidHelmNun
AcidHelmNun
Would you like fries with that?
Any idea what kind of cooling those boxes use?
Vexorg
Vexorg
Got Milk?
Lwatson wrote:
200 Sound Cards.....

200 IEE1398/USB 2.0 Cards

200 Scsi Cards

hope you have 200 Visa and/or Mastercards




And if anyone ever tries to get me to troubleshoot the IRQ conflicts if you try that, I'm headed for the hills.

Any idea what kind of cooling those boxes use?

The local joke is that it uses the power meter's needle as a fan Smiley

Sonic
Sonic
Hungry For Code
That is one machine I'd like to see the FPS rate of Doom 3 on Wink

What's it being used for actually?
ethan
ethan
.::The right place to be::.
Thx a lot to you prog_dotnet and to scobleizer !!!
i dreamt for many years to see such photos!!
it's Xmas before december, nice!!
ZippyV
ZippyV
Fired Up
What does Microsoft Enterpies Engineering Center mean?
Farshad
Farshad
http://news.p​ersianforum.n​et

Its wonderful
pics are great

can you upload more pics?

Sixteen racks loaded with IBM Blue Gene/L supercomputer (70.7 trillion calculations per second )



Teraflop Cluster installed in Tromsø

At 14:00 on Wednesday November 10th, Norway`s first 64-bit Teraflop cluster was installed at the University of Tromsø. In less than 4 hours the HPC-group at UiT had upgraded the system from 100 Gigaflop to 1 Teraflop. The installation was broadcasted live on the Internet, and was watched by over 750 people.

3 hours later the system load was 73%, with 45 jobs spanning 158 processors. The largest job was a 16 processor chemistry application (dirac).

Teraflop cake
Eirik, Roy, Steinar, Svenn, Sølvi, Tor, and Maria celebrates the first Teraflop (Photo by: Lars Slettjord, UiT)



The new Snowstorm-cluster is an HP Itanium2-cluster delivered by IT-partner Tromsø and HP Norway. With the use of the award winning Rocks management toolkit, the existing 22 processor cluster was upgraded to 204 processors without taking down the system. The Rocks cluster management tool allows you to add new HW to the cluster on the fly - without any downtime.

new_snowstorm2

New Snowstorm-cluster


 

Per Node

Total *

Machine

1 HP rx4640

49 HP rx4640

Processors

4 1,3 GHz Itanium2

196 1,3 GHz Itanium2

Memory

4 GB

196 GB Distributed

Storage

144 GB

7 TB Distributed + 2 TB centralized

Interconnect

Gigabit Ethernet

Gigabit Ethernet

Performance

20,8 GFlops/s

1,02 Teraflops/s



*) Snowstorm also consists of 2 interactive nodes (HP rx5670), each with 4 1,3 GHz Itanium2 processors, making the aggregated performace of the cluster to 1,06 TFlops/s.



09:50: In the beginning there were only floorspace. This is also the view of the web-cam which had more than 750 visitors during the day. Sølvi is preparing the cables for what is about to come.

 


09:55: Last pep-talk before the action.


10:00: Rack one is installed by Roy, Maria and Steinar. Soon all the 49 HP rx4640 machines will be in place. Each node has 4 1,3 GHz Itanium2 processors and 4 GB of memory. Aggregated performance for the new cluster is 1,02 Teraflop. With the old equipment in addition it's a total of 1,16 Teraflop.

 


10:05: Eirik and Maria are getting the second rack. The chiefs
wisely decide to keep their hands in their pockets. IT-director Magnar Antonsen to the left.

 


10:15: Eirik, Tor and Roy install rack three.

 


10:15: Steinar is mounting the front of the rack, while Sølvi handles the cables at the back. Eirik and Maria are ready to get going with the next rack.

 


10:20: Our biggest investor is paying us a visit. Tor looks really happy to see Santa. Maybe the last piece is in his sack? Roy on the other hand look anxious to get his hands on the gigabit switches on the table.

 


10:40: Roy is connecting the three Gigabit switches.

 


10:45: Maria is watching while Eirik and Sølvi get the networking
in place. At this point the front-end infrastructure is about to be switched to the new networking equipment.

 


11:15: Roy is ready to "Rocks and Roll". Each node took about 15 minutes to upload with Rocks. By then it was completely installed with software and tools ready to be inserted into the compute-pool. Installation of Rocks was done in parallel as soon as the nodes were turned on. The white rack to the left is the old cluster. At the top seven rx2600's which will be returned to HP as part of the deal. At the bottom the two rx5670's (similar specs to the compute nodes) that are going to be reserved for interactive use in the future.

 


11:55: Svenn is posing in front of the cluster. He looks very satisfied. The first user job has just started on the new nodes. Steinar looks a bit more concerned in the background, one of the nodes did not boot "out of the box".


12:30: Maria scissorhand goes to work. Some of the labels obviously need to be changed. This is rack 3 which you can see in the other pictures has no machine installed at the top. Instead it has three gigabit switches mounted towards the back.

 


13:55: Sølvi and Maria celebrates 1 Teraflop up and running. Only three hours from now the system-load will reach staggering 73%.

 


14:30: The Teraflop cake is being consumed in parallel. From left: Eirik Fossgaard, Roy Dragseth, Steinar Trædal-Henden, Svenn Hanssen, Sølvi Bersås, Tor Johansen and Maria Wulff-Hauglann.

 


14:35: All the good helpers get their share. Jan Andreassen (cutting cake) and Martin Eilertsen were responsible for the web-cast. In the background Børge Brunes who had a hand in the networking part.



SOURCE: http://uit.no/itavd/HPC-News/

ZippyV wrote:
What does Microsoft Enterpies Engineering Center mean?



Before the development of Windows Server 2003, Microsoft relied on inhouse stress tests, the Rapid Deployment Program (RDP), and early adopters and beta testers to gauge how its software would meet customer needs on release. With Windows 2003, however, Microsoft instituted a much more comprehensive set of testing procedures to ensure that the OS would be as solid as possible so that enterprises could confidently roll it out before the first service pack release. An important aspect of Microsoft's new testing initiative is the Microsoft Enterprise Engineering Center (EEC).

What It Is
Microsoft uses the EEC, a facility on the Redmond campus, to interact with customers during the development of Microsoft enterprise products. The idea is to bring a customer in house, learn first-hand the customer's concerns about a product's deployment, then test the product against the customer's exact hardware and network configurations. Microsoft believes that this interaction will help make Windows 2003, Microsoft Exchange Server 2003, and future products more desirable and reliable. For customers, using the EEC offers unprecedented access to Microsoft's product teams. If a problem arises during testing, appropriate Microsoft product team members are notified and help investigate the problem.

The EEC currently consists of three Enterprise Customer Labs (ECLs) that feature IBM, Hewlett-Packard (HP), and Dell workstations and servers, plus a massive server room with banks of rack-mounted servers. Because Microsoft customers generally have heterogeneous environments, the company also maintains a collection of Sun Microsystems machines and other non-Wintel equipment so that the EEC can more readily duplicate customer configurations. If customers need hardware that isn't yet available in the EEC, Microsoft makes it available. Since the facility opened in April 2002, Microsoft has had more than 20 customers on site for Windows 2003 testing. The EEC has hosted as many as five customers at one time, and Microsoft is building new ECLs to support more concurrent testing.

Whom It's For
The EEC is open to virtually any enterprise that's rolling out Microsoft technology. George Santino, who runs the EEC at Microsoft, explained how customers come to the center. "Our account teams in the field send us referrals, while a number of customers come because they're part of the JDP [Joint Development Program]," he said. "And we have some that come from our OEM hardware partners. They provide us with their own customers."

The EEC costs participants nothing more than the price of sending employees to Redmond for several days. In return, Microsoft asks to use the customers' lab environments in future testing situations. "We want to understand how they use our stuff," Santino said. "And we want real data to stress the products, not a simulation.

How It Works
After Microsoft and a customer agree to conduct testing at the EEC, Microsoft establishes a schedule and builds a copy of the customer's environment in an ECL. Then, customer representatives come to the EEC and test various scenarios using actual data. This testing period typically lasts 1 to 2 weeks. "When they leave, we want to keep their environments around for at least a week so our product teams can test on it. If there are enough issues, we like to keep the environment around for a couple of months," said Santino. "We are growing an inventory of real-world environments so that, over time, when a product team is working on product X, we can test various things, go back through the inventory, and build it back up in a couple of days."

For customers, the EEC is a great opportunity to interact with the people who develop the products they use. "Any time we had a server problem, a person from the Windows Server 2003 product team would come across campus within 20 minutes, ... dig into the code, and fix the problem," Tim Cornett, the Active Directory Architect for the Kentucky Department of Education (KDE), told me while discussing his December 2002 experience at the EEC. "It was pretty intense. They were willing to stay with us 24 * 7 if we wanted them to, and they were just excellent." Cornett said that he'll likely return to the EEC to test his Exchange 2003 deployment.

SOURCE:

http://www.winntmag.com/Articles/Print.cfm?ArticleID=39163

Enterprise Engineering Center Intro

My name is Bryce Milton and I'm the test manager for Microsoft's Enterprise Engineering Center in Redmond.  We have a large customer-facing testing facility designed to engage in situations in which we need to conclusively answer technical questions about Microsoft server technology when enterprise customer requirements push the edge of the envelope, drive various and interesting technological “skunk works“ programs, and champion critical customer feedback on server technologies.

By way of background, here's a recent link to an EWeek article about us, and MSN Search which has a lot of hits on various EEC activities and announcements of our various hardware partnerships (The EEC has some amazing capabilities and great OEM partnerships in our arsenal that I'll go in to more detail on in future installments).

Generally, what I'd like to do is use this blog to discuss (in non-customer specific terms) the highlights and lowlights of recent engagements here at the EEC to give the larger Windows Server community a sense of what we do here and how it impacts core Microsoft technology development by inserting Enterprise customers into the development process and helping them solve their tough IT platform and application problems.

-Bryce

SOURCE: http://blogs.msdn.com/eec/archive/2004/09/22/233181.aspx

iStation
iStation
Fuujin
ZippyV wrote:
What does Microsoft Enterpies Engineering Center mean?


Maybe it means a ERP development/hosting center for SMEs (Small Middle Enterprises).

Windows Server + .NET + VSTO + SAP = $$$!

:]
ethan
ethan
.::The right place to be::.
It would be very cool if such video was available here!!!
page 1 of 2
Comments: 29 | Views: 132488
Microsoft Communities