Some of these vendors I do have direct experience with. Others have
just contacted me and asked to be added to the list, and I've done so
(without even charging them, which is probably slightly insane on my
part). Still others have contacted me with rich bribe offers such as
tee-shirts, penguins, and coffee mugs or tiny screwdrivers, causing me
to fill them in below with a smile.
Clearly I'm a bit of a slut, but I've still got my pride and
standards. Truthfully, I'll list almost anyone if they're really and
truly quasi-dedicated linux/beowulfish vendors, but (for all of
you who read this)...
...USE THEM AT YOUR OWN RISK. I make no warranty whatsoever
about any given vendor being suitable, cheap, or reliable even if I've
used them before. I could be mistaken, after all, or I could be lying.
The latter isn't horribly unreasonable, if the vendor has bribed me
heavily with tee shirts, coffee mugs, free palm pilots, or (always
useful) money to tell you how great they are. So you shouldn't fully
trust even an open endorsement. Who knows how many stuffed penguins
stand behind it?
- Dell
- Dell provide the Intel Grant hardware that we use for Brahma 2 and Brahma 3. In recent years
Dell has reduced prices so that they are very cost-competitive with even
Intrex (our local vendor) and their hardware is at least as reliable as
anything else available. Dell has very nice onsite service deals
(important to control human costs in large clusters) and offer rackmount
systems that make good nodes. One weakness is that Dell tends to be
pretty inflexible about their offerings -- configurator way or the
highway.
- Intrex
- Intrex is a the local vendor "down the road" that provides very
reasonable systems and parts (literally) over the counter, at least to
folks in North Carolina. Driven strictly by demand, they've started to
put together very reasonably priced rackmount systems, where they'll
basically build you exactly what you want at OTC prices. I recently
(10/00) got a quote from them for rackmount 2U nodes -- 800 MHz PIII's
with 256 MB PC133 ECC, a 10 GB HD and floppy, and a NIC per node for
just about exactly $1K/node. This would be even cheaper now as SDRAM
drops towards $0.50/MB from over a dollar at the time of the quote.
Their "integration charge" per node is basically nothing -- maybe $50 --
as they make their primary profit on the volume of the sale, just as
they should.
I like these guys and do business with them pretty regularly. They
are the sole supplier for the systems in my home beowulf because they
are quite cheap and literally two miles away from my house. This allows
me to get service "instantly". For example (true story), 12/4/00 I was
rebuilding my home beowulf (putting it on a nifty heavy duty steel and
particle board shelf unit, actually, to which I had attached rolling
casters -- very cool and cheap!) and naturally my primary server/desktop
(a dual Celeron on the Abit BP6) refused to come back up -- no power at
all. I suspected a blown power supply, but didn't want to mess with the
swap game at home.
So, I simply picked up the unit and carried it to the car, and
trundled off to their South Square store. I plopped it down on the
counter and said "I think the power supply is blown". Twenty five
minutes later I was out with a new power supply installed at no charge
and no questions asked, as the system was still less than a year old (by
around three weeks:-). Of course I also grabbed some newly cheap SDRAM
and even popped for a 30 GB disk for all my MP3's while I was
there...
This kind of service is why I buy from local vendors if/when I can
find a decent one. Intrex is exceptional even in this regard -- their
systems come with a lifetime labor warranty, and that's the expensive
part of buying ANY system -- hardware (replacement or otherwise) is
cheap, but the time required to swap power supplies, motherboards, and
so forth around to identify and replace a bad part is very, very
expensive. To me -- it is my time. I'll therefore make an exception
for Intrex and "endorse" them even though they don't give me coffee mugs
or mouse pads for free. Their prices are low enough I can afford to buy
my own with what I save.
- Penguin Computing
- We have thus far gotten precisely one (1U) penguin HPC compute node
for testing purposes, and our sysadmin, at least likes it very much.
Seems well engineered, and not horribly overpriced. I will try to
update this as we test the node more. If nothing else, you have to love
the name and the cute little logo!
- PsychoSoftPC
- Catchy name, that. These are serious computing freaks from the
look of it, who sell turnkey clusters (called the ``Psychlone
Cluster''), some sort of 64 bit cluster apparently used by clients in
rendering among other things. I'm including them because they asked to,
because they are a vendor of linux-based supercomputers which is what
this is all about, and because they have a cool drop-in logo I can paste
in here:
Oh, and maybe I can get a cool T-shirt out of them for the link.
Take it for what it is worth...
- RLX
- RLX makes and sells a variety of "single blade computers" --
computers that fit onto a card that can be slotted into a special 6U
chassis. They sell a variety of blades -- ultra-low power blades with
transmeta processors to high power P4 blades -- that can be installed in
densities ranging from 24 per 6U to 10(x2) per 6U (the latter is in 10
dual Xeon P4's per blade, or 20 processors per 6U).
Blade solutions generally permit one to achieve the highest possible
CPU densities. Transmeta clusters can run cool, although it is cool and
relatively slow. However, they are expensive, and hence most likely
will be of interest to people with a highly nonlinear cost profile in
their physical infrastructure -- extensive renovation required to house,
power, cool a more traditional cluster with (otherwise) better cost
benefit. Still, worth looking over, and prices can always change.
- Raritan
MasterConsole
- This is a keyboard, video, mouse switch we bought with/for the
original brahma. Can manage at least 64 nodes from one monitor, but the
switches aren't cheap. Very nice. But not cheap.
- Belkin
- This is another keyboard, video, mouse switch maker (and a ton of
other stuff). I use a four port Belkin in my home compute cluster, Eden
(guess that the naming theme is that use at home:-). It was
fairly cheap (around $100), and works perfectly. I've used really cheap
KVM's (basically rotary serial switches with a knob and everything) and
they aren't worth even the very little money they cost. Get a "real"
KVM or don't bother.
- Scientific Applications on
Linux Site
- The name says it all. This is a somewhat "weird" site (in my
opinion, anyway) with a strange mix of commercial and non-commercial
offerings, but it is a useful place to look for advanced stuff for
linux, some of which you might have to pay for (which is why it is on
the vendors page instead of the links page.
- Paralogic
- Paralogic is a turnkey beowulf vendor. They are as much software
and service as they are hardware, although they'll cheerfully sell you
the hardware as well. Doug Eadline (the president of the company) is a
buddy of mine and has given me tee shirts (full disclosure, here) and
has also given back considerably to the linux community (as one of the
authors of the linux beowulf HOWTO, perennial sponsor of free booze at
Extreme Linux and HPC events, and so forth).
- High Performance Technologies,
Inc
- Another turnkey beowulf vendor. I'd say that it is one of the
highest end of the "true" beowulf vendors, working with bleeding edge
technology to, as they put it, "build Linux clusters to meet the
requirement of people that are accustomed to using high performance
computing (HPC) systems or supercomputers". A good group to talk to if
you are trying to achieve performance competitive with the best of the
big iron that scales to hundreds of processors.
-
ASL Computers
- ASL is a vendor of linux (only!) boxes of all sorts, from desktops
through large rackmount servers. They sell both 1U and 2U units and I
imagine will do systems integration sorts of things for you. I don't
know if they consider themselves a true "turnkey beowulf vendor" (I'd
generally reserve this term for companies with a true beowulf expert on
staff and on the beowulf list who can help with more than just the
hardware configuration and initial software install) but they can
certainly sell you cluster components with linux preloaded. If they
have or establish an outsourcing relationship with e.g. Scyld, they
might even qualify without an on-staff expert...
These guys get some extra kudos from me as they've been extra nice to
me over several years. No tee shirt(s), but they've given me the chance
to work with certain bleeding edge hardware for free before it even
"existed". Very nice folks, good quality hardware.
- Atipa Computers/DCG Computers
- A turnkey linux/beowulf vendor. They do (well, probably "did" at
this point) high end alpha clusters as well as pretty much whatever you
like in the way of linux workstations or beowulfish clusters. Wow!
They sent me tee shirts (the note below must be working:-)! They must
be decent guys, and my kids will henceforth advertise their stuff in
school as much as this site does on the web. I henceforth endorse them
as somebody to talk to if you are looking for turnkey quotes.
-
Linux NetworX
- Formerly known as Alta Tech, LN is yet another linux/beowulf
turnkey vendor. They sent me some very cool tee shirts with these
centipedes on the front. They are worth a call if you are looking for
quotes on a ready-to-run cluster system at pretty much any level of
performance (up to the very high end). They also have some very pretty
node-mounting boxen (not exactly racks -- custom boxes -- which is what
I first knew them for as Alta Tech). Here is a synopsis of their
business that they contributed (note that I did not write this, they
did:-):
"Linux NetworX brings its powerful cluster technology to those
demanding high-availability and high-performance systems. With the use
of cluster computer technology, Linux NetworX provides solutions for
companies with high-computing needs including Internet servers,
research, industry, government and other technological fields. Through
innovative hardware, complete cluster management software, service and
support, Linux NetworX provides end-to-end clustering solutions. To
date, the company has built some of the largest cluster systems in the
world and has developed unique hardware and versatile software to
facilitate overall system management. Linux NetworX has offices in Utah,
New York, Calif. and Texas and worldwide distributors."
"About ClusterWorX: ClusterWorX allows users of Linux NetworX
clusters to control the cluster as a single system, and provides remote
monitoring and management capabilities. ClusterWorX can be accessed
through an easy-to-use graphical user interface (GUI), command line and
HTML. Other management tools include remote access, disk cloning and
serial access to nodes, including remote monitoring and resetting of
individual nodes without effecting the uptime of the entire system. Disk
cloning is a valuable feature for large cluster systems because it
allows software and other updates to be installed on one node and
automatically distributed to the entire system. All features are
architecture independent."
- Microway
- Designs and manufactures very high end, custom Beowulf clusters
based on Alpha, Intel and Athlon processors. Here is their cluster
synopsis (note that THEY wrote this, not me):
"Connectivity is provided using ethernet, Myrinet or Dolphin
interconnect technology. The company was formed in 1982 by Stephen
Fried, physicist and coinventor of the HF chemical laser (Star Wars).
Microway has provided state of the art products for high speed numeric
processing to the university and government marketplaces for over 18
years. Microway is API's largest US customer for UP2000 and 21264DP
Alpha motherboards and processors. The company designs proprietary rack
chassis for maximum nodes per cubic foot, and has a fine reputation for
delivering fully configured clusters that work! Microway has configured
Linux based systems since the early days of the Red Hat releases. Today
we specialize in large Beowulf clusters for customers with HTPC
applications from chemical engineering, mechanical engineering, CFD,
molecular modelling, simulations, biogenetic research to designing
America's cup racing keels, and jet engines. Our customers also include
ISPs and other ecommerce companies."
For what it is worth, I do remember Microway from years of reading PC
Magazine and drooling over their co-processor (i810 and other) boards
and fancy matching compilers. Although I have no clue as to where they
stand in the grand scheme of cost-beneficial turnkey beowulf systems (so
you'll have to visit their website and talk to them and find out for
yourself, which you'd do anyway with ALL of these vendors if you had any
sense) I will say that in one sense, they were a direct philosophical
predecessor of beowulfery. They made an attempt to achieve the same
goal of providing commodity supercomputing at an affordable price in a
readily available platform. Of course, they tried to achieve it by
putting 2-4 very high end processors on a single ISA bus card with very
much custom and proprietary high end compilers and a unique
architecture, but there is nothing wrong with that, especially in the
context in which they were working. For a long time, they
probably led the world in cost/benefit measured in minimum commercial
cost/FLOP in a single platform. And they are sending me a T shirt, so
they must be decent folks.
- Numerical Algorithms Group.
- One of the things Microway sold in the old days was libraries from
the Numerical Algorithms Group (NAG). Here is a communication
from NAG:
NAG is a software company that specializes in reusable mathematical
and statistical software components. We have been in business for 30
years and have been involved in many projects with the Department of
Energy, Department of Defense as well as being able to help many NPACI
and Alliance partners. (There are about 40 site licenses within the
NPACI and Alliance members which makes collaboration and sharing of
software containing NAG very easy.) NAG recently released a new version of
the NAG Parallel Library that utilizes MPI for Beowulf computers.
The product works with either PGI's compiler or the Gnu product.
All of NAG's products are available for "test drives". NAG will also
be coming out with a version of our SMP Library for Intel/Linux next
year.
Best regards, Tony Nilles
VP Sales and Marketing
NAG - The World
Leader in Numerical Software Components
PH: 630-971-2337 x 207
Fax: 630-971-2706
www.nag.com
Note that NAG has promised to "try to find something different from
the usual t-shirts (since your kids seem to be well stocked!)" Clearly
creative and intelligent thinkers!
-
HiPERiSM Consulting
- From my friend George Delic (whom I met at a recent HPC conference
at Wake Tech CC):
HiPERiSM Consulting, LLC, offers expertise, products, and services
in:
- High performance computing
- Software engineering
- Data visualization
- System integration for Linux clusters
- Air Quality Modeling with Linux clusters
If I recall correctly, George and I talked about performance
monitoring and tuning tools for serious MPI code at the conference, so
if you have a serious MPI application where tuning it up is of value to
you, give HiPERiSM a look and see if they can help.
-
Ventura Tech
- Ventura assembles "server class memory" with lots of certification
and guarantees, pretested for various popular high end motherboards
(such as those that might well be selected for a high performance
beowulf cluster). Here is their synopsis (from Sam Lewis):
"Many will say that `memory is
memory'. While that used to be the case, memory today is anything but
standard. With the onslaught of new technologies, it has been a challenge
for integrators and OEMs to maintain true compatibility. To that end,
Ventura Technology Group (VTG) is a tremendous resource, since we design and
manufacture the highest quality server class memory. While our products are
compatible among a wide range of platforms (SUN, IBM, HP, APPLE, COMPAQ,
DELL, HP, IBM, INTEL, TYAN), our focus will always be on higher density
modules (128MB, 256MB, 512MB, 1GB). Rounding out vital service offerings,
VTG stretches ahead of the competition in the following areas:
Every module design is manufactured to exceed OEM specifications
* Certifications with major OEM's (Tyan, Intel, Supermicro) insures constant
compatibility
* All memory technologies are manufactured in our own ISO9001 certified
facility
* All memory modules are manufactured using top grade DRAM and SDRAM
(Samsung, Micron, NEC, etc...)
* International and Domestic supplier relationships insure competitive
pricing and consistent supply
* Lifetime warranty with the benefit of cross shipment where needed
Along with all of the above benefits, VTG continually provides
competitive prices."
- Aduva
- Aduva sells a completely commercial (non-open source) toolset that
can be used to manage mixed, nasty, heterogeneous distributions in a
topdown fashion on a LAN. It would also "work" to manage a cluster in a
manner similar to yum. I would tend to recommend against it in
most cluster or LAN environments, as the primary value it adds is the
ability to cope with and centrally manage versioning on what amount to
bad cluster designs, and there are better ways to do that (e.g. don't
use a bad cluster design -- use e.g. kickstart/yum, scyld,
clustermatic). However, in certain relatively poorly managed corporate
environments it might make sense, so I'll include the link -- it does
give a single decent administrator at least some of the means to manage
and control a nasty environment with lots of mediocre administrators.
- Mountain
View Data
- Mountain View Data is a company that focuses on data delivery
systems in high availibility clusters, primarily in the commercial
world. They make tools like NAS operating systems, real-time backup
solutions, and a "snapshot" tool for capturing an instant image of
critical data (perhaps during a crash). However, they asked to be added
because of their Power Cockpit tool.
This tool, like Aduva's above, seems to enable systems installation and
image management in a cluster (or client/server LAN) environment. To
quote George Sun in his request for inclusion on this page:
Powercockpit can deploy and update complete OS and solution
stack(apps) to bare metal servers on a complete blank hard drive from a
custom repository of saved images to multiple hardware configurations
instantly using multicasting. Also, re-purpose servers on the fly
throughout your local and remote clusters. Don't waste time on tedious
cluster management. Powercockpit is more user friendly and robust than
using Kickstart or System Imager. Although they are useful solutions
they lack: GUI, cluster mgmt support, no rpm control, no cluster status
ability, no multicasting and require tedious hands-on support and
effort.
The primary strength of this tool appears to be how it encapsulates
a lot of expertise and hides in behind a GUI. As was the case with
Aduva's product, I suspect that this product will find a warm reception
in a relatively undermanaged corporate environment, but that most
University and Government shops will prefer to work with lower level,
fully open source tools and "roll their own" cluster, so to speak.
Still, worth looking over. They offer a free 30 day demo at the link
above, so it is easy to try it out.
- ibutton
- An ibutton is a tiny sensor packaging that can be read
electronically via a "1 wire interface" (basically a single twisted pair
with signal and ground). It can hold lots of things -- a crypt key in
ROM, a thermal sensor, account information -- and be detached from it
read/write station and carried around. It can be powered by a battery
for remote operation or vamp power off of the signal line. They
therefore have, in principle, lots of possible uses in the cluster or
LAN environment -- a portable absolutely unique digital ID that can be
used as a login or key for various resources, as thermal sensors
in a cluster room, and more. There is, however, a catch. They don't
tend to come with the wiring already done so you can just plug them into
a serial port or parallel port (they'll run off of either one) and
forget them, with GPL driver software all wrapped in a tarball and ready
to fly. So you MIGHT have to do some interface wiring (they do have a
handy technical spec sheet to guide you) and build the not QUITE so
plug'n'play gcc/userial-based application to be able to read from them.
Still, there are clearly beowulf-list people who have done so as
ibuttons are sometimes mentioned, so they belong here.
- Sensorsoft
- For those who prefer their solutions cooked (and will pay a bit
extra for them) there is Sensorsoft, who makes server room monitoring
hardware with both interface and software. This is a bit more
plug-n-play including linux and bsd software. The bad news is that
Sensorsoft is yet another hardware company that thinks it is "selling
software" for its own devices, so it has silly licensing restrictions on
its non-open source drivers (as far as I can tell without buying one,
anyway). However, they do have a thermal sensor only for less than $100
that has an RJ45 interface (and an RJ45-RS232 cable), and they do
provide technical data that should make it easy to write the drivers
required to read from it. In fact, they are almost certainly the
same drivers used in the ibutton, as all these guys are almost
certainly using the same thermal sensor chip (from Dallas
Semiconductor). I actually have four of these chips at home, together
with a recipe for building my own RS232 "sensor", but haven't had the
time to build one yet. That's the REAL hobbyist way...
- Pico Technologies
- Yes, there is one more possible solution to thermal and
environmental modeling at a fairly cheap price: Pico Tech. Pico makes a
leetle jobbie that plugs into either a USB port or serial port, snitches
power from it, and drives a variety of plug-in sensors from it. They
have thermal, they have humidity, they have combined, they have general
purpose ADC and scope products (which can drive fairly arbitrary
voltage-out detectors, up to and including oscilloscope-type stuff).
Truthfully, this looks like a good place to bookmark for its
possible applications in e.g. physics labs as well as in HPC cluster
rooms. They have linux-read open C source drivers for reading the
devices, and apparently porting their output into various forms is
pretty straightforward. Prices (when I looked -- you should obviously
check again) seemed to be less than $200 to equip a three-sensor plug
that should be able to monitor an entire server room -- chilled air
temp at the output ducts, heated air temp at the return duct, ambient
air out where the air is mixed. Alternative configurations add humidity
sensor information or monitor e.g. door openings and closings with a
standard door sensor.
One of these, a TV card, and an X10, and for less than $300 you have
the equivalent of a far more costly Netbotz appliance, presuming that
you have a node or system in your server room with a bus slot and a tiny
bit of attention to spare polling the device(s).
- Kill A
Watt
- Actually, this is a link to EFI.org, one of many folks that resell
the kill-a-watt power sensor. The kill-a-watt is an awesome device and
(in my opinion) essential to cluster engineering, where power
management is of paramount importance. Useful at home or the office,
too. This cheap little device (about $45 shipped from EFI) gives you
instant readouts of line voltage, line current, and rms power drawn by
whatever you plug into it. It can even monitor long term energy usage
(kilowatt-hours) for things powered through it (up to 1875 VA). I
wouldn't, actually recommend running a cluster through it all the time,
but plugging in those prototype nodes to get a clear picture of their
power draw under various loads and conditions during the design phase,
ah, that's worth its weight in gold!
- Mirus
International
- Mirus International makes "harmonic mitigating transformers" for
use in computer server, or cluster rooms. The switching power
supplies of computers have an unusual property -- they draw current only
in the middle third of each half-cycle of the voltage. This is no
problem when running one or two systems on a circuit, but is a
big problem when running a circuit close to capacity. There is a
FAQ on the Mirus website that clearly explains everything. This is a
must read for cluster architects -- even if you opt not to use a
mitigating transformer, the problems described in the FAQ really
do occur and can cost you a signficant amount of downtime, hardware
failures, and turn you bald and wrinkled before your time. A glimpse at
my picture should convince you -- it happened to us.
- Direct TextBook
- To quote from their website: "Direct Textbook is the fastest
textbook and book price comparison site on the internet. We help you
find the lowest prices on books before you buy! Our price search
software compares dozens of discount book stores, used books stores,
wholesale books stores and other online book stores to find the lowest
price available." This is relevant to beowulfery as they focus on texts
and technical books. In fact, Sterling's book "Beowulf Cluster
Computing with Linux" was the first hit on a search on the word
"beowulf", with "How to Build a Beowulf" by Sterling, Salmon, Becker and
Savarese the third hit. Complete with instant price checks -- you can
select the "best" online vendor to buy from. Not bad, actually.
Feel free to send requests for inclusion to my email address below
and to send non-monetary bribes to Robert G. Brown at 3209 Annandale
Road, Durham NC 27705, USA. Monetary bribes will have to wait until I
get my bank account in the Caymans opened...;-)