[Beowulf] 10GbE Adapter Market

Prentice Bisbal prentice.bisbal at rutgers.edu
Mon Nov 25 11:56:39 PST 2013


I hope I'm not too late to this party to add my 2 cents!

While I don't dispute the legal definition of COTS as stand for 
"commercial off the-shelf", I think most people, especially in the Linux 
cluster community are more familiar with COTS meaning "COMMODITY 
off-the-shelf."

That is a subtle but important distinction, and it's that distinction 
that lead to the growth of Linux clusters. Googling for the definition 
of commodity, there are several definitions. Using the definition I am 
most familiar with, "commodity" means standardized goods that are 
essentially interchangeable, available from multiple vendors, and sell 
at relatively low margins dues to cheap production through economies of 
scale. Brand is usually irrelevant. Using 'define:commodity' in google 
(Thanks for the tip, Ellis!) , this definition comes up:

"a raw material or primary agricultural product that can be bought and 
sold, such as copper or coffee."

Many manufactured goods fit this description too: steel, gasoline, 
tulips, cotton fabric, nails, lumber, memory chips, etc. With respect to 
computers, the x86 Wintel desktop computers were often considered 
commodity items since the components were standardized and you could 
(within reason) swap components between Dell, HP, Gateway (remember 
them?), and 'brand-x" computers with (almost) no issues. This 
commoditization lead to the sub-$1,000 computer in the 90s, and 
eventually the sub-$500 computer (anyone remember e-Machines?).

It is from these cheap "COTS" (where C=commodity) systems that the first 
Linux clusters were born and provided a tremendous cost savings over 
other "COTS" (where C = Commercial) supercomputers. For examples of the 
latter, there were plenty of systems by Cray, SGI, Sun and others that 
were "commercial", but definitely not "commodity" systems. Most of these 
systems no longer exist thanks to Linux cluster built from 'commodity' 
x86 hardware.

So you see, while the US govt. might define the C in COTS as 
"commercial", it was when that C stood for "commodity" that Linux 
clusters were able to take off.

Prentice Bisbal
Manager of Information Technology
Rutgers Discovery Informatics Institute (RDI2)
Rutgers University
http://rdi2.rutgers.edu

On 11/19/2013 09:26 AM, Bob Drzyzgula wrote:
> Well, it is certainly the case that very little high-end product is 
> sitting around on shelves. As someone who has placed his share of 
> orders with five to seven figure bottom lines, I can attest to the 
> fact that it is perfectly typical to have to wait for your place in 
> the production queue to come up before you ever see your stuff. 
> However, that actually isn't the point of COTS, at least as it is 
> defined in the FAR. Rather, the point is that the product was 
> engineered by the manufacturer on their own initiative, with the 
> intent of selling it to the general public. It doesn't even matter if, 
> for example, the product is Configure-to-Order, assembled out of COTS 
> parts as specified by customer, it is still COTS, because that 
> ordering and configuration process is set up for sales to the general 
> public, on the manufacturer's initiative.
>
> Something is *not* COTS if the customer tells the manufacturer what to 
> build, and how to build it, to meet a unique need of the customer, and 
> the manufacturer then has no expectation that there would be any 
> market for it beyond that one customer, or possibly even that they 
> will be granted no rights to sell it to the general public. This 
> process is where those mil-spec hammers and toilet seats came from, 
> and why they were so expensive. The process is far from dead, however 
> -- you can be certain that the military is still ordering stuff -- 
> radios for example -- that are built to unique and even classified 
> specifications. It's just that, under a lot of scrutiny, the 
> government eventually came to its senses and figured out that maybe 
> the hammers that Eastwing sells at the corner hardware store will work 
> just fine, or at least that the procedures that require the use of a 
> hammer can adjusted, at minimal cost, such that those will work. Then 
> again, from another perspective it is perhaps the case that this only 
> became possible when the quality control standards used in commercial 
> manufacturing processes finally rose to a level that one actually 
> *could* depend on a COTS product when the life of a soldier or 
> astronaut depended on it.
>
> But from this perspective, very little computing equipment in use 
> today is anything but COTS. Perhaps some of the giant web services 
> companies are specifying custom computing devices, but I'd guess that 
> the volumes involved there are high enough that distinction becomes 
> meaningless. It is almost as if the Beowulf community taught the 
> industry and the market something that they more or less learned.
>
> Perhaps the bigger remaining distinction, from an HPC perspective, is 
> between devices built around merchant silicon vs custom or captive 
> designs. While there is still a market for devices built around Power 
> and SPARC processors they are certainly not the focus of HPC cluster 
> computing (I note that less than 10% of the current Top 500 use 
> anything but Intel or AMD CPUs; Power/PowerPC accounts for most of 
> those). Network equipment has remained something of a bastion for 
> products built around custom ASICs, but even that is starting to 
> crumble as designs built around the latest merchant chips such as the 
> Broadcom Trident II, together with protocols such as SPB & TRILL, 
> become competitive with big-iron chassis switch configurations.
>
> When I think about the "SOHO" designation, though, I am thinking more 
> in terms of the commoditization of trailing-edge technology. Take a 
> meander down to MicroCenter and you'll find that "SOHO" network 
> switches are still being sold in high volumes with 100 Mbps ports; 
> only the premium models have gigabit ports. As a practical matter, a 
> 100 Mbps switch is perfectly suited to the task of connecting the half 
> dozen workstations of an insurance agent's office to a couple of 
> printers and a 25 megabit Comcast Internet link. This is the kind of 
> application that those magazine writers were thinking of when they 
> coined the term, and it is only incidental that some of that equipment 
> will occasionally be useful to on-a-shoestring level HPC.
>
> --Bob
>
>
> On Mon, Nov 18, 2013 at 10:49 PM, Mark Hahn <hahn at mcmaster.ca 
> <mailto:hahn at mcmaster.ca>> wrote:
>
>
>     interesting stuff about GSA lists.
>
>     > A Netgear 16-port gigabit switch that sells for $200 is both
>     SOHO and COTS.
>     > A Cisco Nexus 7718 18-slot chassis switch is still COTS but in
>     no way,
>     > shape, manner or form SOHO.
>
>     I like to treat COTS as more than merely "not bespoke",
>     but really pret a porter, and the Target end of it too ;)
>
>     that is, I suspect that there are not a bunch of $30k cisco switches
>     on the shelf of the big distributors, and especially not in any
>     local resellers or retail outlets.  the point is really that a COTS
>     device is produced in large volume, with low margins.
>     also, to some degree with conforming to public standards
>     and being available from multiple vendors.
>
>     for me, beowulf is all about doing supercomputer work with COTS
>     hardware.
>
>     regards, mark hahn.
>     _______________________________________________
>     Beowulf mailing list, Beowulf at beowulf.org
>     <mailto:Beowulf at beowulf.org> sponsored by Penguin Computing
>     To change your subscription (digest mode or unsubscribe) visit
>     http://www.beowulf.org/mailman/listinfo/beowulf
>
>
>
>
> _______________________________________________
> Beowulf mailing list, Beowulf at beowulf.org sponsored by Penguin Computing
> To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.beowulf.org/pipermail/beowulf/attachments/20131125/85294459/attachment.html>


More information about the Beowulf mailing list