Many of you will remember the fanfare and bravado surrounding Sun Microsystems' Sep. 2004 announcement of a $1 per hour per processor utility computing plan. What you won't remember is Sun revealing a single customer using the service. That's because it hasn't. More than one year since it first started hyping the "pay-for-use grid computing services" Sun is still weeks away from presenting a customer to the public. The program has proved much tougher to sell that Sun ever imagined.
Post a Comment
Member since:
---
Member since:---
Its mostly the business model and part technology.
1) Companies can build their own grids or clusters.
2) If a company needs those kinds of resources; then they probably already have them. The only time it might be used is if they don't have the resources at a particular time to complete said job.
3) Most companies do not want to have (potentially) sensitive data shipped onto someones grid and then shipped back to them. The exposure to risk is to great for some companies. Something similar happens to Metro-GigE circuits. ALOT of companies don't want to be on shared pipes for various (sometimes perceived) security reasons. Most companies will opt for a dedicated GigE circuit between locations rather than depend on shared networks.
FYI: I hear some pundits saying what about FR/ATM they are shared and those sell like hot cakes, not to mention for gig-e you have VLANS, etc. Well, I will leave that for another day and another debate.
PS: off to work... enjoy all...
PSS: Shared computing resources within a company can be good; shared resources with external companies can be a secuirty risk. Or it might be that no one has a need.
Member since:---
Also, I think its ahead of its time (again!!!) by maybe a year or two, not more than three.
Companies may also fear upsetting Microsoft if they participate. The day when Microsoft's fortunes look really down is when companies will admit publicly and it will be en masse. Remember, the rats and a sinking ship..
Member since:2005-07-06
Member since:---
Member since:2005-07-26
Member since:---
Not sure what this has to do with story but since you asked.
M core N way threaded processors can use any instruction set, however it makes more sense to use a special purpose ISA to get cost down and performance up. The Sparc & Mips ISAs and others are ameniable to that but I am not sure x86 would make any sense. You end up with lots of slower threads with far greater total throughput than theoretical single threaded designs but each thread on paper can look fairly slow. On the other hand for memory bound apps, x86 performance is way overstated and 1 always waiting slow thread that keeps missing on caches may well be no faster than 1 of many threads in a properly designed MTA. It really all depends on the memory system.
And emulating x86 on Niagara sounds like a really bad idea but its been done before Alpha FX etc and can work. FWIW if I wanted hundreds of HW threads, I absolutely wouldn't want x86 compatibility since there is nothing worth bringing over that can exploit vast concurrency. Highly threaded processors need new (actually 20yrs old) ways of doing things.
transputer guy
RE[3]: Is grid technology hyped?
Member since:
---
Looking at the stats at http://www.grid.org/stats/ we see that the total CPU time is 440,069 years and some change. Sun would charge over 160 billion dollars for something volunteers did for free 
)
Member since:
2005-07-06
Member since:---
That's exactly what I was thinking! I mean it might be a good deal for some small companies but they are not likely to require a lot of computational power.
On the other hand if you have the system running 24/7 you're at 24x30=720$ using their system. That's more than it would cost to buy the thing!
1$/hr sounds good but when you do the math...
Member since:
---
Member since:
---
Member since:
2005-07-06
when the article and the ads first sprouted up on slashdot, i was on teh phone with sun's reps because it seemed to solve a lot of problems for our small architecture firm. (off-site backup, cheaper than buying a rendering machine of our own). at the time, the service wasn't available, and it took 4 sales reps to find that out.
now they've actually launched the service, but with no fanfare at all, since they blew their wad a year ago.
the other thing that seems to be a problem is that you need a solaris machine to connect to the grid. fine for off-site backup, not so good for renderings.
Member since:
---
Member since:2005-07-06
.........I think utility computing is a terrific idea but Sun's attempt to commercialize Globus is about 5-10 years ahead of it's time.
I beg to differ i don't see it coming anytime soon unless there is a huge leap in AI and only for very specialized businesses. Its the consumer market where the money is to be made. When you can have a terabyte of storage in your PC for less then $400 and a CPU more powerful then large computers 20-30 years old you don't have much of a need for large centralized processing.
Member since:2005-07-06
I beg to differ; the most expensive part of the equation when looking at this grid computing is the bandwidth; just look at the costs of bandwidth in NZ for example; its a bloody fortune, little wonder the uptake of highspeed connections by business is so low, and those that do have it, strinctly control what it is used for.
The idea is right, the serviced used to deliver SUN"s utility computing is still too expensive - for it to happen, they need to team up with Telcom's and offer it to businesses, along with the telcom - then the telecom will be able to cut the initial monthly costs in favour of using VAS to boost thei underlying profits.
Member since:
---
Here's a link to a FAQ on basic Sun grid info:
http://www.sun.com/emrkt/campaign_docs/expertexchange/knowledge/sol...
Member since:---
Member since:---
Member since:
---
Member since:
---
Member since:
2005-07-06
It's technical and marketing. On the one hand, it's apparently still in beta as they work out issues. On the other you simply have the infrastructure demands to make it usable. Anything that's going to need a lot of processors to crunch on, odds are, is going to need a lot of data to crunch.
There are a lot of issues for provisioning to make this practical as well, ensuring software exists, licensing issues, version numbers, etc.
What it sounds like is that they've been trying to find a high profile customer to launch the site with "See, Chrysler uses our tech..." or whatever, but have found it's easier to sell containers filled with Galaxy servers instead (for assorted reasons).
It's really a shame though, because I think there is potential for this. I think there could be a large VAR/reseller opportunity here (to, for example, help out the previous posters Architecture firm).
As much as we'd like $1/cpu/hr to be drag and drop, I think there will be more to that upfront. And also, while computing is expensive, it's still cheap enough to either go underpowered, or pay for idle servers. The idea of the Sun Grid is simply pay for the scale you use, particularly if you have spiky demands rather than have a room filled with rack heaters half the time or wait twice as long for your results when you really need them.
But, they have their foot in to whatever market there may be (if any) and they're in good position to expand it.
"Results? You want results? I have results. I know thousands of things that don't work!" Sun knows this now too :-).
Member since:
---
Member since:
---
Member since:
---
Sun's grid computing sounds a look a startup company I worked for in the late 1990's called Tilion run by Chris Stone. We had developed software that collected information from companies, converted it to XML format, and then processed it and returned the results. In the end the sofware was not the problem. The XML format wasn't the problem. It came down to the fact that large companies didn't want to send out secret information to be processed and returned. I thought this was kind of paranoid at first but it is actually a real problem for companies to feel comfortable with outside resources. There is also good reason for these security concerns. I worked as a consultant for a large IT company in the state of MA and I was suprised to find out one of the servers had been rooted one day. If that had happened to our startup with a major company like Ford Motors it would have been the instant death of the company. No one would send use work.
Member since:
---
Member since:2005-07-06
Member since:2005-08-19
Yeah. Almost everyone can afford clusters now. Buy few ultra-cheap PCs, collect some unneeded from friends or even use your older hardware which is somewhere in the basement. Buy ethernet cable, few switches and get it running 
I'm going to set up cluster some day just for fun and recompiling linux kernel 
Member since:
---
I work at an investment bank, I can assure OSNews readers that Sun are selling their grid quite well - both storage and compute. This is a major market which has been tapped well, the risk calculations we perform at EOD are tremendous and growing every day.
I fail to see how the article is 'news'
Member since:
2005-07-09
I think it could also be a matter of not enough exposure of grid computing for just 'regular' IT. I am sure there are a lot of researches that are using grid computing like the calculation of primes, seti, folding and various bioinformatics mining projects. I dont know who is using Sun's grid but yes there are definitely advantages to this technology. But it seems to me that the processing required for most jobs is just not enough to warrant paying a 3rd party to use their grid. Home grown solutions like a bunch of Xserves, Linux Beowulf clusters are just cheaper.
Member since:
---
