The cost of starting a BOINC project.
13 Nov 2010 16:36:28 UTC
Topic 195450
(moderation:
How much money does it cost to start up a BOINC project from scratch? Getting the servers and other equipment plus electricity and internet. Software and everything else.
What total costs would it be? Even ball park figures and speculation would be appreciated.
The cost of starting a BOINC project.
)
At the low end you can rent a VPS for as little as $10/month, or even use a dynamic dns service to run it from a home computer. After that it depends on what your server load and bandwidth costs are. If you're not up to admining a unix server yourself there are managed VPS options available for as little as $25/mo but I suspect the cheapest ones have a fairly narrow definition of what software they support and that the boinc server might not qualify (probably comes down to if it includes components outside of php scripts and the mysql database).
RE: How much money does it
)
I guess the dominant factors for the cost is how many users will be served by the project and what the traffic will be.
If we are talking about (say) a demo project (e.g. a school project) that will connect only a few clients, the costs wrt money (not talking about the time it takes to learn, set up and configure stuff) is the money you have to spend for a single PC and the electricity to run it. Not a big deal at all.
But if you make it a public project and you give a lot of credits to crunchers, you'll soon be flooded with new users and it will be getting sort of expensive :-)
Maybe if you could give a few more details about the project, it would be easier to come up with an estimation.
CU
HBE
Hi Bikeman, say about the
)
Hi Bikeman, say about the size of a medium to large project, eg.,Cosmology, Milkyway, Orbit, something along those sizes. I'm not thinking of starting my very own BOINC but I've been looking at a project that has been stalled for over 5 years and I'm curious about how much money it needs to be kick started into life. Is such a large sum of money required the reason that project is stalled or is it other reasons? Can you guess which BOINC project I'm referring to?
RE: Hi Bikeman, say about
)
I only follow a couple of projects, so I don't have a clue really which project you are referring to, and if I knew I'd be even less inclined to speculate on wh it is in whatever state it is now. There might be several reasons other than financial that slow down progress for a science BOINC project, like
* difficulties to port the science code to the platform that is most widely used by users (Windows)
* difficulties to adapt the algorithms so that work can be divided among thousands of PCs that cannot communicate to each other during the computation
* difficulties to adapt the algorithms so that work on the individual PCs require only a modest amount of RAM and finish within a reasonable time
From the perspective of the science organization that would be hosting the project, the focus might not be cost but actually the amount of money *saved* by using BOINC. If the problem can be solved more cost-efficiently on more conventional computer resources readily available to the institution in question, why bother using BOINC?
To get an idea about the workload and cost of setting up a big BOINC project, you can have a look at a presentation of a talk that Prof Allen gave to the 3rd Boinc workshop in 2007. Needless to say you will have to scale down the costs a bit account for falling prices for CPU power and disk storage.
http://boinc.berkeley.edu/ws_07/Bruce%20Allen%20-%20Einstein@home.ppt
CU
HBE
CU
Bikeman
RE: Hi Bikeman, say about
)
Recently some projects that have been begging for money have said they need 80 Euros per month just for the internet costs. One place you might ask is places like FreeHal, it is a small project and run by only one guy, I think, in Europe. Collatz is another smaller project that is both a gpu and cpu project run by one guy in the US. Another might be wuprop, it is small and collects some interesting data.
One thing you haven't talked about is what do you do with the data after you collect it? Say you want to start a project, what is the point of the crunching? And what do you do with the data after it is crunched and then compiled?
Is it possible to limit the
)
Is it possible to limit the participants of such a project to the members of an instition like firm or a resaerch institut? There maybe also some 100 PCs available. I would believe, that this can be managed by 1 person. But setting this up ???
Martin
RE: Is it possible to limit
)
I think so because when a project is in the the Aplha or Beta stage they can limit the number of people they let 'join' the project.
I would think once it is set up and running then yes one person could do it. This would be NOT counting the tech support for each pc, just the Boinc project, but yes one person should be able to handle 100 pc's in a Boinc project.
[entirely my personal
)
[entirely my personal viewpoint]
One issue I take especial note of is who/what 'owns' the results? Has one furthered (wo)mankind's store of commonly viewable knowledge, or maybe given someone a cheap ride to a personal patent bonanza? There are real immediate costs to donating computing resources to a BOINC/DC project, and collectively can be well worth donning the mantle of 'important research' to induce/obtain. In my personal opinion ( unrelated to my mod role here ) you have to publicly answer these leading issues in explicit & attributable detail before people will substantially commit.
For example : being too busy or unavailable to reply is inappropriate. Here is a post on a project's message board ( my codename BLOB ) that I won't name and shame, which has left the content of this user's question unanswered for well over 3 months now ( there's been replies but no substantiative answers, and nothing whatever from the project's leader ) :
NB As a moderator here, I would normally with-hold adverse comment on other projects for the sake of ( the appearance of ) propriety. But I'll break that rule with the obvious caveats you can see in this post, as I stumbled on this the other day and I think it's worth a mention as to how not to go about it.
[\entirely my personal viewpoint]
Cheers, Mike.
I have made this letter longer than usual because I lack the time to make it shorter ...
... and my other CPU is a Ryzen 5950X :-) Blaise Pascal
RE: [entirely my personal
)
This is a VERY good point Mike and one that ALOT of us crunchers wonder about all the time! One other issue that has come up for the different projects is when the papers are published do they tell how the data was gathered? ie do they say that Boinc was used? If more projects WOULD do that then Boinc would get a bigger following and more people would be crunchers!
To emphasize your point Seti got its cooper wire stolen right out of the ground one time, years ago now, the project had access to the internet thru other means but the project was down until the wire was replaced. At NO TIME did Seti post anything on its webpage about the problems or what was going on, yes other people asked MANY times what the problem was, but the project itself was mum on the scope or reasons for the outage! Only AFTER they were back up did anyone say anything and that was only AFTER the community badgered them into it! Apparently it was in the local Berkeley paper and that is how the word got out. MANY times over the years Seti has gone down and their explanation afterwards for no postings on the subject was always been 'we can post or we can fix, which would you prefer'. To me that is dumb and counter productive to running a good project.
RE: This is a .........
)
Well, point taken, though I don't want to start a general theme of 'why I don't like project X' [ except in the case of E@H, as these boards are precisely where such issues should be aired! :-) ] but wanted to demonstrate a very specific point about not answering the question of 'who is to benefit from my donation?' This was something I picked up on when first enrolling nearly five years ago, and my profile still contains my original comment on that. I'm pretty sure that this question is uppermost in many contributor's mind when they are making the value judgment of whom to gift their resources. Not answering such a crucial question, I reckon, is the quickest way to drain contributor confidence.
Thus FWIW I mention this for ( the other ) Mike's sake so any outlay won't be wasted by no one coming along to contribute - the cost of not answering that question. One really can't allow the impression to be gained that one is inducing others to plump up one's personal finances, at whatever level of indirection.
[aside]
I take your point that one cost, of the rather more difficult to quantify personal time and effort variety, is that of maintaining good public relations. I will praise the management of E@H in this regard, as they've taken many opportunities to thank and attribute the contributions of crunchers - for their invaluable assistance in helping to obtain results - in scientific publishing, at professional conferences, and in the general media. I've read more than a few technical papers that put E@H right up front in the description and certainly not merely footnoted, and well before the E@H/Aricebo pulsar co-discovery too. So it's been a policy in place for a long time.
[/aside]
Cheers, Mike.
I have made this letter longer than usual because I lack the time to make it shorter ...
... and my other CPU is a Ryzen 5950X :-) Blaise Pascal