Infrastructure information


log in

Advanced search

Message boards : Number crunching : Infrastructure information

Author Message
Profile huns0004
Send message
Joined: 14 Jun 01
Posts: 46
Credit: 1,848,591
RAC: 454
United States
Message 13701 - Posted: 4 Aug 2004, 14:03:08 UTC

Is there anybody out there who understands what the BOINC infrastructure is? Is there a list, are there threads, is there a website? This is BOINC, after all, and in the very name is Open Infrastructure.

I see a lot of terms bandied about, like validator, splitter, and so forth, and although I know what these mean for the most part, as yet, I don't truly grok them. An overview of how things are set up would be nice, including hardware, software, and so forth. People who are involved in SETI, especially those of use masochistic enough to be involved with BOINC, are generally interested in these techy things, so I don't think this is too much to ask.

Knowing this stuff would help. Information is to complaining what water is to fire.

Of course, this could be bad, if we find out that WUs are being split on a TI-85, which is connected to an Access database running on Windows 3.1 by a floppy sneakernet, while the web site is running on a Coleco ADAM connected to the world with a 300 baud acoustic coupler modem. But even so, at least we would know what was going on.

Note to Berkeley: we're geeks. We want to help. Let us in on your secrets.

Ken Phillips m0mcw
Volunteer tester
Avatar
Send message
Joined: 2 Feb 00
Posts: 267
Credit: 415,678
RAC: 0
United Kingdom
Message 13704 - Posted: 4 Aug 2004, 14:13:45 UTC - in response to Message 13701.


> Of course, this could be bad, if we find out that WUs are being split on a
> TI-85, which is connected to an Access database running on Windows 3.1 by a
> floppy sneakernet, while the web site is running on a Coleco ADAM connected to
> the world with a 300 baud acoustic coupler modem. But even so, at least we
> would know what was going on.
>
> Note to Berkeley: we're geeks. We want to help. Let us in on your secrets.
>
LOL

Is the definition of being old, that I can remember this equipment, and similiar, new, for sale in shops, even possessed some of it? It's certainly a blast from the past, which bit of the project do you think would be run by a sinclair 16k zx spectrum, now that was sheer raw power :-0

rotfl

Ken Phillips


MikeW
Send message
Joined: 7 Apr 04
Posts: 71
Credit: 10,406
RAC: 0
United Kingdom
Message 13705 - Posted: 4 Aug 2004, 14:15:12 UTC

Try http://boinc.berkeley.edu



Giskard the first telepathic robot.

Profile huns0004
Send message
Joined: 14 Jun 01
Posts: 46
Credit: 1,848,591
RAC: 454
United States
Message 13720 - Posted: 4 Aug 2004, 15:19:42 UTC - in response to Message 13705.

> Try http://boinc.berkeley.edu

I did, and there is absolutely no information there about what hardware, software, database schemas, network capabilities, and so on the SETI BOINC project is using.

Profile Toby
Volunteer tester
Avatar
Send message
Joined: 26 Oct 00
Posts: 1005
Credit: 5,622,795
RAC: 2
United States
Message 13741 - Posted: 4 Aug 2004, 18:12:10 UTC

Well the source code is there. This is what the Open Infrastructure means. It is open source so anyone can look at it, tweak it and use it as they see fit. Granted, that will only help those of us who can actually read and understand C++/php/SQL/python but the information is there :)

Also, you need to distinguish between BOINC and seti@home. The splitter, for example, is specific to the seti@home project and has nothing to do with BOINC. You can get a lot of information about seti@home fromt the seti-classic site. (http://setiathome.ssl.berkeley.edu ) The basic setup of the work unit creation and data analysis hasn't changed with BOINC. Just the underlying system that takes the information and distributes it to us.




-------------------------------------------
- A member of The Knights Who Say NI!
Possibly the best stats site in the universe:
http://boinc-kwsn.no-ip.info

Profile huns0004
Send message
Joined: 14 Jun 01
Posts: 46
Credit: 1,848,591
RAC: 454
United States
Message 13745 - Posted: 4 Aug 2004, 18:33:31 UTC - in response to Message 13741.

> Well the source code is there. This is what the Open Infrastructure means.
> It is open source so anyone can look at it, tweak it and use it as they see
> fit. Granted, that will only help those of us who can actually read and
> understand C++/php/SQL/python but the information is there :)
>
> Also, you need to distinguish between BOINC and seti@home. The splitter, for
> example, is specific to the seti@home project and has nothing to do with
> BOINC. You can get a lot of information about seti@home fromt the
> seti-classic site. (<a> href="http://setiathome.ssl.berkeley.edu">http://setiathome.ssl.berkeley.edu[/url]
> ) The basic setup of the work unit creation and data analysis hasn't changed
> with BOINC. Just the underlying system that takes the information and
> distributes it to us.

I understand that the Open Infrastructure part refers to the source code for BOINC, not necessarily anything else, but in keeping with the whole open source idea, I still think it is nice to know the whole setup.

I already know the basics about how SETI works. As you say in your post, though, BOINC changes the underlying system that distributes things to us. Given that the entire problem with BOINC at this stage is that "underlying system," that's what I am trying to find information about.

I've been using SETI for over three years, so I'm not a newbie who hasn't a clue. I simply want to know what they are using and why it's broken, if for no other reason than so I can learn from their mistakes.

Profile KWSN - MajorKong
Volunteer tester
Avatar
Send message
Joined: 5 Jan 00
Posts: 1406
Credit: 793,366
RAC: 0
United States
Message 13779 - Posted: 4 Aug 2004, 22:12:17 UTC - in response to Message 13745.
Last modified: 4 Aug 2004, 22:18:42 UTC

>
> I understand that the Open Infrastructure part refers to the source code for
> BOINC, not necessarily anything else, but in keeping with the whole open
> source idea, I still think it is nice to know the whole setup.
>

As Toby said, the source code is available for download. In your initial post, you asked about certain terms such as 'validator'. These are various software processes in the BOINC (or the Seti@Home) system. With BOINC, these processes are flexible as to where they 'run'. They can all run on one machine (such as what the Pirates@Home people are doing), several of them can run on one machine, and the rest on others, each one can have its own machine, and in some cases (ie. the 'splitter', for instance) several copies can run on several different machines. From what you ask on your post, in addition to the 'software' setup, you wish to know the hardware setup as well. This information is not very easy to come up with, and is frequently changing. In addition to moving various processes from machine to machine (to better utilize existing hardware capacity), hardware fails and must be replaced. Also, there are occasional 'new bits' of hardware added (such as the upcoming Snap Appliance for the database disk storage -- it should HELP a LOT!). The hardware setup is fairly dynamic, and it would take quite a bit of effort to keep it current. Maybe once the project has settled down....

One thing about the hardware setup that I HAVE heard is that most (not sure if its all) of the hardware runs the Solaris OS from Sun. One can infer from this that a good deal of the equipment is from Sun. I have no other details, and this is ALL I am anywhere NEAR confident about saying about their setup. I know its not much.

> I already know the basics about how SETI works. As you say in your post,
> though, BOINC changes the underlying system that distributes things to us.
> Given that the entire problem with BOINC at this stage is that "underlying
> system," that's what I am trying to find information about.
>
> I've been using SETI for over three years, so I'm not a newbie who hasn't a
> clue. I simply want to know what they are using and why it's broken, if for
> no other reason than so I can learn from their mistakes.
>

Per my understanding of the system, the 'what is broken' is not so much the BOINC server software (it works, or WOULD if given the chance) but its two things:

1. Seriously UNDERPOWERED machines they are using for the BOINC/S@H project servers (S@H's GOOD stuff is currently still powering the S@H-Classic project and will likely remain so until S@H-Classic goes offline -- which I HOPE will be SOON). This much the BOINC/S@H staff and BOINC developers have admitted to.

2. My opinion in comparing the performance right before the last outage to the current performance is that the BOINC/S@H project has suffered a nasty hardware failure, and they have had to 'double up' on some things. From the behavior of both the website and the clients (and a slightly cryptic statement from Matt Lebofsky over on the S@H-Classic forums), I would guess that the failure had something to do with the database data storage, thereby putting a cramp on the DB read and/or write speed. This is only a guess.

I am sorry, but this is the best information that I have (and I pay CLOSE attention to the information that *IS* released by S@H staff and developers). Sadly, it is, in my opinion, highly improbbable that the S@H staff and developers will be forthcoming with anything more detailed.
------------
KWSN-MajorKong
KWSN Forum Admin (retired)
http://www.kwsnforum.com

BOINC Beta tester

Profile Michael Foerster
Send message
Joined: 16 May 99
Posts: 10
Credit: 276,250
RAC: 0
Israel
Message 13790 - Posted: 4 Aug 2004, 23:06:08 UTC - in response to Message 13701.

> Of course, this could be bad, if we find out that WUs are being split on a
> TI-85, which is connected to an Access database running on Windows 3.1 by a
> floppy sneakernet, while the web site is running on a Coleco ADAM connected to
> the world with a 300 baud acoustic coupler modem. But even so, at least we
> would know what was going on.
>
> Note to Berkeley: we're geeks. We want to help. Let us in on your secrets.

I know it's not exactly what you're looking for but does seem to indicate SETI is a bit more up-to-date... :-)

from http://news.zdnet.co.uk/internet/0,39020369,39118617,00.htm

"Seti@home already uses more than ten of Sun's enterprise servers and 70 workstations, running Solaris"


<a> [/url]

Message boards : Number crunching : Infrastructure information

Copyright © 2014 University of California