Should BOINC be viewed as a provider of cloud services?
log in

Advanced search

Message boards : Cafe : Should BOINC be viewed as a provider of cloud services?

Author Message
Profile jay
Send message
Joined: 12 Jul 14
Posts: 13
Credit: 51,444
RAC: 0
Message 1878 - Posted: 3 Feb 2015, 21:25:28 UTC

Greetings.

I would like your opinions....

In another post, Ben wrote that he has little choice about using the VINA tasks.
I understand his thoughts and VINA is fine with me.

But, the manner in which another project uses VitualBox irritated me.
Then, I sat back and asked why.....

With BOINC, I am a volunteer.
I do my best to find worthwhile projects by looking into the project's website that explains what it is doing.

I like BOINC, because it can limit the internet utilization and the CPU utilization and it allows me to prioritize projects.

Sometimes.

Some of the CERN projects use Virtual Box.
Virtual box does its own I/O on the internet and does not follow the
restrictions the user places through BOINC.
Virtual Box can run multiple tasks, (I believe sequentially) as a single WU.

What I find irritating is that project's work is presented to scientists as a cloud service.
The BOINC volunteer has little or no information of what the work is doing, other than it is a job for the CERN scientists.

I can't quite define why the virtual box approach is so irritating to me.

Perhaps it is my ego.

I might be spoiled by the transparency provided by World Community Grid.

I would like your opinions/discussion.

Questions:
-What do you think about using Virtual box and BOINC as a cloud service?
-Is that asking too much of volunteers?
-How much should a project tell their volunteers about the end results and the methods?

I welcome your responses.
Jay

Aurel
Send message
Joined: 10 Feb 13
Posts: 84
Credit: 275,734
RAC: 0
Message 1892 - Posted: 4 Feb 2015, 17:57:24 UTC - in response to Message 1878.

Hello Jay,

so much questions...

I´m an BOINC and CERN volunteer. I reached the 4. place by the CERN public computing challenge and tested some VM´s. [I coded some parts from the CMS offline software, too.]

An VM is good, but notice it will need memory. Mostly Linux is used in an box (container), e.g. scientific linux. Vina, the main application, runs under Linux and Windows, but under Linux I saw some bugs (e.g. wu progress bar).

An VM for this project is useless. The CERN uses VM´s for an simple reason:

Only when all tasks been computed on the the same OS you will got no errors. If different OS and bit-system are used, you will get many errors.

The reason why we don´t support use distributed computing, yet. We need to find a way. It´s to expensive to run an wourld-wide computing grid!
Quote from ATLAS-director on "Ideen-Expo" 2012 in Hannover

The CERN uses one of the biggest computing grid on the world; I´ll set an EDMS-Link to it. [if it´s open]



Open Science is an dream. Some projects uses open databases, like this. http://chemware.ucd.ie/FiNDaH/Malaria.php
I think it´s good to tell the user what you are making, see NumberFields@Home or climateprediction. [or just CERN]

Profile Ben
Project administrator
Project developer
Project tester
Project scientist
Send message
Joined: 17 Nov 14
Posts: 316
Credit: 1
RAC: 0
Message 1907 - Posted: 6 Feb 2015, 9:36:23 UTC - in response to Message 1892.

VM is good and easy. You don't need to build your binary for each OS, that's a good point.

I also like when it's open, you can understand and you can contribute. But running in a closed box is OK if you explain why and how.

Open Science is not a dream, but the way is long ^^.
The European Bioinformatic Institute is doing mainly Open Science (ChEMBL, ChEBI, Ensembl, ...)

(e.g. wu progress bar)


It's not a bug :(. I don't know how to do that because of libboost... (I'll check again today)

yo2013
Send message
Joined: 24 Dec 14
Posts: 11
Credit: 302,346
RAC: 0
Message 1926 - Posted: 12 Feb 2015, 17:44:31 UTC

Well, Virtualbox doesn't work on my machines (at least not with BOINC), so I can't say much about it. That's why I participate in Sixtrack but not in ATLAS or vLHC.

As for the more general question of informing the users and let the users choose, I usually don't participate in umbrella projects that don't let the user choose the subprojects to participate in. For example, Einstein@Home lets you choose what applications to run but (AFAIK) Folding@Home doesn't.

Also, I only participate on projects that use free software and that publish all of their research on open access journals (or at least provide free preprints).

Open science is not a dream, it's the only way to do science. If you don't let others know your work, we will be allways reinventing the wheel.

Profile Ben
Project administrator
Project developer
Project tester
Project scientist
Send message
Joined: 17 Nov 14
Posts: 316
Credit: 1
RAC: 0
Message 1927 - Posted: 12 Feb 2015, 18:36:10 UTC - in response to Message 1926.

Open science is not a dream, it's the only way to do science. If you don't let others know your work, we will be always reinventing the wheel.


True. Sharing is what is the most powerful. But some stupid people dinosaurs don't understand that.

Look at the evolution of the GNU/Linux Kernel, that's amazing how it's fast.

Message boards : Cafe : Should BOINC be viewed as a provider of cloud services?


Main page · Your account · Message boards


Copyright © 2017 Dr Anthony Chubb