> To add another two cents worth because I forgot this bit;
>
> Seti and Boinc are many things to many people. The project is not wholly bad
> and its aims are admirable in many ways however please nobody forget that it
> is just a business, providing salaries and funding. Would you like to carry on
> donating your funding after this?
>
> Neil

Could you provide some evidence to back up these claims? Any funds that come to the university through this are a matter of public record. You might be correct that this project is providing a few lecture topics and one or two graduate students with projects but this is NOT a cash cow for Berkeley. They have gotten a few NSF grants and some funding from the Planetary Society but I'm pretty sure that has all gone straight into operations costs just pay the electric and ISP bills. Using 60 Mb/sec 24/7 is NOT cheap! I believe ALL positions on the seti@home team are purely volunteer-based. They might have 1 person on salary, I'm not entirely sure. They are in this for the science. Are you?

I would also like to point out that estimating server load is not always an exact science. Going from a couple hundred beta testers to tens of thousands of users can introduce new bottlenecks where you least anticipate them. Add to that the fact that they had a bad stick of RAM, and you get major problems that are beyond your control. It seems we are all used to the (relatively) smooth operation of seti@home-classic. This smooth operation was only achieved after months and even a couple years of problems and errors much like the ones we are experiencing now. Sure I like to keep my computers busy but for crying out loud... the world will not end if they sit idle for a day or two - or a week for that matter - while some bugs are worked out! I'm with Andrew Waddington on his "chill pill" comment above.

Toby

Your criticism is probably correct. I have made my assumption based on the the mentality of universities in the uk. There would be no possibility of instigating a project such as Seti over here unless it was fully costed with cash flows and the ability for the university to feel that research money could flow in whilst scientific papers could flow out.

If I seem bitter and twisted then I apologise and my comments were not meant as a slur however this is still a shambles.

Neil

Just to play devil's advocate here Toby, I'm going to comment on a couple things...

"I believe ALL positions on the seti@home team are purely volunteer-based. They might have 1 person on salary, I'm not entirely sure. They are in this for the science. Are you?"
___________________

I believe all the dev team is paid. No, the S@H project doesn't make money as you mentioned, but the key people draw salaries. You couldn't really expect otherwise. A project of this magnitude is more than could be dealt with by volunteers working in their spare time. Full time programmers would be a necessity.
___________________
"I would also like to point out that estimating server load is not always an exact science. Going from a couple hundred beta testers to tens of thousands of users can introduce new bottlenecks where you least anticipate them."
___________________

There are/were approximately 18,000 beta testers. That's a bit different than hundreds. I haven't heard any numbers on how many users the project currently has, but I think it's naive to say berkeley didn't know what to expect.

I believe they simply made a bad decision to release before they were ready. I have no idea what the impetus for the decision was, but it's been rumored that berkeley themselves didn't know they were going to go public until a day before the release. They didn't even announce it to beta until after it happened! I believe they didn't tell beta in order to avoid the criticism, because I don't think there's a single beta tester who felt the software was ready for public consumption.

So you can try put a positive spin on it and tender excuses for the problems, but I think they simply made a mistake. And that mistake is currently costing them dearly. I think they've left a sour taste in many mouths, and as a result the project may never achieve what it could have been.

But what's done is done, and we can't change that. All we can do is continue on.

> Just to play devil's advocate here Toby, I'm going to comment on a couple
> things...

Please! :)

You are correct. As I said, I wasn't sure about the paid positions. I have since checked with someone who knows more about the situation in Berkeley than I do. There are indeed several paid positions. However they are most likely funded through grants (I know BOINC got a couple from the NSF) and most certainly not 'high paying'. As I said... They are in it for the science.

> There are/were approximately 18,000 beta testers. That's a bit different
> than hundreds. I haven't heard any numbers on how many users the project
> currently has, but I think it's naive to say berkeley didn't know what to
> expect.

My 'hundreds' was definitely wrong. I was a beta tester for several months, however I only participated with part of my computing power. The rest I left running seti-classic. Once the project went live, my drain on the server resources probably tripled or even quadrupled. I suspect many beta testers may have done the same. What you would really have to look at is the number of active hosts in beta vs. final.

I will agree with you that the release was probably a bit pre-mature and executed in a hurry. I think it is in my nature to give people the benefit of the doubt so I think they probably did at least a few predictions and thought they could handle it for at least the first week or two, allowing them to slowly integrate more hardware and software optimizations. In hindsight this was obviously a bad call. Call me an optimist I guess... I do know that the current 'final' release of the project DID undergo some testing for several days before its release to the public. I know a few of the people who participated in this test. This was probably just a small test to make sure they didn't have any obvious glaring errors and not a stress test.

My last 2 cents for the day. I think the birds are starting to chirp outside - a sure sign that it is far past my bed time :)

> I believe all the dev team is paid. No, the S@H project doesn't make money as
> you mentioned, but the key people draw salaries. You couldn't really expect
> otherwise. A project of this magnitude is more than could be dealt with by
> volunteers working in their spare time. Full time programmers would be a
> necessity.

Just want to point out that a large part of the people who helped making this possible aren't getting paid for it. Many have volounteered to do work for free - simply because the idea of BOINC is interesting.

> make better software and Hardware Things
> and start the projekt all things running better
> and have more network power for a worldwide projekt
> its shit to run a worldwide projekt like a 386 with 4 mb ram and a 2400 baud
> modem you know what i mean
> greetings guido
> Sorry for my bad english
>

Once the dust settles I'd be interested to know how many BOINC users there are compared to classic SETI. I suspect a lot of people will have just given up. I also suspect there will be less processing power donated.

A member of my team would have given up long ago had it not been for the support of the team - it took her 6 days just to create an account. Some members of my team who have access to large numbers of PCs are refusing to make the switch because they don't believe that BOINC is anything other than Beta, nor do they believe that BOINC is suitable for running on machines in a commerical environment.

You also have to remember that BOINC is part of a wider community - i.e. distributed processing. How much damage has BOINC done to the popularity and image of such projects?

It seems that a suitable infrastructure for live BOINC wasn't in place and the existing infrastructure has no flexibility. These are things you can plan for even though it's not an exact science. Prima facia it seems that American HE is as bad as UK HE when it comes to essential skills and expertise such as project management.


j

There are/were approximately 18,000 beta testers. That's a bit different than hundreds. I haven't heard any numbers on how many users the project currently has, but I think it's naive to say berkeley didn't know what to expect.
----------

Yes, I agree with that. With close to 5 Million Seti@home Classic users did Berkeley think that only 10,000 of them would want to run BOINC Seti@home...D'oh

They should have known before hand that the Server side of the Project was simply not going to handle the load & should have prepared for it before hand instead of going Live and letting everybody sink with the ship.

I can't imagine the untold Thousands if not millions of Seti@home Classic users who have tasted BOINC and spit it out & won't be back...

I'm chilling out and going back to "classic". Give me a call when BOING is ready!
And I joined the project December 2 (16:42UTC) 2001, not November 30.
...but I'm still chilling!!

Everybody needs to pull their finger out and chill.
Everybody can or has the option to crunch the Classic version.
Everybody in the BOINC team is doing their best.

So what is the problem? I just dont understand.

Alex

German Board is online:

visit: www.boinc-forum.de

German Board is online:

visit: www.boinc-forum.de

> Yes, I agree with that. With close to 5 Million Seti@home Classic users did
> Berkeley think that only 10,000 of them would want to run BOINC
> Seti@home...D'oh
>
> They should have known before hand that the Server side of the Project was
> simply not going to handle the load & should have prepared for it before
> hand instead of going Live and letting everybody sink with the ship.
>

This is what bothers me most about the BOINC roll-out. I'm willing to give a project team a break when minor glitches crop up from time to time, but this entire project has been so poorly managed I should make it a case study in my Project Management Professional class.

Given the past roll-out troubles with SETI Classic, how on EARTH could they not have predicted similar troubles with BOINC, assessed the risk and been prepared? To so massively miscalculate the required infrastructure resources, for example, is simply astonishing.

I'd like to comment on the idea of whether or not the BOINC project team is getting paid for their work or it's voluntary - it's completely irrelevant. If you commit yourself to a project, then you should be professional enough to do it right, regardless if you're getting paid or not. If you can't give 100% of your time to a project, that's fine - that happens all the time, even with salaried team members - but then the project scope needs to be adjusted to reflect that on a more realistic basis. I mean, who among us would have really cared if they pushed the release back a month or more to work out these issues?

Whatever the real reasons are for the problems with this roll-out, I expected more from these people. It's really disappointing that it's worked out this way. And, before someone tells me to chill... I am chilled. Disappointed, but chilled.

-G-

I myself was a long time beta tester. I don't think so much the problem lies with the client software, but the fault lies with the server prep work.....

No units to distribute, no load tests performed, ect....

When it went live, the seti site was on v. 3.19, while the Beta site was still using 3.18.....that kind of made me chuckle right there......lmfao

I just don't think they had the wet work ready to go live, and someone jumped the gun....

> This is what bothers me most about the BOINC roll-out. I'm willing to give a
> project team a break when minor glitches crop up from time to time, but this
> entire project has been so poorly managed I should make it a case study in my
> Project Management Professional class.

I wish you could. In the software development field we see, still today, no learning from past mistakes. On one porject I worked on I did a post-project "lesssons Learned" and of Steve McConnell's list of mistakes we only missed out on two. Things like this are repeated so often you have to wonder ...

> Given the past roll-out troubles with SETI Classic, how on EARTH could they
> not have predicted similar troubles with BOINC, assessed the risk and been
> prepared? To so massively miscalculate the required infrastructure resources,
> for example, is simply astonishing.

Worse, they had the "old" data and oould have made a transistion of the data in to the new database and performed a load test. The "good" news is that the database is mostly used as a "flat-file" repository and all of the queries I looked at (I will admit very late in the Beta, just a couple weeks ago, is it days only, oh well) did not do joins. But for the life of me I cannot see how they could have made the other mistakes in the database architecture.

The most obvious one is right in the database constraints, 7 indexes on the "Results" table (8 if you count the PK). On any table that has a high update rate that is, um, un-good ...

The second is the use of "Count(*)" on the results table instead of making that specific count an attribute of the participant, team, etc. and updating (increment & decrement) counts in that (smaller) table instead of calculating them on-the-fly on the "Results" table.

The third is the use of "Select *" in all queries I looked at. I grant you, I only looked at about 20-30, but, that is enough. The worst of it is that many of the tables have BLOBs that are not going to be used in the presentation, so now you have done a fetch of the row, all of the blobs, shipped it to the client and then not used the data. So, you have consumed lots of I/O on both ends of the database. And as can be seen while working on the web site the consequences.

> Whatever the real reasons are for the problems with this roll-out, I expected
> more from these people. It's really disappointing that it's worked out this
> way. And, before someone tells me to chill... I am chilled. Disappointed,
> but chilled.

The reason a deployment like this is called "big bang" is for the sound the project makes when it hits the floor. And like you, I am disappointed too ...

> With close to 5 Million Seti@home Classic users did Berkeley think that only
> 10,000 of them would want to run BOINC Seti@home...D'oh

You should know, that only a handfull of those 5 million are actually active users crunching WU's. Most of them don't as far as i know.

--J

COOL DOWN THE HOT HEADS--possible the Server helps a little Quake ;-) to run better
like a good cocktail mix it babe ;-)
[/url] [/url]
another OS like NASA SYSTEM 5 or bsd will follow

Might as well put in my two cents :)

I have been running BOINC for over six months without a hitch. The problem that many have referred to as BOINC is actually not the BOINC software, but a problem with the Seti@Home project servers, as others have mentioned. I believe this to be true, because I have been running an Alpha version of Predictor@Home ( http://predictor.scripps.edu/ ) on BOINC with very little problems. No problems with the BOINC client software. Their project has not been affected at all during this S@H outage, except to upgrade the server-side software, which seems to have went well. The point is, other projects are able to use BOINC, and S@H has been having bad luck.

The BOINC software project was tested, in a selected group of Alpha testers. These people were chosen for their ability to provide useful, constructive feedback. I, for one, think Beta group was left out, because there was very little constructive feedback provided by that group, especially in recent months. Alpha was the first to know about this site, and the "live" project. The project may have "slipped" into the mainstream due to an innocent web surfer that happened upon the website and made a post on a German forum about it. When that happened, a storm of people hit the website, which was at the time only intended for Alpha test users.

It is your choice to be here, and people will probably abandon the project due to the situation it is currently in. BOINC is a very interesting project to me, and I will stay here until it is gone. If you have interest in the project, but are turned off by the problems it has had, I urge you to at the very least, come back to the project later, after the major switchover issues have been worked out.

Guess it worked out to a little more than $0.02. :)
***********************************************
Join US-Distributed

[url=http://www.us-distributed.com]

Being a Debian GNU/Linux geek, my tolerance for buggy software which is released too soon is very low. Interesting that boinc, which I believe was born on a *nix puter, can't properly extract benchmarking perameters from a fellow *nix puter, report its timezone correctly or keep proper track of its credit stats.

The Berkeley software team has done a most outstanding job thus far but I'm afraid boinc/seti@home would rate a barely passing grade. All this trouble could have been avoided if they hadn't been in such a hurry to release before their software/hardware were ready.

But, I've been crunchin' for five years and I'm not about to quit now. I'll just keep switching between classic and boink until all the problems are solved.

Well, this kinda' turned into a pissin' & moanin' thread, so I'll vent here, too.

IMO, the s/w is reasonably decent for an R1, especially when taking into account the nature of the project (DC, University run, non-profit, non-commercial, freeware, etc., etc.).

BUT the infrastrucure COMPLETELY sucks, w/a capital 'C'!

There are two possible reasons, AFAIC:

1] Gross Incompetence

2] Arrogant Disregard

I highly suspect #2.

Unfortunately, I really like the science behind SETI, and believe its purpose to be compelling, important, and noble.

And there's the rub...

My $0.02.

Strat