Vote using thread (+) or (–) on the thread rating.

Disregarding the current problems and so forth answer this question.

With the hardware and software available today, is a project the size and scope of the anticipated Boinc/Seti achievable?

(+) = yes
(-) = no

Thanks for voting!

---

Boinc....Boinc....Boinc....Boinc....

Vote using thread (+) or (–) on the thread rating.

Disregarding the current problems and so forth answer this question.

With the hardware and software available today, is a project the size and scope of the anticipated Boinc/Seti achievable?

(+) = yes
(-) = no

Thanks for voting!

yes they evently need seporate machines for each prosess instead of share like they do now and in the future may need more than one each for upload download validate and delete.

---

Psssttt! Don't look now, but a project like this is achievable because we're already doing it!

---

+ -

---

- +

---

:P

Vote using thread (+) or (–) on the thread rating.

Disregarding the current problems and so forth answer this question.

With the hardware and software available today, is a project the size and scope of the anticipated Boinc/Seti achievable?

(+) = yes
(-) = no

Thanks for voting!

For what's available in general today (HW and people), yes (+)
For what's available at UCB today (HW and people), no (-)

There are still a boatload of people running classic that don't read any seti forums that will switch over when the "notification" (email or thru the classic cruncher) goes out, or maybe when they stop getting classic work, and they check the website.

As the URL that many are used to now point's to seti/boinc, it might be a good thing to have some note on the front page to welcome classic users, and let then know why they are now pointed at a new website. (not a news item that can scroll off), but a notice on the site itself) While the "notification" might also include this info, there will be some that don't get the notification, and hit the site when then can no longer upload/download work for classic.

They're still recovering from a 5 hour outage days ago, and have a much longer outage planned starting this weekend. This will be interesting. I myself have seti/boinc turned off right now for all systems.

BTW, Mr. David Knittle - the word " anticipated" was used in the question!

---

Sorry, az... I don't anticipate what I'm already doing. ;)

If you're aware of a distinction between what the BOINC SETI@home project is now and what it is *anticipated* to be, please enlighten us. I think if you ask 10 people what they anticipate from this project and you're bound to get a dozen different answers. Trying to measure up to such speculation is rather pointless...

---

Sorry, az... I don't anticipate what I'm already doing. ;)

If you're aware of a distinction between what the BOINC SETI@home project is now and what it is *anticipated* to be, please enlighten us. I think if you ask 10 people what they anticipate from this project and you're bound to get a dozzen different answers. Trying to measure up to such speculation is rather pointless...

Simple... Only about 1/2 of the classic crunchers have moved to seti/boinc. The question WAS about anticipated load!

Darn.. You got the pompom to far in your ear that it damaged some brain cells!

Even if the current HW could handle the current load (without shutting down services for days to recover from a 5 hour outage), seti/boinc is not handling even 1/2 the load of classic. (and it seems that adding servers might help little, if the bottleneck is the interface bwtween the servers! (100mb)

The classic seti HW if all switched to seti/boinc over night, probably couldn't handle the same number of users as classic. The backend (server) system with seti/boinc seem to be much more complex than classic, and will require more CPU/MEMORY/DISK/IO than classic.

---

Odd. I'm able perform work on this project perfectly well enough. Evidently you cannot. Who's the one with the brain damage?

---

Odd. I'm able perform work on this project perfectly well enough. Evidently you cannot. Who's the one with the brain damage?

You are.. When was the last time the validator granted you credits? Why would a 5 hour outage cause this kind of a scramble by UCB if the current system could handle the "current load", much less the "anticipated load"?

You look interesting in that cheerleader outfit, but I'd sugest that you get a bikini wax if you keep getting your panties in a bunch!

Remember you are are one that missed the "antisipated" part of the question, and now you're are doing your best to obsure that part of the question, as you can't admit you were wrong!

You messed up and either let it drop, or admit it!

---

With the hardware and software available today, is a project the size and scope of the anticipated Boinc/Seti achievable?

When you say "available today" do you mean the hardware they already own, or hardware that is available to be bought...

If you mean "what they already own" I dont think it's going to be up to the job in the future, as it is already not up to it in the present. :'-(

FYI The "anticipated" load is about 1 million work units per day, regardless of the number of users. I THINK we are actually approaching this number right now but I can't say for sure. But yes, it is very possible to achieve using current technology although certainly not cheap. Currently there is also the issue of the server closet being full. Seti classic shutting down will help with that :)

---

A member of The Knights Who Say NI!
For rankings, history graphs and more, check out:
My BOINC stats site

Have faith, they will get it right. If you dont think that...WHY are you here?

---

Not real clear that shutting down classic will actually help.

That is, while that will certainly free up additional hardware for SETI BOINC, it may also redirect the still quite large population of SETI classic workstations to SETI BOINC.

The thing is, with the validation process, running BOINC in more resource intensive -- both in terms of CPU utilization and system I/O. Lot's of cross checking, lots of units going inbound and outbound and (seemingly) more complex code.

Then again, I don't know that BOINC is the issue -- as there are other BOINC projects (admittedly currently only about 1/4 the host load each as SETI) which are running essentially rock solid and have been for months.

As it is, the SETI folks have been struggling with a batch of issues, hardware, software, connectivity and electrical which no doubt have been causing a fair amount of frustration -- not only for the SETI BOINC user population, but also for the the folks at the lab.

The current validator outage seems to be to have at least the *scent* of 'let's try this, it might work' about it rather than a case of an 'ah ha!' solution about it.

Also, a few weeks ago, Berkeley posted a notice about the difficulty in keeping up with requests for work -- if the remaining active SETI classic folks were to switch over, then that issue also would be a problem here.

And of course in addition to these issues which are 'can we keep up with the *current* workload' issues, there are the 'rifle shots' of hardware failures, electrical issues, connectivity issues which in their own way add to the keep up with the workload issues by generating surges.

So, I guess at this point, given the resources and environment that Berkeley is working with, I am NOT sanguine about SETI BOINC being able to keep up with a future increase in workload (heck they have more than enough problems keeping up with the current workload).

FYI The "anticipated" load is about 1 million work units per day, regardless of the number of users. I THINK we are actually approaching this number right now but I can't say for sure. But yes, it is very possible to achieve using current technology although certainly not cheap. Currently there is also the issue of the server closet being full. Seti classic shutting down will help with that :)

---

Have faith, they will get it right. If you dont think that...WHY are you here?

One of the major reasons as to why I'm here is support for Distributed Computing (DC).

Computers and computing is one of the greatest achievments of the 20th Century.
Seti@Home has led to the implementation of BOINC, one of the leading DC platforms.
Far too many of Humanity's resources are being wasted or lost.
It would be a terrible shame to not use the available resource of idle cpu cycles for the benefit of all humanity.

We are still in the early days of DC, and like all scientific projects, it will continue to be developed and improved.

I want to do my (little) bit to help.

:)

---

Join TeamACC

Sometimes I think we are alone in the universe, and sometimes I think we are not. In either case the idea is quite staggering.

Have faith, they will get it right. If you dont think that...WHY are you here?

I am here to try to help find signs of intelligent life out there among the stars.

As I see it part of the problem the project at large and the UCB people in particular are experiencing is that too much is expected of the old, outdated equipment they're trying to run this project on. Stop and think for a minute. Look at the server status page. Now think about your computer. They're running 300-440MHz processors. What are you running?

Do not think for a minute that I am faulting the UCB SETI personnel. Quite the opposite is true. I think they are making every effort humanly possible to solve the problems that exist. The problem is that there's only so much that can be expected from the current equipment.

My answer to the thread's topic:

For what's available in general today (HW), yes (+)
For what's available at UCB today (HW), no (-)

A little info from the sun website:

Sun E3500

The Sun Enterprise 3500 Server has been retired and is no longer orderable.

Sun plans to take orders for new Sun Enterprise 3500 to 6500 servers through June 2002. Sun plans to take orders for key components for these servers through June 2004. Sun plans to offer support services for these servers through June 2007.

Sun U60

GA Date:
June 1998

EOL:
Last Order: July 2002
Last Shipment: October 2002

Sun U10

GA Date:
January 1998 [original]
February 1998 [Ultra 10S]
April 1998 [Ultra 10 with Elite3D-m3]
November 1998 [Model 333]
November 1998 [Model 360]
May 1999 [Model 440]

EOL:
Last Order: August 2002
Last Shipment: November 2002

Sun E450

GA Date:
September 1997

EOL:
Last Order: August 2002
Last Shipment: November 2002

Sun D220R

Not found on the Sun site. The closest match I could find was the
Sun Enterprise[tm] 220R Server*

GA Date:
November 1999

EOL:
Last Order: August 2002
Last Shipment: November 2002

*If these are what's in use, a number of problems with these servers are documented on the Sun website.

---

Nothing travels faster than the speed of light with the possible exception of bad news, which obeys its own special laws.
Douglas Adams (1952 - 2001)

Stop and think for a minute. Look at the server status page. Now think about your computer. They're running 300-440MHz processors. What are you running?

You're comparing the wrong fruits.
What are you doing with your computer and what do you think a server is doing? Do you think these servers are crunching? Do you think they only use one of their CPUs (those that have 2 to 6)? What kind of CPU are these servers running, compare that to your CPU...

These servers don't run your 'run of the mill' x86 CPUs, you know? Their architecture is build upon RISC, not your own CPU which is designed on CISC. Compare the SPARC CPUs used to what your computer is capable of. Not so bad, huh? ;)

---

Ugh, not this rancid cisc/risc argument again. It made sense twenty years ago when the Patterson (?) wrote the famous CA text book, but reality moves on. Today, x86's are basically risc cores surrounded by a bunch of backwards compatibility circuits and cache that provide superior performance to the widest possible marketplace. I haven't looked for a few years, but I don't think there is a pure risc processor left in the production.

Yet don't take my word for it, look at the market place. Sun, unfortunately or not, has progressively retrenched as a company under the withering competition of comparable or better x86 based servers, both in price and performance. Perhaps a Celeron running Linux is not the best server (maybe it is?!) but well designed x86 servers seem to be eating Sun's lunch.

Nevertheless, this thread was to address whether boinc/seti is ready for the big time. I can't ignore recent events, because they expose the issues the next scale-up presents, and I have to conclude boincSeti is not ready. Command central has repeatedly indicated they are resource limited, and seem to be proving it, even at today's user base. Tripling the userbase will be 'wild'. But additional hardware may not cure the ills, since command central has not demonstrated they understand how this new boinc works. That is, they seem to be turning on and off queues, rearranging network switches, and so forth without demonstrating actual improvement. So at this point, I would like to see them get new or more hardware, but I don't really think they are ready for an easy transition to the next phases (classicSeti shutdown, new astronomical data sources, more sophisticated data analysis, and so on).

---

May this Farce be with You

...With the hardware and software available today, is a project the size and scope of the anticipated Boinc/Seti achievable?...

Definitely yes, but not with this partly rotten concept. I don't want to go into details, but the concept simply has too many bottlenecks to handle the kind of load, that'll come, once SETI classic is closed.
I hate to be that kind of prophet and will get loads of '-' for that, but time will prove :P

---

Aloha, Uli