On Bandwidth (Jul 08 2009)

Message boards : Technical News : On Bandwidth (Jul 08 2009)
Message board moderation

To post messages, you must log in.

Previous · 1 · 2 · 3 · 4 · Next

AuthorMessage
John McLeod VII
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 15 Jul 99
Posts: 24806
Credit: 790,712
RAC: 0
United States
Message 915920 - Posted: 8 Jul 2009, 23:40:59 UTC - in response to Message 915906.  

Using DSL, you're lucky to get 8Mbit/s over 1km of phone line cable.

I wouldn't be so sure; I'm more than 1 click away from the local hub and I have 20Mbit already 9the modem/router can go as high as 100Mbit for internet). Bog standard telephone lines. The fiber lies next to it, but it's not in use yet.

But we aren't necessarily talking xDSL here, just data transfer. Just a (home) network and for that you can use old electricity cables, old telephone cables, old TV cables, whatever is around. See network Media in Home network for some examples and their maximum speeds.

As for the telephone wires and their maximum speed, it may not be 10Gbit, but 1Gbit is a possibility. Depends on the state of the old line, though and the amount of hops it takes. But otherwise.... ;-)

Jord,

What we're talking here isn't really signalling, it's radio that happens to run along wires.

... and as the frequency goes up, the wavelength goes down. Shorter wavelength means smaller antenna.

At 10 MHz, a wavelength is right at 30 meters, and a wire that is an odd multiple of 7.5 meters long is going to radiate (and receive) signals.

At 10 GHz, it's more like every 7.5 millimeters.

As the wavelength gets shorter, the tolerances in a good transmission line get smaller. You can tolerate bridge taps (on phone lines) at 10 MHz, but at 10 GHz every little bump is a place where signal gets reflected back instead of moving toward the receiver.

Lots of tricks to get around that, and occasionally, we get lucky and get a pristine wire pair and perfect conditions.

... but it's tough.

-- Ned

The higher the frequency, the tighter the twist that is required between the pairs to avoid attenuation. If I recallcorrectly, the twist has to be less than 1/4 wavelength.

With more complicated encoding, the number of bits per baud can go up (max I know about is 16 bits per baud).

That still leaves POTS maxing out well below 1Gb.


BOINC WIKI
ID: 915920 · Report as offensive
Profile ML1
Volunteer moderator
Volunteer tester

Send message
Joined: 25 Nov 01
Posts: 21227
Credit: 7,508,002
RAC: 20
United Kingdom
Message 915921 - Posted: 8 Jul 2009, 23:41:08 UTC - in response to Message 915898.  
Last modified: 8 Jul 2009, 23:42:34 UTC

Using DSL, you're lucky to get 8Mbit/s over 1km of phone line cable.

I wouldn't be so sure; I'm more than 1 click away from the local hub and I have 20Mbit already 9the modem/router can go as high as 100Mbit for internet). Bog standard telephone lines. The fiber lies next to it, but it's not in use yet. ...

Then, you're pretty lucky. The 100Mbit/s will just be the ethernet LAN side of the modem router. And as for the optical fibre... then you're VERY lucky :-) Why not yet in use?

From your table:

ITU-T G.hn  Electrical wiring/Telephone line/coaxial cable   up to 1 Gbit/s   N/A
HomePNA          Telephone line                                   10 Mbit/s      300 m


So... 1 Gbit/s but over how far?...


Using the existing grey fibre looks to be the best bet.

With DWDM, you can get 8 Tb/s down one fibre... Could the Berkeley servers keep up with that?

Keep searchin',
Martin
See new freedom: Mageia Linux
Take a look for yourself: Linux Format
The Future is what We all make IT (GPLv3)
ID: 915921 · Report as offensive
Profile HAL9000
Volunteer tester
Avatar

Send message
Joined: 11 Sep 99
Posts: 6534
Credit: 196,805,888
RAC: 57
United States
Message 915945 - Posted: 9 Jul 2009, 0:36:52 UTC - in response to Message 915800.  

Great job making everything work as it is.
It must be hard to steer the boat when you are busy plugging holes.

Do your current routers allow for on the fly configuration. So that during peak loads the connection could be switched from redundancy and then back once it calmed down? It would be hard to say if that would be useful or if you would just have 200mb maxed out all the time.

Could something be done on the client side. Perhaps a check to see if the connection is getting slammed and then instead of trying to up/download. It would instead wait a bit. Perhaps this is already done. However, From the details shown in the message log currently it seems like, for an upload, it's just trying to hand the server all the data if it is there or not.
SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[
ID: 915945 · Report as offensive
Sirius B Project Donor
Volunteer tester
Avatar

Send message
Joined: 26 Dec 00
Posts: 24912
Credit: 3,081,182
RAC: 7
Ireland
Message 915958 - Posted: 9 Jul 2009, 1:02:43 UTC - in response to Message 915855.  

No no no I didn't mean that at all. I'm not saying you're wasting electricity. We still haven't fully covered the Arecibo sky yet with multibeam observations, for example.

What I am saying is we don't need to adhere to a 24/7/365 uptime policy because of the fear that we'll lose our participants forever. I have enough professional experience to prove that if you want to win customers (or fans) back after a long hiatus, you simply ask them nicely, and they return. In fact I've proven this time and time again with the reminder e-mails. Others feel we need to keep the workflow happening at all costs or we're ruined. What do you think?

Basically all I meant was that we can afford to take more long hiatuses to take care of long standing sub projects (like final data analysis?!), and perhaps come up with new/improved SETI@home-esque projects - all of which would be far easier to do if we were not as preoccupied with keeping data flow happening 100% of the time.

- Matt



Thanks for both the posts Matt. I've been saying this on & off for 2 years. Whenever you get problems this bad, shut down, at least externally for a month. YOU won't lose any crunchers as the ones who say they'll leave will be back as it's too addictive.

As for the rest, we'll just continue crunching other projects until your return. It worked for other projects, so why not Seti? & it'll also take the pressure off you guys.
ID: 915958 · Report as offensive
Cosmic_Ocean
Avatar

Send message
Joined: 23 Dec 00
Posts: 3027
Credit: 13,516,867
RAC: 13
United States
Message 915962 - Posted: 9 Jul 2009, 1:10:53 UTC

I can agree with making the MB tasks 2-8 times longer, or if we don't really need those, just move over to AP. Sure those WUs are big, but they take significantly longer to crunch than MB. I believe the crunch time:file size ratio is higher for AP than MB, though it depends on the host, too.
Linux laptop:
record uptime: 1511d 20h 19m (ended due to the power brick giving-up)
ID: 915962 · Report as offensive
Pr1nce0f4mb3r

Send message
Joined: 3 Apr 99
Posts: 1
Credit: 79,298
RAC: 0
United States
Message 915972 - Posted: 9 Jul 2009, 1:23:58 UTC

ahhhh Thank you. This explains why I have not received a work unit in a few days. I'll just keep it on and if I get a work unit this system will be ready.

Also in regards to the email reminder you send out I thank you. Awhile ago I wasn't aware that my account wasn't active. Some glitch happened and it never reported the work unit and never got another one. I don't get on the computer that this runs on very often so it wasn't until your email that said come on back did I actually go and look to see why I would be getting it. It was happily crunching away the same unit for quite awhile. A reset and I was back in the game.

Keep up the great work.
ID: 915972 · Report as offensive
Profile HAL9000
Volunteer tester
Avatar

Send message
Joined: 11 Sep 99
Posts: 6534
Credit: 196,805,888
RAC: 57
United States
Message 916021 - Posted: 9 Jul 2009, 3:07:24 UTC - in response to Message 915958.  

No no no I didn't mean that at all. I'm not saying you're wasting electricity. We still haven't fully covered the Arecibo sky yet with multibeam observations, for example.

What I am saying is we don't need to adhere to a 24/7/365 uptime policy because of the fear that we'll lose our participants forever. I have enough professional experience to prove that if you want to win customers (or fans) back after a long hiatus, you simply ask them nicely, and they return. In fact I've proven this time and time again with the reminder e-mails. Others feel we need to keep the workflow happening at all costs or we're ruined. What do you think?

Basically all I meant was that we can afford to take more long hiatuses to take care of long standing sub projects (like final data analysis?!), and perhaps come up with new/improved SETI@home-esque projects - all of which would be far easier to do if we were not as preoccupied with keeping data flow happening 100% of the time.

- Matt



Thanks for both the posts Matt. I've been saying this on & off for 2 years. Whenever you get problems this bad, shut down, at least externally for a month. YOU won't lose any crunchers as the ones who say they'll leave will be back as it's too addictive.

As for the rest, we'll just continue crunching other projects until your return. It worked for other projects, so why not Seti? & it'll also take the pressure off you guys.


If they were to shutdown for a month to work on things. Wouldn't they need to plan that in advance for the wu's that would expire while out in the field?
SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[
ID: 916021 · Report as offensive
zpm
Volunteer tester
Avatar

Send message
Joined: 25 Apr 08
Posts: 284
Credit: 1,659,024
RAC: 0
United States
Message 916025 - Posted: 9 Jul 2009, 3:10:29 UTC - in response to Message 916021.  

they could leave the upload up.... just disable download...

I recommend Secunia PSI: http://secunia.com/vulnerability_scanning/personal/
Go Georgia Tech.
ID: 916025 · Report as offensive
Profile HAL9000
Volunteer tester
Avatar

Send message
Joined: 11 Sep 99
Posts: 6534
Credit: 196,805,888
RAC: 57
United States
Message 916029 - Posted: 9 Jul 2009, 3:12:04 UTC - in response to Message 916025.  

they could leave the upload up.... just disable download...


Hmmmm Yes. That would make sense.

I think that was done when SETI@Home Classic was winding down.
SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[
ID: 916029 · Report as offensive
John McLeod VII
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 15 Jul 99
Posts: 24806
Credit: 790,712
RAC: 0
United States
Message 916031 - Posted: 9 Jul 2009, 3:20:16 UTC - in response to Message 916029.  

they could leave the upload up.... just disable download...


Hmmmm Yes. That would make sense.

I think that was done when SETI@Home Classic was winding down.

Just turn off the splitters for a while...


BOINC WIKI
ID: 916031 · Report as offensive
Chris Campbell

Send message
Joined: 14 May 99
Posts: 17
Credit: 4,677,968
RAC: 2
United States
Message 916039 - Posted: 9 Jul 2009, 3:39:21 UTC
Last modified: 9 Jul 2009, 3:42:27 UTC

Hi all -

First, I'd like to thank Matt for tipping us off to the anniversary presentation videos that were posted(1) about a week ago. I finished watching them this morning.

Someone posted here (in another day's thread) requesting the powerpoint presentation files. Which I thought was a bit much to ask, until I actually saw the videos.

I am sure that all of the students involved in the process meant well, however the person who manned the camera during Eric K's and David A's presentations should be punished by ... being forced to watch his footage. We get to watch Eric K for 71 minutes while he points to his slides OFF SCREEN. Literally, for the entire 71 minutes, the only time you get to see a slide is when Eric waked in front of the projector. Which happened exactly once at the 35 minute mark.

Now, in the interest of empathy, I'm going to guess that perhaps this camera operator thought that there were going to be multiple cameras, or assumed that everyone would indeed have the presentation, or who knows. But the sad result is that we just don't have the meat of Eric's presentation, especially because Eric's style is to put up the slide and then comment further on it, without actually pointing out what's on the slide. Which is fair, we all have eyeballs and can see for ourselves, right?

Alas, we can't. Matt or whoever, please beg Eric for his presentation file (exported to PDF?) and post it alongside his video on the 10yr page.

Also, it would be awesome if someone found these videos on Youtube, so we could link to those instead and not suck more bandwidth (doh!) out of the S@H project. And it would be good to have both of these things done before the videos get properly mentioned on the home page.

Finally, I'd like to thank Dan W for mentioning Obama's fine speech to the National Academy of Sciences back in April. It's truly heartening to watch.(2)(3)

Thank you ALL so much for everything!

(1) http://setiathome.berkeley.edu/sah_10yr_anniversary.php
(2) http://www.nationalacademies.org/morenews/20090428.html
(3) http://edg1.vcall.com/video/nas/launch.asp
Running S@H since Day 1 -- May 14th, 1999!
ID: 916039 · Report as offensive
Profile Johnney Guinness
Volunteer tester
Avatar

Send message
Joined: 11 Sep 06
Posts: 3093
Credit: 2,652,287
RAC: 0
Ireland
Message 916053 - Posted: 9 Jul 2009, 4:26:08 UTC
Last modified: 9 Jul 2009, 4:47:49 UTC

Well at least its progress that the powers that be are starting to assess getting the gigabyte link up onto the hill.

The Mathematical Sciences Research Institute share the same building as SSL. Bet those guys would love more bandwidth.

Really you guys need a fiber connection to the top of the hill, no messing with copper wires or wireless stuff.

Here is a map of the cable route between SSL on the top of the hill and the Sather Campanile Tower down on the lower campus;




The route marked on the map between points A and B is 2.2 miles along the road. Its 1.2 miles as the crow flies.

2.2 miles = 3,500 meters of fiber cable

Type of cable needed; Fiber Optic Cable, 6-Strand, Singlemode, 8.3/125, Armored Outside Plant (OSP), Polyethylene (PE) Jacket, 1,640'


@ $600 per 500 Meters = $4,200 from cablesplususa.com

+ Labor for 3 Sparks to run the cable. Might be two days work depending on the route they take with the cable

Labor = About $1,500

Plus routers and fiber cable jointing kits, plus unforeseen stuff.

Guesstimate total needed approximately = $8,000 - $10,000 (If the Electrical contractor is willing to cut corners and use existing ducts and drains)
*************
There are many variables left out of this estimate due to lack of info and detail about the nearest Gigabyte link and street ducting and stuff.

John.
ID: 916053 · Report as offensive
Profile Gary Charpentier Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 25 Dec 00
Posts: 31009
Credit: 53,134,872
RAC: 32
United States
Message 916065 - Posted: 9 Jul 2009, 5:15:16 UTC - in response to Message 916053.  

Well at least its progress that the powers that be are starting to assess getting the gigabyte link up onto the hill.

The Mathematical Sciences Research Institute share the same building as SSL. Bet those guys would love more bandwidth.

Now there's a thought. Get another BOINC project up at the building. I'm sure there is a nice math theorem that a pile of computation can help prove just waiting for someone to start it up. See if that escalates the level on the ticket. Our luck it would half the bandwidth to SETI!

@elsethread
As to the videos they are on campus bandwidth and not SETI bandwidth or so says traceroute.
Interesting, 17 hops to the campus, 9 to SETI server from here. HE may be good choice.

Yes, I'd love to see the powerpoint's. Just hope the files got backed up.

ID: 916065 · Report as offensive
1mp0£173
Volunteer tester

Send message
Joined: 3 Apr 99
Posts: 8423
Credit: 356,897
RAC: 0
United States
Message 916076 - Posted: 9 Jul 2009, 6:07:11 UTC - in response to Message 916065.  

Well at least its progress that the powers that be are starting to assess getting the gigabyte link up onto the hill.

The Mathematical Sciences Research Institute share the same building as SSL. Bet those guys would love more bandwidth.

Now there's a thought. Get another BOINC project up at the building.

Not to be too cynical, but make sure the other project is very well funded, so they can pay for the fiber.
ID: 916076 · Report as offensive
Profile Jord
Volunteer tester
Avatar

Send message
Joined: 9 Jun 99
Posts: 15184
Credit: 4,362,181
RAC: 3
Netherlands
Message 916085 - Posted: 9 Jul 2009, 7:29:00 UTC - in response to Message 915906.  

I am just thinking outside the box. If for the moment it's unfeasible to lay down a new line, anything above 100Mbit would be great. If at that time an old line can be used, even only to be used when the 100Mbit line is starting to get clogged up, and in anticipation of the new line coming... It's a cheap solution. Even if the throughput is between 166Mbit or higher.

Now, we know that a phone line can be used, as Seti has done it before. Although perhaps not in the way I pointed out. Of course, with 'old phoneline/coax cable/electricity cable" I meant one that preferably isn't in use by anything else.

And while laying down fiber is nice, I am just assuming that they'll be taking the 100Mbit off line during that period, as it may be easier to lay the new line in an existing trench than it is to dig a whole new one through rocky terrain. ;-)

ML1 wrote:
Then, you're pretty lucky.

They've always said you needed the fiber for high-speed internet, but the only place that's used is going out of the local exchange, which lies about a 1,200 yards that way, as the crow flies. So not even as the telephone cables run. ;-)

The local junction box is next to the flat, which is where my pair of wires meets the rest of pairs of wires, before being bundled into a bigger cable that goes off to the local exchange. Distance from my modem to the junction? About 10 meters. With that I am lucky.
ID: 916085 · Report as offensive
djmotiska

Send message
Joined: 26 Jul 01
Posts: 20
Credit: 29,378,647
RAC: 105
Finland
Message 916098 - Posted: 9 Jul 2009, 9:04:21 UTC

While reading threads about this bandwidth issue, some questions arise. Why is that Hurricane link ending at the campus, why not in SSL? Wasn't it donated to Seti? Could the hardware at both ends of the current 100mbit fibre updated to gigabit? How about adding adding another link to campus, via a cable modem or something like that over for example a TV cabling, like suggested earlier?

I remember the time before the current Hurricane link and all the problems it had. Now all those quad core and CUDA crunchers have taken bandwidth requirements up a lot and bottleneck is now elsewhere. How about adding a daily break when scheduler is off and some small maintenance done, like DB purge or something? No new work is assigned then but pending downloads and uploads could be done at that short time. That could help a bit and keep DB size smaller. At the moment, there are 4,032,901 results out in the field, 3,083,494 results waiting for validation and 1,317,418 results waiting for DB purge. Big numbers, in my opinion.

In another thread I and someone else suggested using mirrors. Nothing like P2P or Torrent, but just normal servers outside SSL. Taking a laptop with external HDD filled with splitted data and uploaded to mirror server(s) via the existing gigabit link, of course from outside of the uncongested 100mbit link. Scheduler could send clients MD5 or CRC hashes to ensure the mirrored work is not corrupted or altered. Of course the simpliest way could be to add extra DL server to campus, where the gigabit link exists, until the current bandwidth bottleneck issue is solved.

I'm now doing my duty to help on this bandwidth issue by crunching on other projects and hope most of you are doing the same. I won't be abandoning Seti.
ID: 916098 · Report as offensive
Profile Gary Charpentier Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 25 Dec 00
Posts: 31009
Credit: 53,134,872
RAC: 32
United States
Message 916099 - Posted: 9 Jul 2009, 9:04:28 UTC - in response to Message 916085.  

I am just thinking outside the box. If for the moment it's unfeasible to lay down a new line, anything above 100Mbit would be great. If at that time an old line can be used, even only to be used when the 100Mbit line is starting to get clogged up, and in anticipation of the new line coming... It's a cheap solution. Even if the throughput is between 166Mbit or higher.

Now, we know that a phone line can be used, as Seti has done it before. Although perhaps not in the way I pointed out. Of course, with 'old phoneline/coax cable/electricity cable" I meant one that preferably isn't in use by anything else.

And while laying down fiber is nice, I am just assuming that they'll be taking the 100Mbit off line during that period, as it may be easier to lay the new line in an existing trench than it is to dig a whole new one through rocky terrain. ;-)

You are assuming that someone didn't have the foresight when they dug the original trench and laid the conduit(s) to not oversize for future expansion. Since trenching is expensive my guess is they did. But even if they didn't I'd guess today there is room because old telephone PBX lines of one hundred analog pairs has been replaced by a digital line freeing up physical room in the conduit. I'd also suspect that the cost of pulling a fiber up to SSL while the rest of the campus is being built out is pretty low compared to having to come back later and do it, so it wouldn't surprise me if it is in place but unused. Not really Matt's job to know though. Finally at 2+ miles I doubt they are on copper at all today.

I suspect the politics of the hangup is that once gigabit is in the building, the building will have to be re-cabled to get the gigabit everywhere and that may cost more than getting it to the building in the first place. SETI being in the closet wouldn't need that, but everyone else might feel slighted if only SETI had access to it.

ID: 916099 · Report as offensive
W-K 666 Project Donor
Volunteer tester

Send message
Joined: 18 May 99
Posts: 19401
Credit: 40,757,560
RAC: 67
United Kingdom
Message 916109 - Posted: 9 Jul 2009, 10:06:38 UTC

If it is political, that there isn't better comms to the building, maybe we need to shame the 'powers that be' into action.
A few thousand e-mails from the us users might wake them up. Or ask the students to have a demonstration on our behalf.
ID: 916109 · Report as offensive
Profile ML1
Volunteer moderator
Volunteer tester

Send message
Joined: 25 Nov 01
Posts: 21227
Credit: 7,508,002
RAC: 20
United Kingdom
Message 916120 - Posted: 9 Jul 2009, 11:29:39 UTC

Other ideas for working with what is there at present...

Smooth out the traffic peaks with traffic shaping so that the link can be maxed out WITHOUT dropping packets and suffering a disgraceful degradation with effectively lost bandwidth. A simple example is to use a variation of the Wondershaper script and the linux "tc" commands. (I've an example for anyone interested. Works very well.)

Use the Ned idea of restricting access by IP ranges to ease the overload peaks. A bit like a rolling blackout!

Improve Boinc so that better traffic management is included in the first place. Automatically throttle the client transfer rate in realtime if packets get dropped/lost?


I prefer the last option as Boinc is supposed to be unobtrusive by design.

A combination of all the above options would be the most thorough traffic solution.


And again... Is not the grey fibre already inplace? It's just a question of replacing the equipment at both ends on campus.

Regards,
Martin

See new freedom: Mageia Linux
Take a look for yourself: Linux Format
The Future is what We all make IT (GPLv3)
ID: 916120 · Report as offensive
Profile ML1
Volunteer moderator
Volunteer tester

Send message
Joined: 25 Nov 01
Posts: 21227
Credit: 7,508,002
RAC: 20
United Kingdom
Message 916123 - Posted: 9 Jul 2009, 11:32:12 UTC

Meanwhile...

Isn't the biggest headache the db bottleneck and maintenance nightmare that appears to be quickly heading towards a 24hour operation in itself.

Do we get a duplicate Matt when the maintenance burden exceeds 24hours in any 24hour period?

Happy crunchin',
Martin

See new freedom: Mageia Linux
Take a look for yourself: Linux Format
The Future is what We all make IT (GPLv3)
ID: 916123 · Report as offensive
Previous · 1 · 2 · 3 · 4 · Next

Message boards : Technical News : On Bandwidth (Jul 08 2009)


 
©2024 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.