99% of the country? Hmmm, probably not even 75%. Perhaps you mean 99% of the population?
If telecomms were cheap here, fewer people would have mobile phones... a lot of people I know only have them because it's so much cheaper to call (or especially, text) a mobile from another mobile, than it is calling one from a landline. (Not that it's particularly cheap - it's just that calling a mobile from a landline is so expensive).
The bundled minutes model isn't the only one widely used in at least the countries in Europe I've experience of. Instead of this model (i.e. subsidised phone, bundled minutes, cheaper call charges, monthly fees, minimum contract length) a lot of people are paying nearly full price for the phone then paying per minute for all their calls, but without monthly fees or minimum contract length.
There are interference problems between the original specs of bluetooth and 802.11b. There are working groups at IEEE trying to sort it out (search their website or search google for "802.11b bluetooth interference" and dig away).
Fortunately some modifications have been suggested to Bluetooth which should significantly reduce the problems. Let's just hope that these are incorporated into any mass-market bluetooth devices or it doesn't bode well for wireless internet access via 802.11b in some places, for example, coffee shops, where you are also likely to see a lot of mobile phones...
Frequency-hopping systems generally use frequencies from within a wider band but keep jumping between them, so they don't interfere with any one other user for a long period of time. Trouble is, this doesn't tie in very well with ethernet/TCP protocols where performance is seriously impacted by packet loss. (Sure, the packets are resent, but TCP treats packet loss as congestion and slows down).
Some graphics card drivers will at least let you change the colour tone, which indirectly changes the white to something a little more pleasant, but you can't get white-on-dark with those. (obviously you usually need to write the registry to save these changes, but it's usually only about 5 mouseclicks if you can't). matrox definitely does this, I'm sure I've seen others.
If the powers that be are really that bothered about costs of re-ghosting, they should consider something like tandberg.se's 'safety card ultra' (I am sure there are other similar cards). This is a pci card which redirects HD writes so they are trashed after the current session. Of course this is a bit of a pain for you, but then you might be able to convince them to let you write anything you want keeping to the protected partition, after they have seen it works okay...
Re:How Much Bandwidth Stylesheets Can Save You...
on
Slashdot Updates
·
· Score: 1
Damn right. Stylesheets make a huge difference on anything using tables and font tags. And reading Slashdot on a dialup gets a lot more pleasant if you're connecting through a gzipping proxy (junkbuster-zlib for example - horrible http support and the filtering's not a patch on Proxomitron, but it does gzip). Not too sure about yet more Apache modules, I think front-end accelerator boxes is the way to go (perhaps Oops with a plugin could do the gzip magick - got to be easier than starting from scratch). Even better, offloading the crappy 33k dialups like mine onto lightweight cache boxes drops the RAM load of the Apache-perl servers like nothing else.
Some posts on the BAWUG wireless list have these URLs for 900MHz kit (which I always forget about since it's not allowed in UK): these tend to be slower but work over greater distance.
Novaroam 70 miles LOS Freewave (good for point-to-point, not so good for point-multipoint)
Man, why is it when companies build in NICs on motherboards they always choose the crappiest one they can find?
Look in the bit you quoted from the driver source and you see this:
On the bright side, the 8139 does have a built-in PHY
That's got to help keep the size of the board down.
ETSI BRAN's Hiperlan 2 standard provides guarantees for time-sensitive data (and has support for a lot of other features which make it much more suitable than 802.11a for providing widespread network infrastructure - things like frequency selection, control of transmit level).
It's definitely more fun to blame MSN than it is to troubleshoot problems with inbound mail into AOL. Though, I guess it's more likely to be MSN, if it was AOL I doubt if there would be any report of failure, the mail would just be dropped on the floor.
Make sure you don't skip setting the access-list as tightly as possible, otherwise you're providing an anonymising open relay for anyone who port scans you and sees the SMTP response.
It's *much* better to use authenticated SMTP if at all possible.
If users are forced through MSN's mail servers, those servers can rate-limit the mail. If that's done, it's quite effective at stopping spam (and people running their own mailing lists... oh look, MS provide a commercial mailing list service! oh, what a surprise!).
Yes, under FCC rules you are allowed a few extra dB if the signal is directional. Very handy, this is why US users can use things like 24dBi parabolic-grid antennas. Of course the Canadian regs are slightly different (apparently the authorites are quite helpful, and there is plenty of precedent for Canadian school districts using 2.4GHz wireless LAN, going back to 1996 - before the days of interoperability between different manufacturers kit, back when a 3Mbps bridge would set you back $2500:)
Here in the UK we have a lot of dark fibre run along canals - they're one of the few people other than the railways who own long strips of land going across large areas of the country. There's also fibre wrapped around the powerlines of the national grid, as well as the railways.
It would be so nice if more ISPs started charging different rates for local and transit bandwidth so that more people could afford to use the infrastructure that's there, but that's another story...(as is the rather unusual and slightly annoying situation where we can't use the unlicensed spectrum at 2.4GHz to provide commercial services).
None of the *DSLs will reach that far without repeaters and additional DSL router hops en-route. Even IDSL, which I think would be too slow for a school district anyay, only reaches about 21km *with* 2 repeaters on the way (and, it's usually more difficult to locate DSL repeaters than wireless repeaters).
Very interesting. I think it would give better range with less feed cable (reducing amp power as necessary to keep within FCC limits). This can best be achieved by placing the radios as close as possible to the antennas (in weatherproof boxes). Cleaning up the electricity supply to the amp was a good idea (for anyone else trying this, use a decent UPS - it has to provide a conditioned feed all the time rather than only switching in during abnormal conditions, and I suspect that cheap ones wouldn't give a particularly good waveform on the AC).
There's plenty of 802.11 DS equipment too... Fair point about not having access to the client side (as long as the equipment itself is secure). Though you don't need to authenticate if you just want to sniff traffic.
Commercial laser systems are available, though some types of building construction aren't suitable (at least, I know of someone who tried this locally, and their building vibrated too much).
#1, yes, but this is why you conduct a survey to determine where to place any equipment. The type of GIS the telcos use is generally able to take into account things like tree growth etc.
#2, using fairly tight beams (which you want anyway, there is no point in wasting power sending transmissions where they're not needed) from towers things like mobile phones are not likely to be much problem.
Of course the exact setup will depend on factors such as terrain and which licensing restrictions you are subject to, but providing you can find locations for repeater stations (which can be solar powered, so you don't need a mains electricity supply) this should be feasible.
Here are some URLs you might find interesting:
HPWREN (featured here recently) have a 45mb backbone using western multiplex tsunami kit, and 802.11b access points. They use solar power and batteries to power some backbone nodes.
Using 802.11b or similar tech, you should expect each wireless hop to add about 5ms of latency, maybe a little more depending on distance. You can quite easily build a repeater by connecting two bridges together by a X-over cable. You could probably do this with Linksys WAP11 or similar, but over this type of distance you will find it much easier to use something like the high-spec version of Cisco Aironet 350 bridges (the 100mW versions will push the signal a lot further - 25 miles with 24dBi antennas - you can use Cisco's own, alternatives include Superpass (based in Waterloo), HyperLinkTech and others.
Aironet bridges let you set the distance of the link which modifies timing parameters (a slight problem with standard 802.11b over long distances), and their security is better than WEP.
There's plenty of homebrew opportunities for antennas and other related kit, although I guess they're probably of more use to people who don't have a budget to play with (: There's a collection of links on this page with a particular focus on homebrew kit.
Slashdot Mirror
Sounds like GATech need some bandwidth shaping *badly*... Hmmmm, P2P. Hadn't thought of that - it would explain some 350ms I've been seeing...!!
If telecomms were cheap here, fewer people would have mobile phones... a lot of people I know only have them because it's so much cheaper to call (or especially, text) a mobile from another mobile, than it is calling one from a landline. (Not that it's particularly cheap - it's just that calling a mobile from a landline is so expensive).
The bundled minutes model isn't the only one widely used in at least the countries in Europe I've experience of. Instead of this model (i.e. subsidised phone, bundled minutes, cheaper call charges, monthly fees, minimum contract length) a lot of people are paying nearly full price for the phone then paying per minute for all their calls, but without monthly fees or minimum contract length.
Fortunately some modifications have been suggested to Bluetooth which should significantly reduce the problems. Let's just hope that these are incorporated into any mass-market bluetooth devices or it doesn't bode well for wireless internet access via 802.11b in some places, for example, coffee shops, where you are also likely to see a lot of mobile phones...
Frequency-hopping systems generally use frequencies from within a wider band but keep jumping between them, so they don't interfere with any one other user for a long period of time. Trouble is, this doesn't tie in very well with ethernet/TCP protocols where performance is seriously impacted by packet loss. (Sure, the packets are resent, but TCP treats packet loss as congestion and slows down).
If the powers that be are really that bothered about costs of re-ghosting, they should consider something like tandberg.se's 'safety card ultra' (I am sure there are other similar cards). This is a pci card which redirects HD writes so they are trashed after the current session. Of course this is a bit of a pain for you, but then you might be able to convince them to let you write anything you want keeping to the protected partition, after they have seen it works okay...
Damn right. Stylesheets make a huge difference on anything using tables and font tags. And reading Slashdot on a dialup gets a lot more pleasant if you're connecting through a gzipping proxy (junkbuster-zlib for example - horrible http support and the filtering's not a patch on Proxomitron, but it does gzip). Not too sure about yet more Apache modules, I think front-end accelerator boxes is the way to go (perhaps Oops with a plugin could do the gzip magick - got to be easier than starting from scratch). Even better, offloading the crappy 33k dialups like mine onto lightweight cache boxes drops the RAM load of the Apache-perl servers like nothing else.
here's someone who /might/ be able to do consultation and engineering who seems to know his stuff.
Novaroam 70 miles LOS
Freewave (good for point-to-point, not so good for point-multipoint)
Bit lacking in the PDC department then... better just use a VCR if you want to be sure of catching things :)
ETSI BRAN's Hiperlan 2 standard provides guarantees for time-sensitive data (and has support for a lot of other features which make it much more suitable than 802.11a for providing widespread network infrastructure - things like frequency selection, control of transmit level).
There's not much about it, but the Whistle InterJet pop3 server seems to support it, so there's at least one implementation on UNIX.
It's definitely more fun to blame MSN than it is to troubleshoot problems with inbound mail into AOL. Though, I guess it's more likely to be MSN, if it was AOL I doubt if there would be any report of failure, the mail would just be dropped on the floor.
It's *much* better to use authenticated SMTP if at all possible.
If users are forced through MSN's mail servers, those servers can rate-limit the mail. If that's done, it's quite effective at stopping spam (and people running their own mailing lists... oh look, MS provide a commercial mailing list service! oh, what a surprise!).
View from inside damaged building looking past switches to hole in side of building.
Yes, under FCC rules you are allowed a few extra dB if the signal is directional. Very handy, this is why US users can use things like 24dBi parabolic-grid antennas. Of course the Canadian regs are slightly different (apparently the authorites are quite helpful, and there is plenty of precedent for Canadian school districts using 2.4GHz wireless LAN, going back to 1996 - before the days of interoperability between different manufacturers kit, back when a 3Mbps bridge would set you back $2500 :)
It would be so nice if more ISPs started charging different rates for local and transit bandwidth so that more people could afford to use the infrastructure that's there, but that's another story...(as is the rather unusual and slightly annoying situation where we can't use the unlicensed spectrum at 2.4GHz to provide commercial services).
This just appeared on the BAWUG list, with some info on Canadian regulations for the 2.4GHz band.
None of the *DSLs will reach that far without repeaters and additional DSL router hops en-route. Even IDSL, which I think would be too slow for a school district anyay, only reaches about 21km *with* 2 repeaters on the way (and, it's usually more difficult to locate DSL repeaters than wireless repeaters).
Very interesting. I think it would give better range with less feed cable (reducing amp power as necessary to keep within FCC limits). This can best be achieved by placing the radios as close as possible to the antennas (in weatherproof boxes). Cleaning up the electricity supply to the amp was a good idea (for anyone else trying this, use a decent UPS - it has to provide a conditioned feed all the time rather than only switching in during abnormal conditions, and I suspect that cheap ones wouldn't give a particularly good waveform on the AC).
There's plenty of 802.11 DS equipment too... Fair point about not having access to the client side (as long as the equipment itself is secure). Though you don't need to authenticate if you just want to sniff traffic.
Commercial laser systems are available, though some types of building construction aren't suitable (at least, I know of someone who tried this locally, and their building vibrated too much).
#2, using fairly tight beams (which you want anyway, there is no point in wasting power sending transmissions where they're not needed) from towers things like mobile phones are not likely to be much problem.
Here are some URLs you might find interesting: HPWREN (featured here recently) have a 45mb backbone using western multiplex tsunami kit, and 802.11b access points. They use solar power and batteries to power some backbone nodes.
Some other people using mostly 802.11b kit who will have some information you can use: BAWUG PersonalTelco.net NoCat.net Freenetworks.org
Using 802.11b or similar tech, you should expect each wireless hop to add about 5ms of latency, maybe a little more depending on distance. You can quite easily build a repeater by connecting two bridges together by a X-over cable. You could probably do this with Linksys WAP11 or similar, but over this type of distance you will find it much easier to use something like the high-spec version of Cisco Aironet 350 bridges (the 100mW versions will push the signal a lot further - 25 miles with 24dBi antennas - you can use Cisco's own, alternatives include Superpass (based in Waterloo), HyperLinkTech and others.
Aironet bridges let you set the distance of the link which modifies timing parameters (a slight problem with standard 802.11b over long distances), and their security is better than WEP.
There's plenty of homebrew opportunities for antennas and other related kit, although I guess they're probably of more use to people who don't have a budget to play with (: There's a collection of links on this page with a particular focus on homebrew kit.