First TC50 Suckage: Wifi at conference?

@matteofabiano writes “TC50 wifi overloaded. Way too slow.”

It’s not the only note I’ve seen already about the Wifi sucking and the show hasn’t even started yet.

Jason Calacanis, co-founder of TC50, wrote last night: ““90 mbits of bandwidth at tc50 location. Roof antena and wire line. Wifi for 1500 devices. I hope we stay up!!!”

Has anyone figured out how to do good wifi for an event this size yet? I heard that TC50 spent $50,000 on the wifi. When I was at the IFA show last week in Berlin the Showstoppers co-founders told me that figure isn’t out of line. They told me that wifi is one of the great headaches of putting on a show.

Advertisements

52 thoughts on “First TC50 Suckage: Wifi at conference?

  1. Don’t you guys all have 3g phones anyway? I wouldn’t think anyone with a professional interest in this would go in trusting the site’s connectivity.

    Like

  2. Don’t you guys all have 3g phones anyway? I wouldn’t think anyone with a professional interest in this would go in trusting the site’s connectivity.

    Like

  3. Not surprised at all. And totally agree.

    And when you finally get all the wifi up and running and are scratching your head as to why its overloaded, you realize a fair bit of monkeys are sitting inside the conference hall and trying to check out the live stream, and thats what chokes it 🙂

    Cut down access to video sites, and a good chunk of the bandwidth loosens up.

    Like

  4. Not surprised at all. And totally agree.

    And when you finally get all the wifi up and running and are scratching your head as to why its overloaded, you realize a fair bit of monkeys are sitting inside the conference hall and trying to check out the live stream, and thats what chokes it 🙂

    Cut down access to video sites, and a good chunk of the bandwidth loosens up.

    Like

  5. PyCon managed it a couple of years ago, by building their own:

    http://www.tummy.com/Community/Articles/pycon2007-network/

    They spent $2200 on hardware and supported 600 conference attendees with extremely good results.

    Unfortunately, many venues refuse point blank to let you run your own wifi. They then inevitably charge you a fortune and totally screw it up, having no idea what the demands of a geek conference as opposed to a regular conference are.

    Like

  6. PyCon managed it a couple of years ago, by building their own:

    http://www.tummy.com/Community/Articles/pycon2007-network/

    They spent $2200 on hardware and supported 600 conference attendees with extremely good results.

    Unfortunately, many venues refuse point blank to let you run your own wifi. They then inevitably charge you a fortune and totally screw it up, having no idea what the demands of a geek conference as opposed to a regular conference are.

    Like

  7. I’m going to a local developer conference tomorrow, and they notified us last week that it won’t have wi-fi because the facility can’t support it for 300 attendees. In a way I thought it was kind of lame, but it actually won’t affect me at all, as I’ll just use my phone.

    Like

  8. I’m going to a local developer conference tomorrow, and they notified us last week that it won’t have wi-fi because the facility can’t support it for 300 attendees. In a way I thought it was kind of lame, but it actually won’t affect me at all, as I’ll just use my phone.

    Like

  9. “Has anyone figured out how to do good wifi for an event this size yet?”

    Wifi at conferences seems to be one of the most mysterious things to get right. At least for tech conferences. I get the feeling most venues cater to their average user and are capable of handling a small number of people surfing or checking email, but are completely not prepared to handling the large number of people at tech events who want online.

    Every venue we’ve ever done an event in has had problems with their wifi and even often the ethernet connections as well.

    Like

  10. “Has anyone figured out how to do good wifi for an event this size yet?”

    Wifi at conferences seems to be one of the most mysterious things to get right. At least for tech conferences. I get the feeling most venues cater to their average user and are capable of handling a small number of people surfing or checking email, but are completely not prepared to handling the large number of people at tech events who want online.

    Every venue we’ve ever done an event in has had problems with their wifi and even often the ethernet connections as well.

    Like

  11. hey folks- i run defrag (www.defragcon.com), and wifi isn’t really a mystery or even all that hard. you have to:

    a) sign a contract w/ a location that will let you bring in outside vendors (not hard)
    b) use Swisscomm (they’re 2x expensive, but actually get it right)
    c) ask them for only 25 connections per access point – and FLOOD the place with access points.
    d) if you’re doing tons of streaming video – adjust accordingly.

    ejn

    Like

  12. hey folks- i run defrag (www.defragcon.com), and wifi isn’t really a mystery or even all that hard. you have to:

    a) sign a contract w/ a location that will let you bring in outside vendors (not hard)
    b) use Swisscomm (they’re 2x expensive, but actually get it right)
    c) ask them for only 25 connections per access point – and FLOOD the place with access points.
    d) if you’re doing tons of streaming video – adjust accordingly.

    ejn

    Like

  13. We’re WiFi specialists at http://cfree.ro. And we know that base stations and client devices alike are not too smart. 🙂 1500+ users would demand in the range of 60 consumer routers, amounting around $6000 alone. their placement and load balancing is crucial. The backbone is relevant but distribution is key. If, say 500 visitors hop on 3 routers placed in the main hall the net will crumble in seconds.

    we think we found a way to enhance load balancing but it’s hardly a hardware solution and more of a hack.

    Hope the TC50 guys get it right!
    Dan

    Like

  14. We’re WiFi specialists at http://cfree.ro. And we know that base stations and client devices alike are not too smart. 🙂 1500+ users would demand in the range of 60 consumer routers, amounting around $6000 alone. their placement and load balancing is crucial. The backbone is relevant but distribution is key. If, say 500 visitors hop on 3 routers placed in the main hall the net will crumble in seconds.

    we think we found a way to enhance load balancing but it’s hardly a hardware solution and more of a hack.

    Hope the TC50 guys get it right!
    Dan

    Like

  15. Providing large amounts of bandwidth over RF is a *hard* problem. There’s a reason why companies like AT&T have large in-house engineering departments that do nothing but worry about making their network of base stations work as well as they can. And they have the advantage of working with a system (GSM/GPRS/EDGE and/or 3G) and network equipment that was designed for exactly that purpose.

    The only way to effectively serve several hundred people using wi-fi is to have many access points spread out across your physical location with each AP taking part of the load. However wi-fi was never designed to work in that type of “cellular” type environment, it has limited spectrum availability and very minimal RF power control which is a perfect recipe for a self-interference limited network. Even if you use fancy “switched” APs that coordinate with each other to prevent APs on the same frequency talking at the same time you still end up being limited by the available spectrum and inefficiency of the wi-fi protocol itself.

    Like

  16. Providing large amounts of bandwidth over RF is a *hard* problem. There’s a reason why companies like AT&T have large in-house engineering departments that do nothing but worry about making their network of base stations work as well as they can. And they have the advantage of working with a system (GSM/GPRS/EDGE and/or 3G) and network equipment that was designed for exactly that purpose.

    The only way to effectively serve several hundred people using wi-fi is to have many access points spread out across your physical location with each AP taking part of the load. However wi-fi was never designed to work in that type of “cellular” type environment, it has limited spectrum availability and very minimal RF power control which is a perfect recipe for a self-interference limited network. Even if you use fancy “switched” APs that coordinate with each other to prevent APs on the same frequency talking at the same time you still end up being limited by the available spectrum and inefficiency of the wi-fi protocol itself.

    Like

  17. With each WiFi hotspot taking up 3 channels, and around 11-13 channels available, that gives the conference around 4 channels of WiFi (so four access points) optimally. With N, this translates out to about 248 Mb/s (top speed, but that never happens), and with four (again, an optimal situation) this is about 992 Mb/s for the whole conference (in the 2.4 GHz band).

    With the proposed, say 1500 devices, this isn’t very much; in fact, this is about 0.66 Mb/s or 675.84 kb/s, or about 12 times faster then dial up, or a very slow “broadband” connection. And that would be optimal.

    Of course, there are other ways to hit the problem; using N on the 5 GHz band where there is less interference, using “3G” cellular technologies, etc.

    But this still runs into the problem of the shear amount of people in such a small space expecting to use a large amount of bandwidth with decent latency. Of course there are solutions, just very few made specifically for the convention environment (the same problem exists a little more spread out in large cities; cellular networks deal with it fine by having smaller but more numerous access points), raising the same problem at every one.

    Of course, the convention is an ideal place to work on a new technology (well, not so much new, but more scalable to ever smaller access points and ever more of them; on the cheap) focusing on fixing this problem for when it becomes more prevalent elsewhere, but that defeats the purpose of WiFi in the first place — almost everyone has it. Additionally, developing such a thing would be much more expensive then $50k.

    So what solution would be ideal in this mess?

    Network management? No; it merely pisses off users more; nobody likes being told to stay off of YouTube, and being blocked completely is censorship.

    I’ve got one better. Add more capability not through the air, but… GASP! with good old wires. Providing Ethernet to seated guests in addition to an outlet is a good start, as their power and bandwidth hungry devices can gobble it up as they report on your tech.

    If they need to get up and go, they can disconnect, and start using the WiFi again. Might be slower, but it will provide a bit of connectivity until they get to the next demonstration.

    It is a someone “low” tech solution to a lingering problem. But who knows. Maybe someone will adapt it next year.

    Like

  18. With each WiFi hotspot taking up 3 channels, and around 11-13 channels available, that gives the conference around 4 channels of WiFi (so four access points) optimally. With N, this translates out to about 248 Mb/s (top speed, but that never happens), and with four (again, an optimal situation) this is about 992 Mb/s for the whole conference (in the 2.4 GHz band).

    With the proposed, say 1500 devices, this isn’t very much; in fact, this is about 0.66 Mb/s or 675.84 kb/s, or about 12 times faster then dial up, or a very slow “broadband” connection. And that would be optimal.

    Of course, there are other ways to hit the problem; using N on the 5 GHz band where there is less interference, using “3G” cellular technologies, etc.

    But this still runs into the problem of the shear amount of people in such a small space expecting to use a large amount of bandwidth with decent latency. Of course there are solutions, just very few made specifically for the convention environment (the same problem exists a little more spread out in large cities; cellular networks deal with it fine by having smaller but more numerous access points), raising the same problem at every one.

    Of course, the convention is an ideal place to work on a new technology (well, not so much new, but more scalable to ever smaller access points and ever more of them; on the cheap) focusing on fixing this problem for when it becomes more prevalent elsewhere, but that defeats the purpose of WiFi in the first place — almost everyone has it. Additionally, developing such a thing would be much more expensive then $50k.

    So what solution would be ideal in this mess?

    Network management? No; it merely pisses off users more; nobody likes being told to stay off of YouTube, and being blocked completely is censorship.

    I’ve got one better. Add more capability not through the air, but… GASP! with good old wires. Providing Ethernet to seated guests in addition to an outlet is a good start, as their power and bandwidth hungry devices can gobble it up as they report on your tech.

    If they need to get up and go, they can disconnect, and start using the WiFi again. Might be slower, but it will provide a bit of connectivity until they get to the next demonstration.

    It is a someone “low” tech solution to a lingering problem. But who knows. Maybe someone will adapt it next year.

    Like

  19. Several hundred iPhone and Sprint data card users all trying to connect to one cell tower? Yeah, that’s gonna work well…

    Like

  20. Several hundred iPhone and Sprint data card users all trying to connect to one cell tower? Yeah, that’s gonna work well…

    Like

  21. @Andrew has it right doing wifi as the primary netacess for a very techie conference is going to be Very Very hard – this isn’t putting some kiddie Ap/Router in your home which any Sub can handle.

    @yert

    err I think your confused 802.11B/G only has 3 channels non overlapping 802.11A gets you 8 non overlapping channels cisco say 10-15 hosts per ap.

    N just makes it even worse as it uses wider channels

    So in a large room you to have to do some complex signal level adjustments to have non overlapping cells 802.11a makes it a lot easier as there are more channels.

    After that you then get into roaming and mobile IP so you don’t lose a connection as you move from AP to AP.

    Like

  22. @Andrew has it right doing wifi as the primary netacess for a very techie conference is going to be Very Very hard – this isn’t putting some kiddie Ap/Router in your home which any Sub can handle.

    @yert

    err I think your confused 802.11B/G only has 3 channels non overlapping 802.11A gets you 8 non overlapping channels cisco say 10-15 hosts per ap.

    N just makes it even worse as it uses wider channels

    So in a large room you to have to do some complex signal level adjustments to have non overlapping cells 802.11a makes it a lot easier as there are more channels.

    After that you then get into roaming and mobile IP so you don’t lose a connection as you move from AP to AP.

    Like

  23. Extricom (www.extricom.com) has a solution similar that makes wi-fi easy to implement in a large scale. The available frequencies are used “horizontaly” as “blanket” instead as the traditionnal “cell planning”. This enables large coverage with the same frequency (the controller decides which AP serves the client) and the other available frequencies can then be used for other matters (e.g. one freq./blanket for voice, one for data, one for specific users etc.).
    It is worth checking it out as it is also very easy to install and maintain!

    Like

  24. Extricom (www.extricom.com) has a solution similar that makes wi-fi easy to implement in a large scale. The available frequencies are used “horizontaly” as “blanket” instead as the traditionnal “cell planning”. This enables large coverage with the same frequency (the controller decides which AP serves the client) and the other available frequencies can then be used for other matters (e.g. one freq./blanket for voice, one for data, one for specific users etc.).
    It is worth checking it out as it is also very easy to install and maintain!

    Like

  25. Meru has a box that has 10 APs in it – sure they do, it lives under a plastic case in their wild dreams of shipping a not shitty product.

    Extricom is another POS, they also do the “fair sharing of limited bandwidth”…doesn’t anyone remember ATM? Guess not, all the Dushes that bought that story lost their jobs.

    So, avoid losing your job by purchasing pipe dreams and just buy the product that supported InterOp’s show floor flawlessly 5 times – XIRRUS, the only product with the same bandwidth as a 24 port of 48 port switch.

    In fact they just announced a 24 radio array today.

    Or, just buy Meru, I hear they have run out of cash and will likely dry up a blow away unless Juniper is actually dumb enough to buy them.

    Have a great day! And, any dumbies can dry up and blow away.

    Like

  26. Meru has a box that has 10 APs in it – sure they do, it lives under a plastic case in their wild dreams of shipping a not shitty product.

    Extricom is another POS, they also do the “fair sharing of limited bandwidth”…doesn’t anyone remember ATM? Guess not, all the Dushes that bought that story lost their jobs.

    So, avoid losing your job by purchasing pipe dreams and just buy the product that supported InterOp’s show floor flawlessly 5 times – XIRRUS, the only product with the same bandwidth as a 24 port of 48 port switch.

    In fact they just announced a 24 radio array today.

    Or, just buy Meru, I hear they have run out of cash and will likely dry up a blow away unless Juniper is actually dumb enough to buy them.

    Have a great day! And, any dumbies can dry up and blow away.

    Like

  27. If you are using 802.11A you now have 24 non-overlapping channels. If you throw enough radios at the site with non-overlapping channels you will get enough bandwidth for all of the clients. The trick is getting all of the radios in the space without RF interference. Extricom’s and Meru’s issue with their channel blankets is you have limited overall throughput per blanket (each blanket has limited bandwidth). You are better off doing some good planning with Cisco, Aruba, or Xirrus and getting much higher throughput to the conference attendees.

    Like

  28. If you are using 802.11A you now have 24 non-overlapping channels. If you throw enough radios at the site with non-overlapping channels you will get enough bandwidth for all of the clients. The trick is getting all of the radios in the space without RF interference. Extricom’s and Meru’s issue with their channel blankets is you have limited overall throughput per blanket (each blanket has limited bandwidth). You are better off doing some good planning with Cisco, Aruba, or Xirrus and getting much higher throughput to the conference attendees.

    Like

  29. Agree re: Aruba. Have done some consulting with KPMG Aruba V Cisco and the planning and actual practical application of Aruba was very solid. Supporting (admittedly) small conferences and *many* drop-in first year accountants and BAs at their main offices as they come in to do a lot of syncing.

    Like

  30. Agree re: Aruba. Have done some consulting with KPMG Aruba V Cisco and the planning and actual practical application of Aruba was very solid. Supporting (admittedly) small conferences and *many* drop-in first year accountants and BAs at their main offices as they come in to do a lot of syncing.

    Like

  31. @Maurice – In some parts of the world there are 4 non-overlapping channels for b/g; N can use the same band as a can, but I didn’t know how many channels that had, so thanks for pointing that one out.

    Of course, my estimates were very rough, and only to make the point that tech-convention type areas don’t have a proper solution available most of the time. My suggestion was pretty much to cop out and use a wired solution, as most laptops don’t necessarily have the hardware to deal with the 5 GHz band, and adopting WiFi to a cellular environment is yet to be done right.

    Like

  32. @Maurice – In some parts of the world there are 4 non-overlapping channels for b/g; N can use the same band as a can, but I didn’t know how many channels that had, so thanks for pointing that one out.

    Of course, my estimates were very rough, and only to make the point that tech-convention type areas don’t have a proper solution available most of the time. My suggestion was pretty much to cop out and use a wired solution, as most laptops don’t necessarily have the hardware to deal with the 5 GHz band, and adopting WiFi to a cellular environment is yet to be done right.

    Like

  33. WOW, I knew, of course, that technology was expensive at conferences but never considered the cost of setting up Wifi networks!

    I would think it’s even worse at conferences where people are live blogging, sending hundreds of tweets, uploading audio, photos and video!

    I haven’t had a noticeable problem at any conferences I’ve attended so far this year but now I’m wondering about BlogWorld next weekend…

    Like

  34. WOW, I knew, of course, that technology was expensive at conferences but never considered the cost of setting up Wifi networks!

    I would think it’s even worse at conferences where people are live blogging, sending hundreds of tweets, uploading audio, photos and video!

    I haven’t had a noticeable problem at any conferences I’ve attended so far this year but now I’m wondering about BlogWorld next weekend…

    Like

  35. If your Wi-Fi use is increasing, you’re not alone. As the numbers (http://webworkerdaily.com/2007/09/19/wi-fi-hotspot-usage-the-numbers-are-upway-up) show, business and home use are increasing with tremendous speed and are critical tools for workers everywhere. So the question is…are you monitoring your Wi-Fi to know what is around you and what performance you’re getting?

    Here is a great FREE tool that you can use wherever you are to make monitoring the air around you easy….

    Xirrus Wi-Fi Monitor (http://www.xirrus.com/library/wifitools.php) does the usual norm that many of the “look for network availability” tools do, and then some. It continuously scans and displays a list of available networks as well as delivers lots of information about the networks available to you — type of network, security needed, and more.

    For users of public Wi-Fi hotspots, this is a great tool as it shows you the networks within range and what is needed to connect to them. The tool can be downloaded as a Windows Vista or XP Gadget, a Mac OSX Widget, or a Desklet for Linux.

    Try it — it’s free, easy to install, and will save you frustrations trying to figure out what networks are in range.

    Its a suggestion and will help you all figure out if an area has good or bad Wi-Fi

    Like

  36. If your Wi-Fi use is increasing, you’re not alone. As the numbers (http://webworkerdaily.com/2007/09/19/wi-fi-hotspot-usage-the-numbers-are-upway-up) show, business and home use are increasing with tremendous speed and are critical tools for workers everywhere. So the question is…are you monitoring your Wi-Fi to know what is around you and what performance you’re getting?

    Here is a great FREE tool that you can use wherever you are to make monitoring the air around you easy….

    Xirrus Wi-Fi Monitor (http://www.xirrus.com/library/wifitools.php) does the usual norm that many of the “look for network availability” tools do, and then some. It continuously scans and displays a list of available networks as well as delivers lots of information about the networks available to you — type of network, security needed, and more.

    For users of public Wi-Fi hotspots, this is a great tool as it shows you the networks within range and what is needed to connect to them. The tool can be downloaded as a Windows Vista or XP Gadget, a Mac OSX Widget, or a Desklet for Linux.

    Try it — it’s free, easy to install, and will save you frustrations trying to figure out what networks are in range.

    Its a suggestion and will help you all figure out if an area has good or bad Wi-Fi

    Like

  37. Yert –

    You said “most laptops don’t necessarily have the hardware to deal with the 5 GHz band”.

    The only way this would be true is if you either are buying $400 note books or have a population of very old notebooks (+3 years).

    .11a (5Ghz) support eclipsed 2.4Ghz only support in June of 2004 when Intel changed all centrino to a/b/g support. And, any MAC shipped in the last 2 years has a/b/g/n built in.

    It’s surprising how many people have no idea they are tri-mode already.

    Like

  38. Yert –

    You said “most laptops don’t necessarily have the hardware to deal with the 5 GHz band”.

    The only way this would be true is if you either are buying $400 note books or have a population of very old notebooks (+3 years).

    .11a (5Ghz) support eclipsed 2.4Ghz only support in June of 2004 when Intel changed all centrino to a/b/g support. And, any MAC shipped in the last 2 years has a/b/g/n built in.

    It’s surprising how many people have no idea they are tri-mode already.

    Like

Comments are closed.