I think the major issue is that Wi-Fi ^[1] is probably the wrong technology for the job, if you're really talking about 3,000 clients in a small area like a ballroom. For fewer clients spread over a large space, I think it's feasible.

Covering a ballroom with potentially thousands of clients is going to be a stretch for Wi-Fi, assuming that the clients are actually using the network. You've only got 3 non-overlapping channels (in the US), and I've never seen an access point (AP) reasonably support more than 50 clients effectively. You're going to end up with a lot of access points sitting on the same channel and a lot of contention for the air. That's a lot of client devices to have in a small area.

If you could rig some kind of highly directional antennas and radio power was clamped down to target small numbers of clients you might make this better. For a temporary event like a conference, the level of obsessive care that such a site survey would require would, I'd imagine, be unreasonably expensive.

Assuming you're covering a lower client density than 3,000 clients in a single open-air space, you'd want to space APs with coverage zones sized to handle a significant fraction of the possible number of clients that AP can support (by tweaking radio power / antennas), and you'll want to try and keep adjacent APs on non-overlapping channels. The more APs the better, and don't overload the APs with too many clients. (Tweaking radio power / antennas to make coverage zones seems non-intuitive to anybody who hasn't tried to scale Wi-Fi to handle a large number of clients in a small physical area.)

From a layer 2 broadcast perspective, it would make sense to broadcast multiple SSIDs and back-end them into different VLANs / IP subnets. That would depend on the number of client devices and the character of the traffic. Personally, I wouldn't put more than about 500 devices in a single layer 2 broadcast domain on a corporate LAN. I can only imagine that a conference Wi-Fi network would be worse.

DHCP should be a no-brainer, though redundancy is a concern. I'd probably use the ISC dhcpd and work out a failover arrangement to a second server. I think I'd be on the lookout for rogue DHCP servers, too. On wired Ethernet you could easily disable the ports that rogue DHCP servers show up on. For wireless Ethernet, it's a little more problematic. Anybody know if there are APs that support blocking mobile units based on MAC address? (That doesn't help if the rogue DHCP server spoofs its MAC once detected, but it's a start...)

Obviously, the firewall / edge router should be able to handle the number of NAT table entries that such a number of clients might generate. A consumer toy NAT router isn't going to handle it. A redundant router protocol (HSRP, VRRP, etc) and multiple edge router devices are going to be a necessity to prevent a single point of failure from ruining the whole show.

As for bandwidth contention on the backhaul, you could throttle client bandwidth to the Internet. That should also limit the overall contention on the air, to some extent.

I'd throw something like Squid Cache in place as a transparent proxy for HTTP traffic. That's going to help with utilization of the backhaul. Your HTTP proxy cache shouldn't be a point of failure, so you'll need infrastructure to monitor the cache's health and, if it fails, route around it.

I don't have the energy to fire up a spreadsheet and look at the economics of a bunch of small Ethernet switches and patch cables strewn about, but the more that I read, the more that it sounds like wired Ethernet would be a great way to pull off decent connectivity. There would be, no doubt, major effort needed to run the Ethernet cables and power the switches, but it provides a much more manageable network infrastructure, more reliable bandwidth, and requires a lot less obsessive tweaking than wireless. You could get away with using low-end gear for the edge switches, too, since 100 Mbps service would plenty for the purposes of accessing the Internet.

Cisco has a little 8 port switch that draws its power from PoE-- the Catalyst 2960PD-8TT-L. That'd be sweet for this application-- putting something like that on each table, drawing its power from a larger PoE-capable switch. I'm guessing that those are pretty expensive for this application, but I'm guessing that there's a "downmarket" option that's not as pricey available from somebody. (Searching for switches powered by PoE seems to be fairly difficult with Gooogle...)

Intel has a 2006-era paper re: providing Wi-Fi access at conferences ^[2]. Looking at their numbers, they had 50 clients on a single AP at one point, and a peak client load under 100 clients total. Those seem like pretty small numbers compared to what you're talking about, and in 2006 everybody wasn't carrying around iPhones, etc.

(For those that are interested, I have finally written up my 2009 report on the wireless at PyCon ^[1]).

I have done the wireless for the PyCon conference most of the years since we moved from George Washington University into hotels, so I have some ideas about this, which have been proven in battle -- though only with around a thousand users.

One thing I hear a lot of people talking about in this discussion is "open air coverage in a ballroom". One theory I operate under is that the ballroom is NOT an open air environment. Human bodies soak up 802.11b/g and 802.11a quite nicely.

Here are some of my thoughts, but more details are available in my conference reports if you search google for "pycon wireless" -- the tummy.com links are what you want.

I use just the non-overlapping channels, and spread the APs out. For 802.11b/g, I run the radios at the lowest power settings. For 802.11a I run it at the higest power setting because we have so many channels.

I try to keep the APs fairly low, so the bodies can help reduce interference between APs on the same channel.

I set all the APs to the same ESSID so that people can "roam" to different APs as loads (number of associated clients) go up or coverage goes down (more people coming in, etc).

Lots and lots of APs. The first year we had the hotel do the networking, they eventually brought in 6 APs, but they had started with only a couple. Despite that we had told them that we would be heavily using their wireless. But we also had other problems like the DHCP server giving out leases with a gateway in a different network than the address. (Calls to support resulted in "I'll just reboot everything.").

We are running relatively inexpensive D-Link dual-radio APs, costing around $100 or $200 each. We just haven't really had the budget to buy 20 to 40 of the $600+ high end APs. These D-Link APs have worked surprisingly well.

In 2009 we had a hell of a problem with netbooks. Something about the radios in these just stinks for use at this sort of conference. I've heard reports of people putting Intel wireless cards in the Netbooks and getting much better performance. At PyCon 2009, my netbook couldn't get a reliable connection after the conference started, but my ThinkPad had no problems. I heard similar reports from people with Mac and other "real" laptops, but the cheapest hardware just was not working.

I have NOT done anything with directional antennas. I was expecting to need them, but so far it has worked out well enough.

Most hotels cannot provide enough bandwidth. Don't worry though, there are lots of terrestrial wireless providers that can provide 100mbps. I'm not talking about the places that run 802.11g from some tower, but people with real, serious radios and backhaul to cope with it.

Over the last several years we haven't really had much in the way of wired ports, mostly because of budget and volunteer effort required to cable all those locations. In 2010 we expect to have quite a few wired ports. I like the idea of wiring every seat for wired, but I would doubt we'll be able to cover even 10% simply due to the effort required to wire and maintain such a network. Getting people off the wireless is great.

Getting people off the 802.11b frequencies is good as well. Most people talking about since Joel has brought it up have been saying things like "3 non-overlapping channels", which is true for the 2.4GHz spectrum. However, we have seen a HUGE move towards the 5.2GHz spectrum. The first year I ran the network (2006?), we had around 25% usage. In 2008 we had over 60% in 5.2GHz.

So, yes, running wireless with thousands of people requires some thought. But, giving it some thought seems to have resulted in a fairly high level of satisfaction.

Sean

Michael Arrington, of TechCrunch, hired Mariette Systems ^[1] for TechCrunch 50 and had stellar results ^[2]. From the comments, it appears they had hundreds of CISCO switches providing RJ45 connections at every seat ( picture ^[3]) which probably got enough bandwidth off the air to make it work.

Giving 2,000 hard core Internet users simultaneous access from a single location is very, very hard. I’ve seen grown men cry when they tried and failed.

This year, though, WOW. There was more Internet at TechCrunch50 than you could shake a stick at. And for that, Mariette Systems gets that big wet kiss I promised.

The team: Ernie Mariette, Cliff Skolnick and Tim Pozar. They came in, brought bandwidth (100 Mbps line-of-sight microwave link from WiLine and 30 Mbps from Telekenex), hooked it into a BSD router and distributed it throughout the building via more than 100 Cisco switches and 28 wifi access points. There were hundreds of ethernet connections (and power strips) at attendee tables. Plus dedicated bandwith to Ustream, the DemoPit area and the main stage. And, overall, lots of very happy attendees.

There were more than 1,200 simultaneous connections at peak points, and bursts of up to 88 Mbps inbound bandwidth usage. But no one was ever cut back. And I noticed multiple people in the audience watching the live Ustream feed on their laptops. Others were watching the US Open livestream. In other words, the audience was totally wasting bandwidth. And it was wonderful.

In fact, I was a little disappointed that the audience failed to make our Internet fail. They tried their best, and were found wanting.

Apple's WWDC (at least the last few years) has had fantastic wireless coverage, but they have a team of dorks sitting in a fishbowl all week long staring at Cisco coverage maps. They deploy a ton of access points. They also cleverly block some of the large downloads from the conference site from the wireless net and instead pop up maps of where you can find an ethernet cable.

Good Wi-Fi deployment is to key to a good wireless access in large conference.

When a lot of people is trying to go online, the air traffic would jam due to interference. Imagine everyone try to "talk" to the access point, all the talking would become noise for a proper conversation when the others try to talk. This is exactly the cause in such scenario.

To overcome these problems, you can consider the followings:

1 - Lots of Smart Access Points

Enterprise grade Thin AP ^[1] has radio control, and it detects the Wi-Fi client capacity and try to detect if it's a 802.11a/b/g/n card. If that's the case, the AP controller would try to switch the user to 5G 802.11n or 802.11a channel, which is less common and thus less interference.

2 - 802.11n

When we talk about Wi-Fi, a typical 802.11g AP can NEVER achieve 54Mbps as advertise (like 3G), and it goes to about 22Mbps. Shared by 10 guests, everyone will get a modest 2Mbps. Share by 50 guests, that will be anything around 40kbps.

802.11n improves this a bit, the advertise speed of 150Mbps usually goes to 70Mbps, so you can have a more reasonable speed for more guests.

3 - More AP, More Bandwidth

Deploy more AP and that would eliminate the previous bottleneck. But with multiple AP, you also have to set their channel properly to avoid AP interfering with another AP. Don't stop with more AP, add more bandwidth so the Internet would not become the next bottleneck

Matt

I've just returned from the RIPE ^[1] 59 conference in Lisbon, where I've provided technical support, including wireless networking, in a team of 7 engineers, for about 300 attendees. For as long as I've been involved, complaints about wireless are very rare in most cases. We peak at around 250 wireless clients online, and total 44 Mbit/s traffic.

Basically, we've given up on hotel connectivity many many years ago.

Our modus operandi regarding networking is basically:

• Find a hotel with structured cabling (patches, and test a few)
• Agree with the hotel that you will be allowed to place equipment in their patch room, and support
• Have a local host (ISP or similar experienced decent network) that provides min. 100 Mbit/s into the conference area patch room
• Agree with the hotel that they will switch of all their wireless APs in the conference area
• Bring a set of switches, routers and access points, and build your own network

We have few mandays for the networking setup, including all the patches and setting up wireless. Our equipment consists of:

• 15 Cisco 802.11 a/b/g access points, usually deploying about 10
• Two Juniper J2320 routers for BGP and VRRP ^[2]
• Four Foundry switches, of which two support PoE for access points
• A simple 2.4 GHz spectrum analyzer
• A decent Fluke cable tester, measuring things like short circuits, miswiring, broken wires, cable length, signal loss

We also bring a few kilometers of UTP and lots of RJ45 connectors in case we need to work around hotel cabling.

For 802.11 b/g, we first do a survey, trying to map any interference (from other buildings or floors) and then pick channel 1/6/11 for each AP. We use a fairly high density and low power settings. During the meeting, we monitor the distribution of clients amongst APs, and adjust power levels where needed to avoid overloading a single AP.

For upstream, we have our own IPv4 and IPv6 prefix. Our public network DHCP pool has about 900 addresses. We usually set up BGP ^[3] peerings with our upstream (the local host).

So, this way we avoid crappy wireless or crappy external connectivity by the hotel. However, as you can imagine, there is a considerable cost in equipment, engineering time and perhaps the hotel bill. The good quality we can provide this way, is worth this cost.

The Microsoft Australia Tech.Ed folks have some interesting articles about their Wi-Fi ^[1] experiences at this years event.

Wireless was a heavy focus, given that all 2500+ attendees were supplied with a HP MiniNote wireless-enabled laptop, plus their phones, extra laptops, etc...

The main blog is at http://techedbackstage.net/ with Resolving a Wi-Fi performance hunch ^[2] covering a (somewhat) surprising twist.

Of course, this was further complicated during the event by the usual inconsiderate geeks who feel that a 500 Mbps uplink is for their torrenting needs. :(

Does no one else see the basic problem with hundreds (or tens of hundreds) of people physically connected by being in the same room, but separated in mind by their Internet connections? Why put everyone in the same room if you don't expect them to talk to each other, or give their undivided attention to presenters? The only radio technology that handles thousands of independent connections within a single contiguous broadcast range is cellular, so if you really want everyone in the same room to have their own connections, that's your only option, and good luck finding a cost effective solution to accomplish that. Sheesh.

Sean Reifschneider has done some great work on getting the wifi for past PyCons working. He wrote up extensive reports for PyCon 2007 ^[1], PyCon 2008 ^[2] and recently PyCon 2009 ^[3].

Two of the key takeaways for me are really sort of back to basics, first lots of bandwidth and secondly lots of access points. Getting all those access points to work well and the details of it all are of course important. But if you don't have the base of bandwidth and access points it doesn't much matter how you set it up, it's just not going to work.

In all the conferences I've organized -- about 10 now -- the best solution to the wi-fi problem is holding the conference somewhere that has solved the problem already at a large scale.

The answer? Universities and colleges. They already have to provide wireless access to thousands. Your conference isn't going to do them any harm. Hold your conference at a university or college, get them to issue you a password and you won't have to worry about wi-fi.

This question is asked of us so many times....I am technical director of an event Wi-Fi ^[1] / VoIP ^[2] / Internet Business in the UK. We only provide services to Sporting, Exhibition, Conferences and Experiential events.

Although so many of the other answers on here have technical merit, I wanted to come at it from another direction. The event manager meeting planner listens to the venue sales manager, two people who probably know very little about Wi-Fi or Internet solutions.

Most venues will tell you that they have "Wi-Fi throughout the venue". This Wi-Fi is normally sufficient for the day to day requirements of the hotel visitors but not for the 2000 hardcore techies in the conference suite!

We have been involved with various tech conference in Vancouver, Atlanta, LA, London, Seoul etc and without fail we conduct a site visit with the client, meeting planner and hotel IT Manager. The very first thing to establish is the Internet back-haul available to the event, not shared with the hotel or managed by another party, we must have ownership of it for the duration of the event. At one event we actually took over the entire hotel Internet access and gave the hotel a feed back from our router !

If you don't own it, you can't manage it.

Secondly, understand the requirements of the audience and the requirements of the client and see if the two meet ? The client has budget and performance requirements, the audience have technical and performance requirements. The big difference between the two is budget. This year in particular it has been a very big issue, we advise on the Internet speed required and the cost and the client downgrades the speed due to budget. The audience isn't happy as the Internet is slow and hey presto the Wi-Fi / network is rubbish !!

The next issue is equipment, we have no axe to grind here but generally we only use Cisco equipment because it tends to keep working and can be scaled to the size of the audience. If you go to PC World or Best Buy and get a $50 Belkin router and it fails mid way through the event....big surprise !

Finally the technical crew, every business has its own IT team or tech guru so that when things stop working they have someone who understand it all to come and fix it.

If you have an event network and decide to let the event manager or key note speaker look after the network too then you have problems, they are busy with our things and really don't stand the pressure.

Get an IT team in 24x7 to manage the network, it may not be your priority three months out but half through the keynote when the Internet fails it will become your number one priority and can kill a successful event.

Several people here have mentioned Apple's WWDC. That network covers 5,000+ people on Moscone West's three floors for a week. Each floor is more than 120,000 sq. ft. so they cover >360,000 sq. ft. Virtually everyone there has a laptop (and an iPhone).

The past 3 years or so they've done it with Cisco AP's and controllers on both 2.4 and 5.0 GHz. They also provide 10-top "worktables" with 10/100 wired switch ports to take the load off for the most hard-core of users. It works reliably in the lobbies, hallways and in the meeting rooms.

pictures:

http://developer.apple.com/wwdc/experience/

You might consider attending next year just for the experience - or send someone to take notes. They put on a very well-run conference.

I've been involved in organising BarCamps in London, where we have 150-250 people. One thing we've found is that there is rampant under-estimation of capacity by venues and contractors. If we say that 200 people are coming, they operate on the basis that 50 of them are going to use the Internet. And the extent of their use will be checking their e-mail and posting a few tweets. In reality, 200 geeks means 300 devices (laptop + smartphone + gadgets). If you are organising a conference and getting an outside company to set up the connectivity, tell them that 200 people are coming but to operate as if 600 are coming (etc.).

At a recent BarCamp, we realised that many people wanted to hack on iPhone development, so a lot of people needed to get the 700Mb+ iPhone SDK (many had forgotten to upgrade the SDK after Snow Leopard came out). The nice thing is that geeks can adapt to low bandwidth situations. I told anyone who wanted it to wait an hour or so - then I downloaded it and put it up on an HTTP server on my laptop and told them to download it from there.

Another thing to look out for: ports. I went to a hack day in London recently that blocked the ports CVS, svn and git. No, really. If you are trying to get people to write code, you need to make sure the ports are open for developer tools like version control and SSH/SCP/SFTP.

With non-profit and community events, you can generally find people who know what they are doing. The first BarCamp in Brighton did one thing very right: they got the administrators of the local free wifi zone (Pier to Pier - a free hotspot that covers Brighton beach) to help set up the event wifi. They brought a stack of modified WRT54Gs and a lot of knowledge and experience. If you organise BarCamps, be sure to fill your address book with people who know wifi, who know DHCP, dns, bind etc. Buy them copious amounts of beer and pizza.

Another thing: provide wired connectivity. The first London BarCamp had only wired connectivity because the company hosting it had concerns about wifi security. Having wired available as a backup is helpful. If you've got wired connectivity, and the venue-provided wifi falls over, you can always roll out your own with your own DHCP.

If you've got a sufficiently geeky crowd, consider Peg DHCP ^[1]. ;)

Did you hear about Xirrus ^[1]? (A friend of mine who is connected with Wi-Fi in his job adviced them as a preferable solution). See example of usage on Microsoft Events ^[2]

And Requirements: *

• Ability to simultaneously connect 3,000 users in a single room

• Simple network to be centrally managed

• Quick installation with minimum number of devices

It has from 4 to 24 802.11abg+n radios coupled to a high-gain directional antenna system into a single device along with an onboard multi-gigabit switch, Wi-Fi controller, firewall, dedicated Wi-Fi threat sensor, and an embedded spectrum analyzer.

Update: There is insider information, that even the Cisco guys think that Xirrus is cool! :-))

Agree completely with Mark re the IETF. The networking requirements for hosting an IETF meeting are online at http://iaoc.ietf.org/network_requirements.html
They're extremely detailed, and probably would cause most venue providers to faint dead away (Example: mandate for primary and secondary site internet connections, with the primary being 45mbps - 100mbps and the secondary being >10mbps, bidirectional.)

The amount of energy in that room of 3000 people trying to connect to WiFi would be huge. So many different signals would be interefering that in the end, it would all translate to noise.

The only reasonable way to get proper access would be to set up directional antennae and segregate portions of the room to different channels. Given that, it will still be nearly impossible to get "decent" connections. Best bet is to provide ethernet drops if you are honestly wanting to negate an "absolutely abysmal" internet connection.

Find out who did the system at Google I/O 2009 and do what they did. There were a number of industrial access points visible up on small towers throughout the venue, and server was incredibly reliable for how many attendees they were serving. (4000?)

Especially considering they gave away Android phones (with WiFi) to every attendee.

Something that a lot of people fail to consider is that it's typically not something that's thrown together overnight and typically takes more than just a few months to iron out the entire process (ever look at the lead time to provision an OC3 or DS3?). Most event coordinators will just mention "oh yeah, we need internet access, too" or the center/hotel will use internet availability as a selling point. The events where people have had excellent experiences has to do with coordinators who knew their target audience and knew what they needed. I've helped set up and manage custom network infrastructures at conferences that had a lead time of less than 14 days and it showed. I've also helped set up and manage custom network infrastructures for conferences where they gave notice a year in advance. The one with advanced notice had seamless, as fast as you can use it, no problem access. The one with 14 days lead time was held together with duct tape, bailing wire and prayer, and it showed. Yet, there is no public celebration when things go great. If I do my job right and people actually plan ahead, only the event coordinators know the name of the company I work for. If the coordinators screw up and don't mention it till the last second, everyone at the convention seems to think the company is incompetent (however, the web has a very short memory).

In my University (École Polytechnique de Montréal), they supposedly had something like 100 access points and there was a global dynamic channel dispatch algorithm created by the school. It would dynamically change the channel of some AP to optimize the available bandwidth and signal in a specific location in the school depending on the time of the day. Cafeteria got the highest bandwidth during lunch and so on, so the whole AP network would adapt to the behaviour of the users. They tried to reduce the channel overlapping as much as possible when high bandwidth was desired.

Something like that could be implemented in a conference where depending on the schedule, you could allocate channels dynamically to different AP.

I don't know the answer to the question asked, I just came here from Joel's blog to add an example of Conference-Wifi-Done-Right, and suggest that you might be able to get some tips from these guys.

Who guys? Apple. WWDC. 1000s of developers hit the Moscone center for a week, each toting a laptop (or two) and an iPhone (or two or three or...), all on wifi. Half of them downloading the huge Videos of the conferences they're not currently attending.

...And it all seems to work, somehow.

I have no idea what their secret is, but I'd guess that it's a small team of guys whose job it is to make sure that the conference-wifi doesn't suck.

That's probably a bit overkill for SO-days (looking fwd to it!), but it would seem that it shouldn't be TOO hard to send maybe a pair of guys ahead to each venue to take care of the main issues, at least. Like IP addresses. And maybe maximum throughput for people who are clearly "hogging the bandwidth", etc.

In short: you can't count on the venue to make your conference not-suck. That's the conference organizers' jobs. At best, you can hope that their fresh fruit isn't too green and they have enough bottled water.

At ApacheCon (1500 or so people), wireless always works great. They bring in an outside company that sets up its own 20 Mbit internet uplink and blankets the conference floors with wireless AP coverage. Depending on the hotel's wireless is virtually guaranteed to be painful. If you want it done right, you need to spend money and bring it in yourself as part of your conference planning.

You should try contacting ASSEMBLY's(http://www.assembly.org) Netcrew. Every year they manage to provide stable Wifi access for about 6000 people(well, not all are using the wireless) scattered around a hockey arena.

Of course there is a lot effort building this kind of network but probably they know some tips and tricks that could be useful for smaller events as well.

A note on IP spoofing: Friends of mine were responsible for networking at linux.conf.au 2006, and they solved the spoofing problem with a utility called dhcparpd:

dhcparpd is a tool to mitigate the effects of IP spoofing attacks on ethernet based networks. dhcparpd was initially written to help manage the conference network for Linux.Conf.AU held in Dunedin in January 2006.

dhcparpd works by spoofing ARP replies based on the information held in the lease database of an ISC DHCPD3 server instance. dhcparpd listens on a pcap socket for incoming ARP queries, queries the DHCP server for a MAC address using the OMAPI protocol and finally sends a spoofed response using the libnet packet injection library.

dhcparpd does not stop the real host sending its own ARP response. You need to configure iptables/ebtables to block these packets.

Check it out at http://research.wand.net.nz/software/dhcparpd.php.

I have been at a recurring big tech conference where there was a lot of wifi users, and it worked flawlessly.

It is the IETF.

In fact, the IETF conference network is used as showcase of exiting and new RFCs. There are special parts of the schedule where it is described, it's stats, lessons learned, and for comments, questions, and critiques by it's users, who are probably THE most technically advanced internet technology users on the planet.

Each room had a creepy looking little Cisco access point on it, with no less than SIX antenna on it. And the main conference room, those little things were everywhere, no less than 30 meters apart.

Taken from my blog: jordaneunson.com ^[1].

I hate, HATE, when I go into a conference, large meeting, campus, whatever and get a Wi-Fi ^[2] signal but my requests seem to go into the database in the sky. Conferences are notorious for this. They pack a large group of people, into a small space with a single wireless access point for them all to share. Or worse, they place multiple access points in the conference hall, all with the same SSID, on the same channel and relatively close to each other. Wi-Fi can support an extremely large amount of clients if setup properly. However you will need to take into account that as more and more people come into your network, they each create a signal with their laptop or phone and thus interference to other people around them.

First, cell phones, I do not mean Wi-Fi enabled phones, no. I mean just standard cell phones that operate on either side of the Wi-Fi 2.4 GHz band. All those cell phones are creating interference at the beginning and end of the Wi-Fi spectrum. With that in mind we first want to setup an access point in the middle of the spectrum, around channel 6. Most access points have a channel setting feature. However, as more clients join the network more noise will be placed into that little channel. Most people think that the Wi-Fi access point is just overloaded and so they add another access point to the network. This will only ever make the problem worse. The problem is not bandwidth. Say it with me now: The problem is NOT bandwidth! The problem is the ’signal to noise ratio.’

When Signal met Noise

The definition of signal to noise ratio (SNR) is the ratio of noise power that is corrupting the signal power. A ratio higher than 1 indicates more signal than noise. You will notice that higher data rates like 54 Mbit/s will drop off quickly the further you move away from the access point, a lot of people then try increasing the transmission power to allow the signal to travel further, but this just introduces more noise and thus the same SNR. The SNR dictates which data rates can be used in a wireless network. As data rates get higher, more complex methods are used for transmission and that requires much higher SNR to properly decode the signal back to the data stream on the receiving side.

Introducing Multiple Access Points

As I previously mentioned, you want to try and setup your access point around channel 6, for the sake of specificity let’s say channel 6. But once there are 25 people or so using this one channel it will be saturated and the SNR is going to go way down, thus reducing the data rates for you clients. Therefore adding another access point on this same channel is not answer, instead you want to add an access point on 2 channels away from 6, one at 4 and perhaps another at 8. Give them different SSID’s so clients can pick and choose which one to use, and name them based on the geography in which they lie. Name’s such “Access Point 1″ are not a good idea, however a name like “South Wall AP” or “Stage Left AP” is. That way the user can figure out which AP is closest to them, thus ensuring maximum data rates.

Location, Location, Location

As previously mentioned, you’ll want to name your access points after where they physically lie. Also though, is to think about electric and magnetic interference. For example, placing an access point in the ceiling directly beside a 3×20A conduit is probably not a good idea. Neither is putting it on top of a microwave or fridge. Give the access points ample space between them and finally try to get some better antenna’s than the stock ones. With these tips you’ll be able to support a large amount of people in same location with ease and without frustrating the hell out of the conference attendees.

It's usually a very simple answer: Providing the infrastructure to support 3,000 users costs money, and a lot of it. Access points, power, phat pipes, proxy, QoS, per-user throttling, etc etc.

Unless they can re-coupe these costs from the people hiring the venue they're not going to bother installing the infrastructure. And if they try to re-coupe the costs people will go elsewhere cos "They're so damn expensive".

Unless they write it off as an expense item, but I can't see that happening, except maybe at a Google or FogBugz owned conference room!

TechEd 2007 in Orlando had no problems. They had a huge number of wired stations and thousands of attendees were spread over the entire convention center. Having stations people could use to print and check email and then the separate demo/workshop areas probably cut down on the amount of time people were noodling around on their own computers when they should have been absorbing the presentations.

Maybe you could bring along an access point of your own, and connect it to a computer that's configured as a router using the venue's wifi as its uplink to the net. Then don't tell any of the attendees the password for the venue's wifi, only tell them the password for the AP you brought.

This won't help you with crappy bandwidth, but at least you won't run out of IP addresses like you did at devdays boston (which was operated perfectly other than the wifi issues.)

Edit to add: As others have mentioned, WWDC has had incredibly good wifi every time I've gone. Find out who handled their wifi, and hire them :)

You need to have lots of access points, but with their signal strength turned down. This way, you can effectively have more access points to a given area. Since there are only effectively 3 different channels, you need to lay them out in a triangle grid to minimize overlap. I would recommend just using access points, not routers so that you can have a robust DHCP solution.

Speak with Liz Frederick (liz @) who runs CFUnited ( cfunited.com ^[1]), an annual Coldfusion developer Conference. The internet and power access at this conference was better than I have ever seen. There is clearly lots of forethought put into the infrastructure to allow for hundreds of developers all trying out new code on their CF web servers back home.

At linux.conf.au 2008 (Located in the lovely internet backwater that is Australia) we easily filled a 100Mbit pipe just with users over WiFi. (The bottleneck was a 100Mbit interface on our edge router, with some better planning we would have had 200-300 Mbit available).

Each attendee used ~250k/sec, probably >750k/sec as usually less then 1/3 were connected at any time.

In my day job I run educational networks, that bandwidth is more then many entire schools use here.

For a true heavy tech conference you really do need to budget that amount of bandwidth.

What was unexpected was that many attendees tried to mirror the conference video which was located office (In Canberra not Melbourne). Once we added a local mirror things improved a lot.

Maybe you should get a cell in a box http://www.calzavara.it/lang/en/products/telecommunications/radio_station_kit.html and provide attendees with wireless 3G USB sticks.

There is something like fifteen channels of Wi-Fi ^[1] now. 3 in the crappy and horrible 2.4 GHz band and 12 in the wonderful 5 GHz band. With all these channels a decent network can be built. This is something over 300 MHz to share. It is however horrible that the iPhone 3GS does not have a dual band 2.4/5 GHz Wi-Fi radio (I think some black berry's do). This is the only reason I have not purchased one.

Not sure if it's been mentioned, but it's worth mentioning the problem of fake access points.

At Tech-Ed in Auckland this year there were at least 2 fake access points, supposedly set up with key loggers.

In fact one of the security speakers apparently had his laptop hacked right in the middle of a talk.

This topic is definitely an issue that grows each day. With the growing number of IPhone, Android phones and other handhelds that use Wifi. Also with the growing importance of social meda to events (even non tech events). This is an issue that has to given more attention by conference managers and organizers.

I agree that most of the time Internet access is just an after thought. Even some high profile tech events I've worked or attended had major wifi issues.

So much great information on this thread....still digesting.

My company is the IS/IT geeks to IEEE, IETF, IGF, etc. So thanks goes out for those who say the networks, work.

Having a great network is not a want, it is a need. Members who attend events, tend to attend many events all over the globe. They are away from home offices and must do double duty by being a member of a group working on a standard and stay connected with the office as work must move forward.

I wish I could get all the work available, but that is not possible. The best recommendations I would give are:

1) References; require a possible vendor to produce references from similar or larger events with similar needs (Wi-Fi, Cyber cafe, audio video streaming, dictation, etc). You dont want to be teaching someone you are paying.

2) Billing; Your first event with a new vendor should have a detailed description of services with start end date and time, number of engineers onsite, response time for after meeting hours. You need to know the final bill before you engage the vendor.

3) Hardware; Enterprise hardware is a must. Working with a larger IT firm will make sure hardware stays currents as they will have enough clients to always afford upgrading to best of breed hardware.

4) Staffing; Not all IT geeks are created equal. Proper planning reduces the head count for onsite staff, also seen it, done it, made it a turn key product makes it so our typical meeting is staff at 1 geek to 250 attending members, now this number will be higher if audio and video streaming, or other special needs are requested. As an example IEEE 802 with 1300 attending members is staffed with 5 VeriLAN staff.

Contact VeriLAN http://www.verilan.com for additional information.

High performance secure networking is not a plug-and-play kind of thing. For example:

You can eliminate a lot of peer-to-peer chatter yet still allow Internet access by partitioning your network up so that each client can only broadcast to itself and the router. I do this to help stop virus/worm propogation and snooping on shared drives for hotel and motel Internet users.

Instead of one class-C or class-B for everbody, you can chop up the address space into little 4 address chunks (/32 instead of /24). That gives you 64 subnets on a class C. Using the 172.27.0.0/16 range, you'd get 256*64 = 16,384 unique subnets. Each of those has its own router IP and broadcast address so your broadcasts are just between you and the router - they don't propogate beyond to the other clients - thus Windows doesn't discover other PCs with their network shares. Instead of 256 clients with:

IP 172.27.1.x netmask 255.255.255.0 gateway 172.27.1.254 and broadcast 172.27.1.255

you would get 64 clients like:

Client 1 : IP 172.27.1.1 gateway 172.27.1.2 netmask 255.255.255.252 gateway 172.27.1.2 broadcast 172.27.1.3

Client 2 : IP 172.27.1.5 gateway 172.27.1.6 netmask 255.255.255.252 gateway 172.27.1.6 broadcast 172.27.1.7

You do end up needing to write scripts to generate the configuration for dhcpd.conf and /etc/network/interfaces since there's an entry for each client. Also with this scheme, ifconfig will show you just how much traffic each client has used (up to 4 GB, after which the transferred bytes and received bytes counters reset).

... that's just messing about on the layer-3 stuff. Wired APs will be better than using WDS ^[1] (easier set-up but halves your bandwidth or worse).

I've attended a couple of Debconfs ^[1], and their wifi network just works.

Their equipment and software setup are detailed here:

• http://wiki.debconf.org/wiki/DebConf7/OpenWRT
• http://wiki.debconf.org/wiki/DebConf8/NetworkSetup
• http://wiki.debconf.org/wiki/DebConf9/OpenWRT

No fancy cisco hardware, just hardware that runs OpenWRT well like Fonera ^[2] devices.

If you want to know how to create a good tech conference network, check out the network review from the 23rd Chaos Communication Congress in Berlin:

ftp://ftp.ccc.de/congress/23C3/video/23C3-1728-en-23c3_network_review.m4v

If you don't have an hour left to watch this presentation, the short answer is: It will be a lot of work.

The Wifi setup is described at the end of the presentation (last 20 mins or so).

The PyCon conferences (almost a 1000 attendees for the last two years) usually have pretty good wireless. The conference networking is done by a firm called tummy.com. You can read Sean's writeups on the last three year's networking:

• http://www.tummy.com/Community/Articles/pycon2009-network/
• http://www.tummy.com/Community/Articles/pycon2008-network/
• http://www.tummy.com/Community/Articles/pycon2007-network/

Joel, there are two parts to solving this problem: 1: Internet access at conferences should be a paid optional extra 2: Internet access should be provided via 3G USB modems

How about the idea of having a telco provide 3G modems to everyone at the conference? Each user pays a deposit for the modem which they get back when they hand it in at the end of the conference.

The assumption seems to be that conferences have an obligation to provide free wireless. Perhaps this assumption should be challenged.

Historically, the assumption was that airlines had to carry for free as much baggage as you could lug. Nowadays baggage is an optional extra that passengers must pay for.

Why should it be any different at conferences? When you buy your conference ticket you have the option of paying your refundable deposit for the 3G modem for use during the conference. Plenty of people already have wireless mobile Internet so they'll bring their own.

The problem is not "how to provide wireless access at conferences (at a reasonable cost)". The problem is "why do conference organisers assume that wireless should be free and not an optional extra"?

I wrote a long article on why conference WiFi sucks and how to improve it. This article includes a Q&A with Tim Pozar.

http://www.muniwireless.com/2009/11/19/why-conference-wifi-sucks-and-how-to-improve-it/

Esme Vos

Joel Spolsky wrote about this issue:

http://joelonsoftware.com/items/2009/10/08.html

James Mitchell

www.jmitchell.me

Why is Wi-Fi so poor at tech conferences?

BitTorrent! That's why! Check out this blog post ^[1] from MS TechEd network engineer that performed an analysis about what was screwing up the conference network.

At this year's TechEd ^[2] they're planning to implement rate limiting, bandwidth quotas & deep packet inspection in an effort to stem the torrent tide.

The NANOG conference is targeted at the network engineers who run the Internet backbones. They have three conferences a year moving around North America and the wifi is almost always flawless. There is no more demanding audience than these network engineers many of whom are actually monitoring their own networks over the wifi in case problems arise during the conference. These are the 3rd level support guys who designed the networks and the buck stops in their mailbox/pager/cellphone.

Some of the people who have been involved in providing the conference wifi over the years have published details of what they do. http://seclists.org/nanog/2007/Feb/281 http://www.nanog.org/meetings/nanog38/presentations/kapela.pdf http://nanog.org/meetings/nanog39/presentations/abley-bof.pdf http://www.gossamer-threads.com/lists/nanog/users/95657

Some of it is planning, some of it is intimate understanding of the technology which is NOT JUST NETWORKING but involves lots of radios too. That's what leads them to put lots of APs around the edges of the room with the power cranked DOWN.