On his blog, Evert waxes about how conferences tend to have poor wifi:
What’s needed is for someone knowledgeable to carry out a site survey and to use the results of this to work out a radio-plan. Take in consideration all possible sources of interference, the network load (i.e. amount of bandwidth needed) and make sure that you implement correct channel management and you should be OK. Coverage area and number of users can just be plugged into this matrix as it is scaleable.
Hm, hold on there. WiFi connectivity was spotty at OCC BBQ and wasn’t that you I saw in a suit pottering around with WiFi access points at the Digital Island Meetup/BizSpark Launch in November? I couldn’t get connected to WiFi there either. It’s not kosher to throw stones and I’m not throwing stones here – I’ve previously documented my own experiences with WiFi at events.
The problem is, in my opinion, shitty equipment. Whether it’s the oft-documented Free Public WiFi misfeature (Thanks, Microsoft) or the fact that some people don’t turn off their torrent applications when they connect to public wifi – I’ve seen it happen. All WiFi equipment is made by the lowest bidder. It’s all turd. And it’s worse when Windows creates ad-hoc networks at random without permission. That’s just arsehole behaviour.
The problem is… WiFi is shit.
Firstly it’s entirely based on collision detection. And it’s on a shared medium. So every time you add a new user on there it’s going to decrease the chance you can get any data. We hated this when it was a wire and we hate it when it’s wireless. But it’s a standard. So what are you going to do.
People should not expect massive data rates. They don’t need massive data rates. You need less than a megabit a second for decent WiFi. Anything faster than that and you’re just pissing into the wind. Use a damn wire already.
http://russellmcquillan.com/2008/12/11/wi-fi-at-conferences/
Matt, first of all you’re comparing apples with oranges.
1) The OCC BBQ was a free event. There was no 1500 euro ticket charge.
2) The wifi at the OCC BBQ was fine in and around the Derg Inn and was only spotty in the hall where the presentations were held. Fully my fault but I ran out of time while setting it up. Being the sole organiser these things tend to happen. Apart from that area there were no complaint about the wifi. (and yes, I am ashamed for that one cock-up). Apart from that area there was 3Mbps solid wifi access.
3) The OCC BBQ was held out in the sticks where we had to put a wireless backhaul in place from 10 miles away. Considering all that we did a lot whole lot better than Le Web or quite a few other conferences.
As for the wifi at the BizSpark launch; I was asked at the last moment if I could help out there as the “public” wifi was 22 euro a pop. I asked for an “open” ethernet connection so that I could run a wifi hotspot from it. As it turns out the wired ethernet was also routed via a Wayport captive portal asking for 22 euro a pop. The result was that the wifi worked but there was no backhaul so I switched it of. Also it was another free event.
My gripe (if you want to call it that) is with overpriced, overhyped tech conferences that cannot even take care of a simple thing as wifi access. It’s not like they don’t have the time and budget to do so. There is no point in comparing these type of conferences with grassroot events such as Open Coffee Clubs or Barcamps and the likes..
Your comments about wifi as a standard/technology are without foundation. Not all wifi equipment is shit. Ruckus makes excellent wifi hardware as do Apple, Buffalo Tech, Ubiquity and a lot of other manufacturers. Yes there is shit wifi hardware for sale but that’s not the point. Wifi is not shit.
“It’s a shared medium”? What is the Internet then? And how about mobile phone networks?
Oh dear.
The Internet is ‘not’ CDMA. It’s also not ‘shared’ in the networking sense Mobile phone networks are also different, much more economic with spectrum, much higher power and in many cases switched. You’re comparing elephants to oranges.
The point being that wifi doesn’t scale, it’s not designed for many users.
A decade of tech conferences (my first WiFi kit I bought for about a grand in 1998) have convinced me that it’s consumer technology. Not designed fir more than 10 users per AP.
Matt, I was not referring to CDMA and neither were you in my opinion. What I was referring to is that the available bandwidth on wifi is shared between all users similar to contention on wired Internet access.
Also wifi uses “single carrier direct sequence spread spectrum” or “multi-carrier Orthogonal Frequency Division Multiplexing” while CDMA is similar to DSSS it is not the same.
Lastly while 1998 vintage wifi hardware was not able to scale past 10 or so users quite a bit of the current hardware is. The main limitation is that the available bandwidth decreases as the number of users increases. radio chipsets and wifi operating systems are well able to handle up to 100 concurrent users in some cases.
The CDMA I refer to is Collision Detect, Multiple Access (I admit I should have used CSMA/CD). This means the number of users drastically reduces bandwidth and it only gets worse when you permit backward compatibility. This is not the same as switched IP networks which are used in a backbone – you can’t compare packet switched networks with multiple users at the bank end of a router.
I’ve been going to commercial conferences for IT and networking with WiFi since 1999 and never found a conference that did it well. Coincidence? I think not.
WiFi: great for home.
I was looking at it from a users perspective. Users don’t give a damn if it is packet switched or something else.
But yes, you are right in saying that multiple users reduce the amount of bandwidth available per users and also that a single 802.11b user on an 802.11g router will bring all users down to 802.11b speeds.
But only on a bog standard ap/router. If you implement bandwidth management you can make sure that every user gets the same amount of bandwidth irrespective if they are the only user or number 20. Intelligent/mananged AP’s can also handle p2p traffic perfectly by only allowing a certain amount of available bandwidth to p2p traffic. I do NOT believe in blocking certain services.
As for your experience with wifi at conferences; you’re absolutely right and it confirms my statement in my blogpost. However the problem is not in wifi hardware in general. It lies squarely with badly managed wifi using unsuitable equipment.
They don’t care and they don’t need to know. But someone who designs networks needs to know the difference.
Again, every IT conference I’ve ever been to has had shit WiFi.
Coincidence? No.
test