Having a Web Summit? Get some decent Wi-Fi! It’s never a good thing for conference organisers when their Wi-Fi goes down – but it’s even worse when you’re positioning yourself as the première European tech conference of the year, Web Summit. Attendees were pretty miffed as Dublin’s massive 22,000-strong event kicked off today and the Wi-Fi at the RDS venue promptly …
It was a total disaster the last time as well. I think it ought to be called the Offline Summit, but credit where credit is due: by running this thing only once a year nobody recalls what a mess it was last time.
Sounds like a business decision paid off again: why spend any money on a decent setup if they sell out anyway?
It could change if anyone with a stand starts reclaiming their money, or at least a substantial part of it. It will only change NEXT year if this year does not generate a profit and given that this is a repeat performance, I'd pay no attention to apologies. Apologies are cheap.
Probably need to add some clarification about the impact: the problem was not just the public network. Imagine having paid quite a bit of money for a stand, and then having to use 3G to demonstrate your product or service because the most fundamental thing of a WEB summit doesn't work..
You seriously think that a typical nearby phone cell is going to offer usable data to 22 thousand delegates?
That is why venues charge a lot for good wifi (though clearly the 2nd part was missing here) as you need a lot of coordinated access points and serious back-end capacity. You know, simplistically 22k users trying to get a miserable-by-3g-standards 100kbit is going to peak at 2.2Gbit/sec for the broadband link out.
@PC
But what if, per chance, there are brief intervals when not precisely 100.0% of the "22k" attendees are all using their mobiles at exactly the same time? For example, perhaps a few of them are participating, I dunno, in the conference. And, for example, the fact that 95% of the people will have never considered the possibility of paying for a generous mobile data plan, leaving the cell site(s) wide open for the few willing to pay for a data plan. What about that there could be more than one mobile service provider in that area of the world? :. Your analysis is rubbish. Off by a couple orders of magnitude. This isn't the first time in modern history that 22,000 people have attended an event.
The point that whistled over your head, is that I was not attempting to solve the problem for the cheapskate unwashed masses wandering around looking for free wifi, but simply to mock them. Too many people with expensive smartphones that are simultaneously too cheap to have a generous data plan.
Downvote this if I've described you. LOL.
>You'd think that anyone interested in technology would have their own mobile data plan and a smartphone of their choosing. (JeffyPoooh)
The RDS is a conference venue, it offers a basic (1 mbps) free service to attendee's so they can check email etc. the Web Summit is an international gathering, so we can assume that a large number of the attendees are not from Ireland and hence would appreciate the free WiFi, particularly given the cost of roaming data...
All rather irrelevant arguments. They known in advance what sort of crowd volume they are going to get (tickets & stands are sold in advance, and it's not the first time this event has been on), managing network contention is not exactly a new science and a ticket price close to 4 digits gives you PLENTY of margin to draw in some good technology to make this work, even after you pay for the halls and all that's in it and staff etc. There is simply no excuse for this.
My own experience showed a number separate issues:
1 - inability to join either the public or the stand WiFi (I had access to both)
2 - inability to stay logged in (tried with 3 separate types of devices, just to be sure)
3 - when logged in, flashbacks to the 90s 28.8k modems because of network speed
Personally, I would have used an entirely separate network and uplink for the stand network, but given that the issues were on both networks it seems reasonable to assume that that wasn't done either, so there wasn't even any redundancy planned in.
You haven't done your best if you have all 3 issues - again. There is really no excuse.
Hmmm, Conor claims 'websummit has moved to Starbucks down the road'. I hope he doesn't mean the one that Google Maps shows as across the road from the RDS, as it doesn't exist, nor the one at the Oval round the corner that seats about a dozen and has very intemittent wifi. His acceptable 'down the road' radius may be a little large for most people.
A venue that probably never has to cope with that many geeks / devices connected at the same time normally cannot cope.
Personally, I wouldn't want to connect to that WiFi with that many geeks there anyway... I bet all is not as it seems, I will stick to 2/3/4G thanks all the same.
I don't think any reasonably-priced infrastructure could deal with the sudden influx of 50,000+ devices (a modest average of 2.5 per attender; it's probably more like 3.5) all opening twenty or thirty connections to various HTTP-borne services.
And, because this is the Web Summit, I suspect many of the cries of "network is down" are really "network is up, just slow, and I'm too impatient to let the sessions establish"
A venue that probably never has to cope with that many geeks / devices connected at the same time normally cannot cope.
- Tickets are sold in advance, so are stand concessions so there is a volume you can anticipate
- This is not the first time this event has taken place, so the problems are not new either.
Especially given the latter, having the EXACT same problem again is just poor show. If I was running this event, I would have called up some seriously heavy players and dared them to do it better. It's Dublin, for christ's sake, everyone knows everyone there. Don't tell me there isn't a clued up tech setup there.
>so there is a volume you can anticipate...
As always with live events, you don't actually know if the technology will work until afterwards, particularly as the simplest way to test the network is to run an event, such as Web Summit, that is likely to push the technology.
It will be interesting to see if anything gets released about the actual causes of the outages. Comments I've seen have point to:
1. Firewalls
2. "unprecedented wifi density compared to similar European tech events."
Any one know who's kit is being used by RDS?
This venues wifi falls over for a 2000 attendee political conference - cue me getting called by every journo in the station unable to file copy. How anyone even vaguely aligned to the media or tech world in the country didn't know this astounds me as it's the main venue for Big conferences - the much better Convention Centre being a tad small.
Last big party conference there was ample 4G capacity but I expect that as time and device support marches on, it won't do either.
I think there are not enough swiss watchmakers for that.
Back in the late 20th (not 18th) century, big iron vendors used to have exhibitions like DECworld/DECville which used more than enough electricity to bring down any venue's power supply (Palais des Festivals et des Congres in Cannes, for example - you've seen it in the Film Festival on the telly). So the organisers trucked in their own on-site power generation and distribution capability for the duration of the show. The setup period, prior to the private power supply, used to get interesting - in the basement, with insufficient electricity for aircon, and barely enough for the computers.
Maybe the Web people could truck in their own network for the duration? WiFi wouldn't be the obvious choice for backbone capacity, for reasons which are already noted here (there are inherently only so many GHz available for WiFi and expecting to use that limited bandwidth to service both vendors and visitors isn't really a bright idea. Cables might work for vendor stands. Maybe that was part of the picture; the article doesn't say.)
Or is reliable service of little interest to marketing hacks and presentation layer people?
Wifi is just plain bad for crowded venues. Why?
The *original* wifi spec had an option that was called "point coordination function", the access point itself would coordinate access to the given channel between itself and everything associated to it. This option has never actually been implemented AFAIK; so, instead, the access point and stations attempt to wait for the channel to be clear then transmit. On a busy channel this means collission city, frequent collissions mean a increasing fraction of the time on the channel transmits no useful data, and eventually you get congestion collapse (connections retransmit and time out without useful information making it through.)
Wifi also has no power control. In a dense environment with many APs, the power on each access point should be turned down since the goal is no longer maximizing range, but covering a specific area with different APs covering the neighboring areas. But the APs typically have no way to tell the client to transmit at reduced power; so usually they don't, they broadcast full power and crap up the channel for everyone else.
Nevertheless, the rudimentary fixes are: 1) Use as many channels as possible (5ghz channels as well as 2.4ghz). 2) Set power control on APs relatively low, since you have many APs and want to reduce overlap. 3) To the extent possible, keep APs on the same channel as seperate as possible. 4) High end APs have various other black magic, best practices, proprietary options, and so on, to try to maximize throughput in a heavy usage environment.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
