Hi Philippe, This is a great discussion and I'm glad to see we're not the only ones struggling with these issues. I'll add my 2 cents on a number of issues brought up in the thread, and yes I think some of your accounting is a bit off.
First off I don't think it helps to have this discussion without recognizing that the life cycle of cabling is generally 15-20 years, sometimes more. There are exceptions, but this is the general practice for a number of reasons. I don't know anybody who is concerned about 11ac phase 1 bandwidth, and even the phase 2 that you mention, there is a lot of skepticism about how much it will really utilize. It is important to recognize however, that full duplex does not help you if the traffic is asymmetric, weighted towards the consumer, which is what we see. You still only get one gigabit at a time in each direction. I think it is also important to consider that the network is 100% utilized whenever there is data transmitted. The game we play is engineering so that that happens infrequently enough so as not to cause excessive queuing, or worse, excessive packet loss that would be noticed by the end user. We often see five minute averages of utilization that round out a lot of this, and it isn't a problem until there are periods of time when it happens enough to cause delay, packet loss, and retransmissions that impact the end users experience. Whether this is going to happen with 11ac phase 2 or not seems unlikely to me, but I think it is a legitimate debate. I would be concerned, except we see utilization at five minute averages that is often not much more than 25% for a whole residence of AP's. But does anyone really think it is wise to assume that an AP won't exceed a gig per AP during the lifespan of the cable? If you look at the rate of growth of bandwidth on the Internet over the past and the projections for the future, assuming they are even close, and the rate of change and innovation, I think this would be a very risky assumption to make. It certainly does seem like more is moving to the cloud. I would say we haven't found it to be cost effective, or secure enough and we still offer a lot of services on our local campus network which are faster, closer and more reliable than the internet. We also have Akamai appliances that cache Internet content and improve the response time for our users while helping cut down on bandwidth utilization. We have all kinds of programs that move large amounts of data including 3d drawings and video. If we did move more to the cloud, we would likely have to upgrade our Internet connection which is already at 6 Gigabits, and we will continue to upgrade over the coming years whether we move to the cloud or not. But I suppose if you keep your bandwidth low enough upstream, then you don't have to worry about these things at all. Right now we do indeed have 10 Gig in the core, and are putting in 10 Gig to a lot of our buildings. We are also putting in some 40 Gig to the core, although that at the moment is admittedly more driven by researchers. In theory we are still oversubscribed by quite a bit in regards to how much every user or AP can theoretically use, and that model works very well as long as the nature of user traffic is as bursty as it has tended to be thus far. But bottlenecks can occur in practice, and we watch for them so they don't impact performance. I have to say that I don't think it is necessarily a bad idea to run one cable (assuming it is cat6a) instead of two, even though we run two. And we may re-evaluate this approach ourselves as we see things develop. If you do run one Cat6a you just have to bank on the AP vendors switching to 10 Gig connectivity before bandwidth becomes an issue for you. I don't think that amounts to a big risk, but we've decided for now that given the incremental cost to pull two, we are going to do it for the insurance. One of the challenges with running two is that we will need space in closets to support the extra cable which could also be a bit of a challenge. Pete Morrissey From: The EDUCAUSE Wireless Issues Constituent Group Listserv [mailto:[email protected]] On Behalf Of Hanset, Philippe C Sent: Friday, February 07, 2014 10:22 AM To: [email protected] Subject: Re: [WIRELESS-LAN] How many drops 802.11ac phase 2 Is the main justification for two drops due to power/bandwidth/the-two? With many services and most killer apps going to the cloud, I would suspect that the bandwidth to the WAN is so limiting, that this excess of capacity on Wireless is a complete overkill (a vendor driven non-sense). Yes, those 802.11ac Phase2 APs can generate a lot more than 1 Gbps, but that's is shared bandwidth (half-duplex), and your uplink is 1 Gbps full-duplex (2 Gbps in Cisco math as we said in the old days). So, you really plan to also uplink your switches with 40 Gbps, and then a core at many times 100 Gbps, all connected to your ISP at a few Gbps... something doesn't add up here. Am I alone making bad accounting here? Philippe Hanset www.eduroam.us<http://www.eduroam.us> On Feb 7, 2014, at 9:58 AM, James Robert Kennon <[email protected]<mailto:[email protected]>> wrote: We just made a call on a new building and decided not to incur cost of 2 cables per drop at this time. Hope we don't regret it later. From: Lee H Badman <[email protected]<mailto:[email protected]>> Reply-To: The EDUCAUSE Wireless Issues Constituent Group Listserv <[email protected]<mailto:[email protected]>> Date: Fri, 7 Feb 2014 14:56:31 +0000 To: <[email protected]<mailto:[email protected]>> Subject: Re: [WIRELESS-LAN] How many drops 802.11ac phase 2 We'll be running two, until some sanity emerges. ________________________________ From: The EDUCAUSE Wireless Issues Constituent Group Listserv <[email protected]<mailto:[email protected]>> on behalf of Brian David <[email protected]<mailto:[email protected]>> Sent: Friday, February 7, 2014 9:54 AM To: [email protected]<mailto:[email protected]> Subject: [WIRELESS-LAN] How many drops 802.11ac phase 2 All, I wanted to see how many people were planning on running 2 drops to 802.11ac phase 2 access points? Currently we are just doing a one for one swap when replacing an older a/b/g AP's with 802.11ac phase 1 AP's When you have new construction, do you plan on running 2 drops so when phase 2 come into play you will be all set for it? Brian J David Network Systems Boston College <image003.jpg> ********** Participation and subscription information for this EDUCAUSE Constituent Group discussion list can be found at http://www.educause.edu/groups/. ********** Participation and subscription information for this EDUCAUSE Constituent Group discussion list can be found at http://www.educause.edu/groups/. ********** Participation and subscription information for this EDUCAUSE Constituent Group discussion list can be found at http://www.educause.edu/groups/. ********** Participation and subscription information for this EDUCAUSE Constituent Group discussion list can be found at http://www.educause.edu/groups/. ********** Participation and subscription information for this EDUCAUSE Constituent Group discussion list can be found at http://www.educause.edu/groups/.
