We didn't pick the single channel architecture for two reasons:
-IEEE 802.11 was designed with small cells in mind, that's the
standard! Though, creating large one channel cells works with 802.11
hosts,
nothing guarantees that it will work in the future. Also, as Bruce
mentioned, the one channel design is not very tolerant to
interferences from neighboring 802.11 networks that one doesn't
control (if your one cell is on channel 1 and a channel 1 neighbor
is interfering, how do you adapt? you spend your time adapting,
comparing to a multi-channel design with controllers that will adapt
for you!
The bursting capability of 802.11n is also diminishing the Time
Division advantage of a 1 channel design!
-Second: at the time of our study (18 months or so ago), Meru didn't
have an IP mobility solution. Large layer2 networks was their answer.
Been there, done that,
it scales up to 2000-3000 users Max and the Wireless quality lost by
Broadcasting is huge.
Philippe Hanset
Univ. of TN
On Jul 30, 2009, at 8:15 AM, Osborne, Bruce W. (NS) wrote:
Jason,
I wholeheartedly agree. We here at Liberty University spent a year
evaluating wireless & NAC solutions. We chose to move from Cisco
“fat” APs & Clean Access to Aruba’s wireless & ECS NAC solutions.
The real challenge is in dense environments. Meru’s “single channel”
becomes “channel stacking” aka multi-channel to provide additional
bandwidth. You then have the client roaming issues again. Also, you
cannot “steer” clients to load balance the clients across available
resources. Aruba’s ARM 2.0 has many options in this situation and
solves many of the issues that Meru’s architecture solves.
With a single channel architecture, you are “stuck” if some
interference appears in that RF range. The system may be able to
change to another channel, but that WOULD CAUDE *all* the clients to
roam. In a multi-channel architecture, only a small number of
clients would be affected.
There is obviously a reason why Meru is the _only_ vendor with
single channel. All the others (including the largest players) use a
multi-channel solution. If Meru’s solution is so great, you would
see others with single-channel too, even if they needed to license
technology from Meru.
Bruce Osborne
Liberty University
From: Jason Appah [mailto:jason.ap...@oit.edu]
Sent: Wednesday, July 29, 2009 1:44 PM
Subject: Re: Single Channel vs Multi-Channel Architecture
I agree wholeheartedly, the Aruba ARM works quite nicely, recently
the neighboring hospital turned up its radios, and ARM switched us
out without missing a beat. We reviewed Merus’s devices and liked
the approach, but were less than wowed with the completeness of the
feature set.
In the end we choose Aruba for four reasons:
Price – pretty self explanatory
Performance/deployment - (this was identical in most and in many of
our use cases better than Meru)
Feature Set – Aruba has obviously spent many hours actually
listening to and implementing user centric changes, I don’t know of
a more feature rich wireless solution
Support – Aruba has in many occasions been proactive, where I have
posted a question to this forum and others to actually go out of
their way to contact me to help me fix a problem, in some instances
where the problem wasn’t even Aruba’s at all…
We haven’t looked back.
From: The EDUCAUSE Wireless Issues Constituent Group Listserv [mailto:WIRELESS-LAN@LISTSERV.EDUCAUSE.EDU
] On Behalf Of Ken Connell
Sent: Wednesday, July 29, 2009 9:15 AM
To: WIRELESS-LAN@LISTSERV.EDUCAUSE.EDU
Subject: Re: [WIRELESS-LAN] Single Channel vs Multi-Channel
Architecture
I don't have much experience with a single channel deployment, but
without even getting into vendor preferences or specifics I can't
see how a single channel can gain any perfomance in such an
unpreditctable and dynamically changing environment as far as other
devices, and wireless networks that will come and go probably a
daily basis with little or no control.
The channel you decide on today, may not be the best suited channel
tomorrow, and if you then need to make a change at that point, then
you've jsut come full circle and are right back where you started.
In my opinion it just makes sense to go with an automated RF type
deployment (Aruba ARM for us) and be able to sleep at night ;)
Ken Connell
Intermediate Network Engineer
Computer & Communication Services
Ryerson University
350 Victoria St
RM AB50
Toronto, Ont
M5B 2K3
416-979-5000 x6709
From: Ryan Holland
Date: Wed, 29 Jul 2009 09:04:34 -0400
To: <WIRELESS-LAN@LISTSERV.EDUCAUSE.EDU>
Subject: Re: [WIRELESS-LAN] Single Channel vs Multi-Channel
Architecture
...interesting thread...
When we were making our decision 3+ years ago, we discounted Meru
primarily on scalability information in their RFP response. So
unfortunately, we did not get a chance to bring them in for a demo.
I am still quite skeptical about a single-channel architecture but
believe I understand why it is promoted: to assist devices in
roaming by creating a seemingly single BSSID. However, once we see
more devices supporting standards such as 802.11k and 802.11r, such
efforts, to me, are negated. Again, however, I have not had the
opportunity to play with this gear, so [disclaimer].
We have been deploying Aruba for sometime and have learned a great
deal about their technology, so I will caution the trusting of
intelligent radio management solutions. Instead, I would suggest one
utilize this technology while maintaining a tight supervision of it.
Using Aruba with whom I am most experienced, their adaptive radio
management (ARM) is quite powerful, as it allows for dynamic
remodeling for channel and power based on the environment. This
means that as other building tenants bring in their own wireless
systems, our network can modify its channel configuration
accordingly. Also, in the event of an AP failure, adjacent APs will
likely perceive a lower aggregate signal strength of neighboring
APs, boost their power, and thus help alleviate the loss of coverage
from said failed AP.
The reason I cautioned earlier is that many administrators simply
"turn on ARM" and leave it. Doing so is assuming the defaults are
applicable for all environments, which I would argue is not true for
most educational institutions. Examples: the range of chosen
transmit power is likely too expansive; the noise threshold at which
an AP would change channels may be too low, especially for "research
areas" like Illinois mentioned; the target coverage index may be
too low for densely deployed installations or too high for sparsely
deployed installations. Aruba is great in that administrators can
configure different ARM profiles for all these different
circumstances and use them suitably. But again, to just turn it on
and expect it to "work" can lead to false assumptions.
I would also add that there are still a lot of those that state
static channel/power assignments is the best way to go. While I
would agree that is true assuming the environment is identical at
installation as it was during survey, it is incredibly likely that
the environment will change and therefore negate the initial survey.
Because our environments are largely unpredictable, I find a dynamic
solution to be preferable. Now, if we had complete control over RF
across campus, my opinion may be different.
(Oh, and because people seem to be concerned with these sorts of
numbers: ~5,000 APs, ~40 controllers).
==========
Ryan Holland
Network Engineer, Wireless
CIO - Infrastructure
The Ohio State University
614-292-9906 holland....@osu.edu
********** 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/.
