[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: [K12OSN] wifi wireless k12ltsp support status



Let's also add to this the fact that the TCP ACK's would create all sorts of collisions, thus reducing the "real" throughput of wireless (remember, it's a hub) even further for thin-client purposes.  But it's actually worse than that, since wireless isn't CSMA/CD, but rather CSMA/CA, like the old LocalTalk.  CSMA/CA is even slower than CSMA/CD.  So, it's even *worse* for thin-client purposes than already outlined.  And yes, that applies to the new "N" wireless, no matter how new and "sexy" you think it is.

Please, *PLEASE* don't screw up a fine system like LTSP by trying to run it on wireless!

--TP
_______________________________
Do you GNU?
Microsoft Free since 2003--the ultimate antivirus protection!


James P. Kinney III wrote:
Wireless has a place when mobility trumps cost. Wireless is not, and
will not in the foreseeable future be a viable transport method for thin
clients. It may become useful in the chubby client (local apps) as they
require less constant bandwidth.

A $70 AP is quite limited in power and connectivity. It still requires a
copper connection. And each client machine requires a $30 NIC instead of
a $10 NIC.

A 54Mbps AP provides 37.8Mbps of usable client bandwidth. Jumbo frames
are not an option in the wireless specs. Since a current thin client
requires a minimum of 7Mbps with no sound or streaming video that puts
the maximum client count per AP at 5. 

With a measured benefit ratio of 3:1 students:clients that puts the max
class size at 15. In Georgia the max size is 24. So that puts the thin
client using wireless into the 4.8:1 ratio which has been shown to be
ineffective. 

So a WAP ($70) plus 5 clients ($30 *5 = 150) requires $220 and is not
expandable and won't provide enough classroom exposure to make a
statistically significant academic benefit.

Most systems ship with an onboard NIC. An 8 port Gig switch is $70 (A
10/100 is $30). A managed 8 port is $110 - retail. So a wire is already
run for the (theoretical) WAP, plug in the switch (save $210 per
classroom - $30 * 7 clients) and get the 3.4:1 back. And at only 49Mbps,
there is expansion room to support streaming video (limited) over a
100Mbps wire for 7 clients or add more clients using a Gbit feed from
the upstream switch.

Yes. Wireless is "sexy". It's mobile. But the reports are in from all
over corporate and college campuses - it's EXPENSIVE to set up and
maintain well.

My money is on the wire.

On Tue, 2008-10-07 at 08:54 -0400, HOWARD DANIEL-GVN674 wrote:
  
IMHO, wireless has it's place in schools and will definitely get better
in many respects over time, as well as cheaper.  I agree with Jim that
when doing an enterprise server model, where the wireless links to
classrooms would be between servers and clients, wireless is a very bad
idea, and likely to stay a bad idea until we get speeds in excess of 400
Mbps, since real speed over wireless is usually about half of the stated
PHY layer speed, so 54 Mbps is really about half the Mbps in reality.

But wireless links to a classroom server are still viable and cheaper if
they were only more user friendly in Linux (a point to Jim's $$$ labor
cost), since $70 for an access point and $30 for a wireless NIC is still
less than half the cost to wire a classroom (usually at least $200), and
subsequent rooms nearby would be only $30.  But getting wireless to work
in Linux is not always easy, which is I think Jim's point, due to things
like chipsets changing rapidly and not being supported in Linux,
although I suspect it will get easier over time as well.  And for rooms
with older wiring, or wiring that runs by noisy electronics, that the
school won't have the budget to rewire anytime soon, wireless is useful
as a backup and alternative.  And wireless to classroom K12LTSP servers
doesn't require batteries, of course.  If we could generate a decent
list of supported wireless NICs for classroom Linux servers (and keep it
up to date) I think wireless would have a strong place for K12LTSP
deployments using the classroom server model.  Plus, if I were in a
school that had a really hostile IT department that said no way to
K12LTSP on their network, one of the ways to set it up w/o them and
their network being affected would be a classroom server with a 3G
broadband wireless access card.  I've seen folks with Linux laptops with
3G access, so I know it's at least possible.  Not the best long term
solution, agreed, but a way to get a foot in the door perhaps.  And if I
ever had the time/resources to really evangelize this, a Laptop server
with 3G Internet (as backup to a wireline) connected to a switch and
thence to a handful of netbooks/UMPCs would be a great roadshow kit to
cart around to show schools.

The point about laptops is important: what makes wireless really
critical is when laptops become cost effective for schools (note when,
not if).  Already we now have companies breaking the $100 barrier for
UMPCs, vis a vis, the HiVision Mini Note:
http://techvideoblog.com/ifa/98-linux-laptop-the-hivision-mininote/

At less than $100, it is a viable solution for elementary and perhaps
even middle schools, and given the lack of moving parts, should be much
more reliable, albeit not as fully reliable as a desktop, but
approaching it.  Better reliability can be had if it's not a 1:1
program, rather a 3:1 or 2:1 with them if they stay in the classrooms
and thus don't get banged around between school and home.

My company has wireless deployed throughout all corporate locations so
that anytime I attend a meeting, whether I'm in someone's office or
small conference room, or board room, I merely boot up my laptop, log
on, and I'm connected.  Most of my customer locations are likewise, but
sometimes they lack an unsecured network for guests, although that too
is changing.  It is not merely a boardroom thing anymore, and I'm sure
this model will spill over into schools when the cost of the laptop
(acquisition and support) gets low enough to make them viable, which I'm
convinced will happen eventually, and I'm guessing in only a few years
now.

Best,
Daniel


_______________________________________________
K12OSN mailing list
K12OSN redhat com
https://www.redhat.com/mailman/listinfo/k12osn
For more info see <http://www.k12os.org>

    

[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]