[ale] Source for flat-panel displays for a school

[Jim Kinney]

Old machines have fans ready to die. Some brands have more than one.

For medium to large scale roll outs, the thin client model can't be beat
for price, installation cost, ongoing maintenance.

The next phase of a roll out like APS would provide something like a
chromebook, some can be purchased new for under $200, with some network
connection wizardry and a spice client that connects to a central Ovirt
cloud providing a full Linux desktop. This will still have battery issues
as an ongoing maintenance issue. Most of the tiny cpu laptops are also
fanless and only use SSD storage. What's missing in these is a sturdy case.

Ok. Enough OT discussion. As far as getting used lcd displays goes, I don't
recommend even looking. As Jeff pointed out, older ones with non-led
backlighting are crappy and heading to an ill-timed death. As far as I've
seen over the years, lcd monitors are used until they die or become too
horrid to use. So finding usable used ones is highly unlikely.

That said, the best place to find bulk used gear is ebay. Business hire
teams to scavange out old gear and those teams dump batches on ebay.

With gift of 59 PCs, I would plan on about 10-15 being unusable and would
set aside another 15 as spares. So only 30 machines to deploy. Microcenter
has cheap new monitors at $70 and a handfull of "refurbished" ones for less.

The most important thing learned from the APS project is what level of
computer use has a measurable, positive impact on student education. The
parents that spearheaded the thin client process did some good analysis of
student performance as they added systems to the classrooms. The tipping
point was 3:1 students to computers. It basically requires students to use
the computers for 2 hours a day, every day to make a change in academic
performance. A computer "lab" that students use an hour a week is a waste
of time. Might as well give them to the office staff and teachers.

So before a computer rollout, the question must be asked "How will they be
used?". APS showed clearly that the teachers that "got it" instantly
incorporated the computers into the daily process and their kids benefited
from looking up everything (universal library) and creative projects
(documents and presentations first, video later). Other teachers used them
as typewriters (old computer lab methods) and showed little student
improvement. Some teachers used the included games (mostly puzzle games and
some math "action" games) as reward time for students. Those students
focused on their desk work more to get the game time. Academic improvement
in that group was also noted.

The most important thing we heard from teachers was anything brought into
the classroom must work all the time with no failures in front of the
students. If a single machine dies, that student sits idle. Chalkboards
don't fail. A tiny bit of teacher training to show how to connect the
systems if a student unplugs it and how to force one to reboot with the
understanding that nothing could fail software wise  on the classroom
machine went a very long way towards teacher acceptance.

Computers are useful in educational environments as long as they are
appropriate in the setting and used for something other than drill and test
exercises. If old systems are used, hot swap spares must be ready to be
drop in replacements and staff must be ready to run to do the swap. With
2000 thin clients rolled out, we saw around 10-20 fail within 10 days. The
servers had a horrible failure rate due to vendor/APS corruption. A server
failure was a school-wide nightmare.

Networking is also fun. The fans in most switches are horrible noise. Wads
of cables are a mess. Kids unplug things and reconnect things wrong. A
single looped switch can bring down the entire building.

The APS project was an outstanding success despite all of the issues. The
followup system wide rollout was a total, corruption induced failure. Jeff,
Aaron and I were not involved in that debacle. If we had been, it would
have worked. I still giggle a bit that a trio of motivated Linux geeks
performed a task that IBM couldn't replicate even with our recipe.

On Oct 5, 2016 6:30 PM, "Jeff Hubbs" <jhubbslist at att.net> wrote:

> Especially after the Atlanta Public Schools masterpiece/debacle (depending
> on how you look at it) that Jim, Aaron, and I did a few years ago, I would
> advise this school to tread very, very carefully. That donation from SSA
> can turn into a leaden albatross very easily - perhaps so easily that there
> is no way to avoid it.
>
> The basic problem always seems to come down to trading up-front cost and
> labor avoidance for some combination of up-front and follow-on labor cost.
> It's the latter that really hurts because it accrues over time.
>
> Let's assume all the PCs are the same, were procured at about the same
> time by SSA off either GSA or some other blanket procurement, and they all
> function perfectly when you receive them. If you deploy so as to depend on
> the disk drives in each one, the far end of the so-called "bathtub curve"
> distribution of drive failures is out there waiting for you and when you
> arrive at it, you'll be sinking a lot of labor into drive replacement and
> system recovery that you could avoid by designing to get rid of the disk
> drives (a separate calculation would help to determine if it is worth it to
> remove or just unplug the drives; in this day and age I doubt it would be
> worth it to make a single massive array out of them and use the array for
> infrastructure).
>
> We deployed 2200 thin-client seats in seven schools, up to 500 per school,
> and we were trying to stick to one app server for every 100 seats (we would
> likely have that number at 400-500 today). Even at just 53 machines it
> would definitely call for some sort of netboot/central-app arrangement like
> the one we built or the labor intensity would just be ridiculous.
>
> But more to your original question, I would first compare the goods and
> labor cost of monitors against that of 53 all-up monitor/thin-client combos
> before even agreeing to take the donation. When we tried to bid for the job
> of building out the rest of the APS district, I worked out many of the
> details that would be associated with assembling monitor/thin-client combos
> at industrial scale - tens of thousands of seats' worth. When you do that,
> you have to sweat the labor intensity of every single step in the process
> (e.g. screwing monitors to thin clients, loading trucks) and plan things so
> as to keep costs from scaling with volume to the greatest degree possible.
>
> Recall that the CCF backlights in monitors dim over time and if you source
> monitors that are already five years old or older, that's a lot of CCF life
> that's already behind them. If you were to go for refurbs, you'd want to
> know if new CCFs were part of the refurb.
>
> The power consumption of an idling tower PC is a lot greater than that of
> a thin client; the latter of which only needs to run a little dippy CPU and
> not much RAM if all that it's really doing is running Xorg and perhaps a
> few other daemons to handle sound and so forth. Even the power supply fans
> in PCs, if you get 8-10 of them in a classroom, make a fair amount of
> noise.
>
> On 10/4/16 9:04 AM, Vernard Martin wrote:
>
> An acquaintance of mine that runs a small private school has recently
> given the opportunity to acquire 59 computers from the Social Security
> Administration but no monitors. I'm not sure that they have an OS currently
> loaded on them and I am, of course, strongly suggesting that she go with an
> open source Linux-based solution either way. However, their immediate needs
> are a supply of flat-panel monitors or the units aren't very useful.
>
> They are based in the Atlanta area. Since I'm no longer living in Atlanta,
> I'm doing all this remotely. Its a challenge as you can imagine :) Does
> anyone have any suggestions on where I can call around looking for
> donations? And barring that, does anyone know where I can purchase around
> 60 refurbish monitors in bulk? I figure 14" 1024x768 would be the low-end
> units that would work, especially if they are being donated. Anything less
> just might not be worth it in the long run.
>
> Any leads would be appreciated.
>
> Vernard
>
>
> _______________________________________________
> Ale mailing listAle at ale.orghttp://mail.ale.org/mailman/listinfo/ale
> See JOBS, ANNOUNCE and SCHOOLS lists athttp://mail.ale.org/mailman/listinfo
>
>
>
> _______________________________________________
> Ale mailing list
> Ale at ale.org
> http://mail.ale.org/mailman/listinfo/ale
> See JOBS, ANNOUNCE and SCHOOLS lists at
> http://mail.ale.org/mailman/listinfo
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.ale.org/pipermail/ale/attachments/20161006/b919c370/attachment.html>