[IAEP] quality of USB keychains for SoaS, was Re: Sugar Labs Mission & The 6 lesson Schoolteacher

[D. Joe]

On Mon, Apr 24, 2017 at 01:03:09PM -0500, James Simmons wrote:

> I tried Sugar on a Stick a few years ago and found that it didn't work all that
> well. Maybe it has improved since then? I think the problems were related more
> to the quality of the USB storage than anything else.

I've found definitely that quality of the USB drive is a factor.

For only a brief time, for reasons I won't go in to just now, I and the
undergraduates in my class had access to computer labs in which the BIOS was
not yet locked down to prevent booting from USB.  I found this out but
hadn't requested new USB drives to give them, so I asked each student to
bring one one of their own in so I could load SoaS on it.  I then ran the
(now deprecated) livecd-iso-to-disk installer to make a (somewhat)
persistent live USB SoaS instance on each (a process whose tedium I was able
to alleviate by stealing into the lab at off hours and running it in
parallel across several systems).
 
The speed at which the installation ran on these drives varied dramatically
not just between USB 2 and USB 3 capable drives, but also from manufacturer
to manufacturer.  I was able to see this pretty directly because performing
this installation process was when I had the most direct access to the
drives in operation.

I then handed the students back their drives and had them boot lab machines
into SoaS from them.  Our aim, then, was not to benchmark USB performance
and so I lost track of how much of a differential there was in the overal
SoaS experience.  My goal then was to give them the opportunity to use SoaS
for getting greater experience with Sugar, for development and testing
activities on newer hardware, and in general complementing the XOs they'd
been lent.  If they had their own hardware and were willing to boot it from
USB drives, that was an option.  

Beyond that, I felt it would be instructive for students who'd never seen a
dual-boot system in action see what the process entailed.  It's one thing to
hear about it or read about it or (these days) see a video demonstration,
but quite another to direct the process oneself, hands on.  Even though the
process of using livecd-iso-to-disk was fiddly and tedious, it was well
worth it to know this was the first time several of the students had seen
such a thing done, and could see the difference in how Sugar performed on
the XOs versus the beefy high-memory, i7-based machines in the labs.

That said, flash media generally, including USB drives, have supplanted
nearly every other form of removable media, but unlike magnetic floppy disks
and optical media, they are not passive devices.  They share with modern
hard drives the inclusion of, from a historical perspective at least,
significant on-board computing power and all the attendant complexity that
carries with it, see, eg:

  https://www.bunniestudios.com/blog/?p=3554

"The embedded microcontroller is typically a heavily modified 8051 or ARM
CPU. In modern implementations, the microcontroller will approach 100 MHz
performance levels, and also have several hardware accelerators on-die. "

As with so many of our computing devices now, consumerism dictates that we
be encouraged to ignore what goes on behind the curtain and to regard them
as passive, cheap, simple, straightforward, reliable, and as easy to use. 
Were our job selling technology, we could leave it at that.  Given our
goals, though, we should keep this complexity in mind, and stand ready to
help Sugar learners better grasp some of these nuances as their
understanding of the technology they use develops.

In the meantime, keeping in mind these complexities can also help us
appreciate why developing and maintaining simple recipes for their use isn't
as straightforward as a consumerist mindset might lead some to think.

-- 
D. Joe