[Gnhlug-jobs] Jim Gettys is hiring

[Jeff Kinz]

Some of you may remember Jim Gettys.  He worked on some kind of
Windowing package for UNIX a while back, 15, 20 years ago.

I forget what the hex its called....

Anyway, Jim is working on another project now and I just happened to run
across the job listing, here it is: 
http://wiki.laptop.org/index.php/Lead_the_effort_around_system_management

His email is jg at laptop org


Here is the text: 

Lead the effort around system management

One of the major challenges we face is system management. Anytime you
start talking about tens of thousands of installations of hundreds or
thousands of machines, scale is a challenge ;-) .

What is more, we face "sometimes connected", mobile, behavior,
in parts of the world where IP addresses are rare.

And finally, we believe that collaborative systems are truly vital, and
that we therefore want end-to-end connectivity, rather than multiple
layers of NAT. Imagine an apartment building with kids from several
schools; we want them to be able to go home at night, and, without any
additional infrastructure, work together as they need to.

The conventional client/server infrastructure we've build up (as
opposed to client/server applications) will often be unavailable. We
can't count on kids being able to reach their schools outside of
class hours.

One consequence of these observations is it may be *easier* to make all
these systems IPv6 "out of the box", than try to continue the
current kludge tower of IPv4. We're still investigating this line of
thought, but so far, so good...

We also need to rethink how we configure much of our infrastructure
pieces to be much less dependent the the conventional central, always
available servers. Here are a couple examples: we want IM to always be
available; but do so even without global connectivity or fixed servers.
And every kid should be able to publish content on their laptop,
(potentially) available to others anywhere.

We want to rethink the configuration of (and sometimes modify) many of
our existing services, to allow them to be used in a much more ad-hoc,
mobile, opportunistic fashion.

Somehow, we have to get to and beyond the management scaling achieved in
Project Athena (where to this day, at MIT more than 1500 systems are
managed by just a few people), but while at the same instant, being much
more fluid and its management less centrally regimented, and allowing
for much more "ad hoc" opportunistic collaboration among people
using the machines. Thankfully, we have many more tools to exploit now;
in the mid '80s at Athena, we had to invent our own window system
(X11), authentication(kerberos), service discovery (Hesiod), IM system
(Zephyr). Our toolkit is very much larger now.

Another aspect to the problem is that there are certainly some amount of
production system management as well. The scale of the OLPC project is
such that we can anticipate needing to host large scale software and
content development projects in support of our goals.

So we're looking for someone who is truly expert and interested in
innovative system management (and is able to, at times, get in and make
changes to code in our infrastructure), whose initial goal is to help
OLPC get its own environment set up to dog-food the environment we
intend to deploy (which may be primarily IPv6, if our current thoughts
hold). This person will help flesh out this vision of systems that are
much less centrally managed, require as little training as possible, and
in which mobile people cross administrative boundaries on a minute by
minute, sometimes second by second basis.

We need someone who really knows how to exploit and configure existing
services for self management, not daunted by the idea of deploying
hundreds of thousands of systems containing hundreds or thousands of
computers, and sees this management problem as one of the keys to the
future of computing. And who knows, there may be some tools we have to
build from scratch; but we're banking on the open source and free
software community to build the tools, if experience of the last 10
years is a good guide.

An ideal person would be interested in a huge challenge, system
management at both a practical and theoretical level, be very familiar
with the state of the art networking (IPv6, mesh networking), able to
meet aggressive development schedules, have a personal passion about
making systems not require explicit management, and familiar with very
large scale systems, while understanding how to work in and be part of
the (growing) community that is at the heart of succeeding at such a
project. As you might gather, serious practical experience is also in
order.

This position will be in Cambridge, Massachusetts, and will involve
worldwide travel.

Please send mail to jg at laptop org

-- 
Jeff Kinz, Emergent Research, Hudson, MA.
Speech Recognition Technology was used to create this e-mail