[PLUG] Report on OSCON 2014
Bill Kielhorn
kielhorn at amerimailbox.com
Sun Jul 27 23:55:31 UTC 2014
Report on OSCON 2014
I attended two and one-half days of the OSCON-2014 conference after
being selected as winner of PLUG's book review contest. Thanks to
O'Reilly Books for sponsoring the ticket and thanks to Michael Dexter
for organizing the contest.
I made full use of the ticket by attending events in each time slot
allowed by the ticket. That included three plenary talks, and 15
session talks as well as lunch and access to the exhibit hall. It was
at times hard to decide which session talk to attend because there were
typically 18 distinct choices in each time slot. For the most part I
attended talks about Functional Programming since I wanted to learn more
about it and quite a few of the talks concerned the subject.
I saw talks about functional languages Haskel, Clojure, and Erlang, as
well as general talks about functional programming. The main advantages
over the more usual imperative style of programing are: 1) Fewer lines
of code with consequent improvement in correctness. 2) Lazy evaluation
for improved efficiency. 3) Robust out-of-order execution with
simplified multi-threading. Nevertheless, imperative style usually wins
in speed contests.
My feeling was that the technical level of the talks was rather low and
I was not alone in the feeling. Several people I met said they would
not return and one company CEO said he would withdraw his sponsorship
because of this. The high ticket cost was also a factor. Nevertheless,
for me OSCON was useful as a way to pick up on the current trends and
interests in open source and software development.
By far, and unexpectedly, the best talk I heard was by CERN's IT
manager. CERN is the European particle physics laboratory that recently
identified the first Higg's boson. The lab's data management task is
enormous. They currently have 100 PetaBytes of legacy data in various
formats that must be maintained and be accessible for the next 20
years. In addition the current rate of new data acquisition is 35
Petabytes per year. (One Petabyte is one million Gigabytes.)
To avoid being overwhelmed they decided to go to open source. They now
hire open source people, they use open source products, and they
contribute code to open source. They have virtualized their system,
they use OpenStack, Puppet, Ceph, and other open source products. In
twelve months time, Puppet will be managing 100,000 cores in their system!
Personally I know very little about any of the products used by CERN,
but the interesting thing to me is that open source was the only
available solution to their problem! They are not using open source to
be nice or to give back. They are using open source because it is the
only way forward! Moreover, it is working! Open source is solving
their problem!
Thanks again to O'Reilly and to Michael, and I will be happy to discuss
OSCON further at the next PLUG meeting.
Bill Kielhorn
More information about the PLUG
mailing list