I started using Linux in the early days. I do
remember reading Linus's first mail while at school. My friends and I hung out
in a computer room that had about 12 IBM PS/2 systems (386/sx) running AIX (at
the time AIX 1.2 still existed on the ps2 as well). We were looking for source
code for an OS and had actually put together, I believe it was, about $400 to
buy the BSD source code tape from BSDI.
Anyway long story short, the BSD tape didn't
arrive for quite some time and Linus's mail was perfect timing so we pulled
Linux in and that was version 0.95+. First we replaced AIX with
Linux on one PS/2 system, then more systems followed and after a while the
entire PS/2 cluster became a Linux-based PS/2 cluster. Slackware was the first
actual distribution that we then pulled in, tons of floppy disks... good times.
What changes have you seen from the early days of Linux to today?
Well, Linux grew up; it was a hobby project which
turned into an obviously hugely successful software product. It's really quite
amazing. It runs on so many chips/architectures, the download used to be a few
100K and now is 70M+ in size. Fascinating really. I think Linus did an amazing
job by managing the project so incredibly well but at the same time allowing
others to do stuff with a great amount of freedom. It prevented forks from
happening early on. Many open source projects end up forking into different
trees because people disagree but Linus has always managed to keep Linux as a
nice single tree moving forward.
The other thing that changed is the number of
people, the early days was a smaller group. However, even today, from what I
can tell there still is a very similar core team that contributes to Linux
often times still the same names.
How important is Linux in today's datacenter?
It's become a core technology, every datacenter
probably has some amount of Linux running everywhere in the world. Whether it's
a regular server or an appliance or embedded device. So it's everywhere. I
don't think there's any particular place that Linux doesn't play a role in
devices that need an operating system. Obviously the server market is visible a
good use case but in terms of units the mobile space is way larger. People
don't see that because it's hidden inside. I am sure everyone has some Linux
running at home, even people that do not have a desktop computer but they have
a Tivo or so.
This also implies that software is ported or built
on Linux by most software vendors and that causes it to be a great ecosystem.
It really is very pervasive today and that includes all areas, test and
development systems, product systems, mission critical, clusters, etc...
How does Oracle invest in Linux and open
I want to highlight a few different areas here:
We port and develop most Oracle products on Linux
and Solaris as a base-development platform. So by building products on Linux we
obviously ensure that our customers can happily deploy on the OS. There is a
big benefit to Linux here that often gets ignored or isn't really appreciated
and that is testing. We have 1000's of servers running Oracle Linux that are
used to test our various products. This helps test the OS as well, during our
Oracle regression tests we run into OS related bugs, we triage and fix them.
There are very, very few companies out there which such a large development
environment and QA farm. QA is an important part for any product to improve and
it's a great, underappreciated contribution.
Oracle started somewhere in late 1990s to port
products to Linux, Oracle database 8.0.5 was first released on Linux in 1998.
We have the Oracle Linux development team which
does a lot of contributions to Linux. Btrfs, all the Linux NFS work, a ton of SCSI
work, etc. there's a long list of stuff that goes back into the mainline tree.
There are two objectives for the Linux development team 1) Help make Linux a
better OS in general. Features needed to continue to be competitive with other
operating systems, Btrfs is a great example. These features might not always be
related to what Oracle products need. 2) Find places where we can improve Linux
to help run Oracle products better and faster.
This work started around 2001 and has been
increasing ever since. The people in the team are part of the Linux kernel
development community like any other kernel developers, same process, same way
of working. It's very smooth and now that we have the Unbreakable Enterprise
Kernel for Oracle Linux, we have a great vehicle to actually show this off more
and we have the ability to get the work we do in the hands of customers
directly. I believe this will help make people realize even more the amount of
effort we put into Linux.
Use of Linux as a deployment platform. We have
1000's of servers in house running Linux, test, development and production just
like we do Solaris. It shows that we are not just telling others to do this but
also do it ourselves; eat your own dog food kind of thing! We also have, and
continue to, offer Oracle Linux in our engineered systems such as Oracle
Exadata and Oracle Exalogic.
Where do you think Linux is going to be in the next 20 years?
It's easy to predict the past, hard to predict the
future! Clearly the mobile and tablet space uses a lot of Linux, so the
embedded market is full of Linux devices, the server market is full of Linux
servers and that will continue to grow. There are always new features to work
on of course and as chips get more cores and threads and servers get more CPU
sockets and memory scaling up will continue to be something to work on and
improve. I am not sure how well Linux on the desktop is going to evolve.
People have been trying to predict that "this" is the year of the
Linux desktop for quite some time now, tablets seem to be making inroads into
that space to a certain extend, so maybe Linux's year of the desktop ends up
being Linux tablets and mobile devices that are slowly replacing desktop use.
Now that Linux is twenty years old, I suppose Tux is twenty too!
Heh! Tux is timeless. As far as I can remember it
has looked the same, hasn’t aged a bit!