|
30 minute interview
“Applying traditional UNIX methods does not make for a good fit”
 |
Michael Tiemann
Vice President, Open Source Affairs, Red Hat |
Red Hat’s Open Source maven says that Penguinistas are innovating furiously
as Linux goes 64-bit with a vengeance.
*Where’s the innovation in Linux?
People are moving away from the conventional UNIX paradigm toward stateless
Linux. On a given client or server, you have a machine without any state. It
doesn’t check your credentials using a local password file; it goes out
on the network to check your credentials. Replacing a machine has no impact;
it’s like replacing a phone. This is preferred by large enterprises.
Applying traditional methods from the UNIX world for real time performance of
applications is not a good fit. Ingo Molnar (a well-known contributor to Linux
kernel development) is making continuous improvements to how latency is handled,
how the kernel handles voluntary pre-emption. 64-bit is starting to happen.
The uptake of 64-bit Linux is driven by larger file systems and virtual memory.
Gnome, lots of progress. Things like the Tango application server. Support for
non-Roman charsets. Last year Red Hat supported ten languages. We recently localised
six Indian languages. Being able to support Indic is very important if Linux
is to gain a wider audience.
Unified front-end with Bluecurve, the Firefox browser and Thunderbird e-mail
client show that there’s always an opportunity for Linux in tools. Internet
Explorer is not the be all and end all.
*Can Linux be monetised?
There are various ways to do this. At Cygnus we had high quality contracts and
undertook specific development projects. We did high end targeted consultancy.
Red Hat has followed two different business models. When we went public we had
a retail product with a six-month refresh cycle. ISVs chose not to port their
software to Linux as the time was too short for them to qualify their product.
With Red Hat Enterprise Linux we have a 12-18 month cycle that is long enough
for hundreds of ISVs [to port their products]. We support technology for seven
years. This is very important for enterprise customers. We have clients with
tens of thousands of servers.
*The Indian contribution to Open Source has been insignificant.
Why?
OSS is a new idea in India. I started a company when nobody was doing commercial
OSS in 1989. I hear the same kind of questions in India that I heard in the
US in the 1990s. It’s a matter of time.
*What’s the case for TCO on Linux?
There are a lot of TCO studies. A TCO study is just a point in time scale of
reference. Companies are looking not only at the cost today but what it will
be tomorrow. Last year IDBI Bank reported 35 percent savings. A US customer
used to spend $12 million per year on hardware maintenance of its SMP 24/48-way
systems. It changed to a horizontal model and terminated the hardware maintenance
contract taking a decision to throw defective machines away. Fifteen months
later it was spending $400,000. A traditional TCO study would miss that.
Prashant L Rao
|
