Hewlett-Packard has formally declared Linux as its third strategic
operating system. Ian Murphy reports
Making sense of the various Unix strategies of the major hardware
and software suppliers has provided long-term employment for an
army of analysts, journalists and hangers-on. But now it appears
that all that competitive spirit is falling by the wayside in an
era of technological Glasnost, as supplier after supplier pledges
its future to that Unix-based upstart, Linux.
The latest company to fall under the spell is Hewlett-Packard.
Although it has been supporting Linux for some time now, HP is
finally putting its money where its mouth is by formally declaring
Linux as its third strategic operating system behind HP-UX and
Windows.
HP has yet to announce which version of Linux it will choose for
its distribution, but it won't be Red Hat as the company has just
signed a distribution deal with IBM.
There is a lot of confusion and concern as to what a Linux strategy
will mean for HP customers and whether Linux can provide an
industrial-strength computing platform.
Much of that concern revolves around the newness of Linux compared
to other versions of Unix and whether it will run on different
hardware platforms. There are also significant concerns over the
availability of applications and their stability.
In an attempt to provide a solid view of what is happening and
prove that HP knows what it is about, it recently sent Bruce
Perens, its strategic advisor for Linux, on a press tour around the
world.
Perens is frank about the issues that Linux faces and understands
the concerns of many customers. The first thing that he clarified
was the role of Linux over the next few years and its relationship
with HP-UX.
Many HP customers have made significant hardware investments in
order to run their systems on HP-UX and there are no plans to phase
out the system. Indeed, to do so would be an extensive project,
requiring testing of the operating system, the production of a
large number of hardware drivers and a co-ordinated effort with all
hardware and software suppliers.
Unlike the Jini initiative, where hardware suppliers supplied
copies of their drivers for others to use, Perens does not believe
that Linux will work the same way. While the operating system will
continue to contain drivers for common commercial hardware such as
printers and monitors, specific drivers for Raid and distributed
hardware will need to be tested against each Linux distribution.
There is also a cost factor that will need to be considered.
However, this has not prevented HP from developing drivers for all
of its consumer technology, and it has now started to look
seriously at some of its enterprise equipment. Included in this
category are HP's San and NAS back-up solutions.
Software suppliers will also need to be sure that Linux can cope
with the requirements of their products. But, with IBM, Dell,
Compaq and HP pushing Linux into the mid-range server market,
suppliers such as Oracle are already porting enterprise software to
Linux.
One issue for commercial software suppliers is the licensing model
that currently underpins Linux. Much of the existing development
starts out as public domain. If you use any of the software
provided under the general public licence you are required to
provide your source code in return. With a software industry that
has spent years saying that software should not be free, we are
likely to see a real battle.
How far up the computing model Linux can go will be determined by
the success of the Intel 64-bit processor. Two years ago, there
were several suppliers in the running to create a 64-bit version of
Unix but, as the launch draws closer and the suppliers have to
produce code, Linux looks like it will be the preferred version for
most of the major computer suppliers.
HP is responsible for the Linux IA-64 kernel, and the success of
this project is likely to have a big impact on the future of HP-UX.
Software suppliers will also have a lot to say, and with
Microsoft's IA-64 bit support falling further back, Linux could
steal the enterprise market away from Microsoft. For HP, supporting
both the Microsoft and Linux camps should mean that it can
consolidate its grip on the enterprise market.
However, HP has yet to follow the lead of Sun and provide the
ability to run Linux applications in native mode on HP-UX. Such a
move would allow corporate IT teams to develop and port software on
their enterprise platforms and then test it against HP-UX for
performance and portability. A substantial benefit of this would be
to accelerate the development of Linux into a distributed computing
platform.
With the current range of low-power, ultra-thin servers about to be
unveiled by Compaq, such a distributed architecture is now within
reach. The only thing that has yet to be determined is the
operating system that will sit on the machines.
As we move down from servers to workstations, there is a
substantial benefit to using embedded Linux in a low-end, thin
client rather than a Windows-based operating system. One of the
advantages is that the client can have limited local storage,
allowing for cacheing of Web data rather than storing it on the
server.
Another advantage is the ability to remotely manage devices and
allow for the connection of a wide range of local peripherals. As
well as the corporate thin-client market, there is substantial
interest in thin-client deployment from hotels, whose main income
is from business travellers. Remote management, the ability to
allow full Internet browsing and VPN (virtual private network)
capabilities are issues that hotels have been struggling with for
some time. Linux is ideally based to provide this.
With HP due to announce its strategy for the low-end market in the
next week, we can expect to see support for both embedded Linux and
Linux on the local storage device.
Further down the computing model, the best-selling HP Jornada PDAs
are due for an overhaul shortly and there are rumours about HP's
commitment to only supplying Windows CE. It is known that HP is
currently looking at a new range of processors for these devices
and, while HP is keeping details quiet, insiders believe that the
next range of processors will be capable of running embedded
Linux.
There are also other reasons for HP wanting to look closely at the
embedded Linux market. Unlike many of its competitors, HP has an
active telecoms equipment business and embedded versions of Unix
have long been used in a wide range of telecoms equipment. With an
embedded Linux strategy, HP would be able to look at the huge
demand for replacement equipment in the telecoms backbone.
This is likely to be extremely lucrative, as the telecoms backbone
tries to move from an IPv4-based network to an IPv6-based network.
Most Linux distributions already have gateways that will map
between these two stacks and HP would be able to take advantage of
its experience in packet-switched data management to simplify that
transition.
HP, like all the other large computer equipment suppliers, is
trying to cover its bets, and the initial impression of its Linux
strategy appears to be one of careful thought. Rather than simply
try to provide Linux support for a small segment of its customers,
HP appears to be trying to position Linux as a complete solution.
This is the most positive position to date of any of the big
players.