June 29, 2005

Fedora Directory Server offers opportunity for Red Hat, community

Author: Brian Jones

The announcement
at Red Hat Summit that the Netscape Directory Service
product purchased by Red Hat last year was destined for GPL glory got a
far milder reaction than the significance of this event merits in terms
of what it means for Red Hat, Linux users, and open source software.

Most people don't deal with a directory service every day, but
probably interact with one all the time. Directory services authenticate
users, complete email addresses when you type in your email client's
"To:" field, and perform many more system-level tasks.

Netscape Directory Server (NDS), which is now Fedora Directory
Service (FDS), is one of the most popular directory service
applications. At one time it was also iPlanet Directory Server, which
evolved into SunONE Directory Server. The code bases for SunONE,
iPlanet, NDS, and FDS are quite similar, and represent a large
percentage of worldwide directory service deployments. That makes FDS a
large boot in the door at many organizations for Red Hat.

Think of all the shops where people are still running Solaris 7 and
iPlanet Directory Server, or even Solaris 8 with SunONE Directory
Server. Why should they buy a proprietary, closed box that contains a
proprietary, closed service, running on a proprietary, closed platform,
and pay for that "privilege", when instead they can get an open source
(and, in the case of FDS, free), full-featured solution to run on a
commodity hardware platform? FDS (and in the commercial world, RHDS)
becomes a hammer swung by Red Hat to drive the final nail into the Sun
coffin.

I don't see how Sun can compete with enterprise-tested directory
services from the likes of Red Hat and Novell. Novell is doing the exact
same thing as Red Hat: clients running its old NetWare OS are natural
customers for all of its products that now also run on Linux.

Novell and Red Hat are now the main competitors in the directory
service marketplace. The question isn't whether IBM SecureWay, SunONE,
or CA's eTrust will be displaced, but rather when and by which
competitor. Linux and open source software will win if FDS is
successful, because FDS will be 100% open source (according to Red Hat,
currently available packages notwithstanding). Novell's product is not
open, and I know of no known plans to make it so.

Novell's and Red Hat's edge may not exist for long. Three years from
now, look for Apple to add a GUI to OpenLDAP, on an Intel platform, and
try to take over. Nor are IBM, Sun, or CA likely to sit back and watch
their products tank.

How Red Hat can fail

No, Red Hat's rosy future is not written in stone. The company must
be consistent and realistic in its pricing and sales distribution
methods. As a Red Hat Network user myself, I'm half-expecting Red Hat
Directory Server (the commercial version of the free Fedora Directory
Server) to show up as a completely separate channel that will
require a flat fee per machine to register for. That would be foolish,
since I can download and use the Novell, SunONE, or
even OpenLDAP software for free. Red Hat should consider making RHDS
free to RHN subscribers up to 100,000 entries or so. This isn't unusual;
eDirectory used to have a free license for up to 250,000 entries, and
SunONE did something similar. Don't nickel and dime me. Yes, I know I
can always download FDS for free, but if I'm running RHAS, I should be able to
run RHDS, which integrates with RHN and probably a few acronyms I'm
forgetting.

A second way Red Hat can fail is by stalling development. RHDS is a
great application, but it's certainly not finished, and I have yet to
hear how the company will build a development community around the
product. The Fedora Directory Server has what would appear to be a development community site
all set up, complete with a changelog page and a bug submission page,
but the account request form says they "are not ready to accept
contributions at this time." I'm sure this is temporary, but why would
you not take advantage of the buzz surrounding the announcement to gain
development momentum?

What about integration? This is where Novell has the benefit of time
and, as a result, a mature and smartly integrated set of tools for
everyone from end users who can barely use email to administrators who
deploy it. Red Hat has a long way to go in this area, and it can't make
the mistake of fighting with Novell over this stuff. It needs to set its
sights on Sun and Microsoft (you didn't think I forgot about Active
Directory, did you?).

Finally, Red Hat could fail to open source those parts of the project
which currently are not GPLed. There are reportedly issues being worked out in
pursuit of open sourcing the most attractive feature of the whole
project: the graphical administration console. For us open source
developers and administrators, it wouldn't be the end of the
world, since the actual running daemon is a GPLed app, and we could
write our own interfaces to that if we absolutely had to. We'd still be
ahead of the game in comparison to OpenLDAP on the basis of cold, hard,
system-level features. But it would absolutely kill any faith we
had in Red Hat being devoted to open source software.

In closing

In a nutshell, Red Hat Directory Server could mean big things for
Red Hat if it can execute a plan to take advantage of the product's
long history and current deployments. Fedora Directory Server, its free
cousin, is a major win for the open source community, many of whom have
come to look at OpenLDAP as something you have to deal with if you want
a free directory server, instead of something that is a truly
competitive product. However, administrators sit cautiously watching and
waiting for Red Hat to follow through with the GPLing of the entire
product, and to see how the commercial product, RHDS, will be
integrated, priced, and delivered. If it comes together, it is an almost
certain win for Red Hat, the open source community, and shops looking
for a commercially-supported, feature-packed solution to complete their
data center Linux migrations.

Category:

  • Enterprise Applications
Click Here!