Paul:
This is not my field (software), but isn't this article (opinion) by John Taschek, sort of like preaching to the choir (Eric Schmidt), and probably 1 year plus late to being relevant here? Would it even make sense for Schmidt to make some sort of deal with the group that left Novell to develop "Wolf Mtn.-like clustering software, come back under some kind of arrangement? Or are all the bridges burned by the past NOVL CEO's? Just curious, as I wouldn't think that Schmidt would let go of a good thing that easily, especially something that would support the progress of NT, and especially highly scalable, highly available Java applications. Or is Novell doing the same thing under their Provo lab as we speak?
Regards,
QuadK
Wide Angle
How Novell messed up a good thing
By John Taschek
06.22.98
Whether Novell's NetWare
5.0 is a great product or
simply an upgrade remains to
be seen. One thing is sure,
though: While mapping its
course for the future, Novell
has obliterated one set of
directions that would have
given it a clear lead in two
crucial areas -- a highly
scalable file system andperformance-based clustering.
Many of the facts have been spun around so many times by jaded
ex-employees and dispensers of corporate legal mumbo-jumbo that it's impossible to know
what really happened. But as far as I can tell, Novell lost a true edge when it closed its Wolf
Mountain clustering project and then sued the engineers who left the company and further
developed the technology.
If you ask Novell officials, they'll say they have clustering and have had it for quite a while.
But ask them where this technology came from, and they acknowledge it came from Vinca.
Vinca's StandbyServer is perhaps the best in the industry, but it delivers only a solution for
high availability. Akin to Microsoft's Wolfpack, in which a backup server takes over if the
primary fails, StandbyServer is not made for clustering (although it does offer some
performance benefits through disk mirroring). (See StandbyServer review.)
But while companies need high-availability clustering, the company that first delivers high
performance along with high availability will be the true innovator. What Novell needs now
is to have a blitzing-fast 64-bit version of its operating system. However, it may need to
completely rewrite NetWare to get the threaded JVM it needs and to solve Java's problems
with trash collection.
Novell then needs to educate developers on how to develop Java network applications.
Novell's history here is pathetic. Round after round of application development woes
practically killed the NOS as an application server.
Novell's lawsuits have shattered Timpanogas, which has only three employees now. They
are developing technology components that will form what Merkey calls VNDI -- Virtual
Network Disk Interface, a storage-area network solution that provides infinite storage
capacity using network technology.
Merkey is also working on aspects of technology to create his own operating system. He's
extremely careful (read: scared to death) about following the court-imposed guidelines and
is working only in nonprohibited areas. Whatever he is able to complete, though, won't be
for Novell -- his technology will be his own or for Novell's competition, Windows NT.
Wouldn't a highly available, highly scalable network OS that ran Java applications really
fast have been intensely interesting? John Taschek can be reached at
john_taschek@zd.com. |
