Linux May Be More Ready Than You Think

November 12, 2001

원본 : http://www.internetweek.com/graymatter/letters111201.htm

I largely disagree with Stacey Quandt on the feasibility of implementing Linux across the enterprise ("Not Yet Ready For Prime Time," Oct. 29).

For one thing, I take issue with her view that enterprisewide Linux implementation requires more Linux-certified IT professionals. Insistence on certification elevates form over substance.

As to Quandt's point that Linux is significantly more limited than Unix, I disagree with her examples with respect to serviceability, scalability and manageability. Linux has 32-bit UID support, as well as a range of global and journaling file systems. A basic Linux SQL server like PostgreSQL will generally hammer the likes of MS SQL Server and Oracle into the ground in performance terms.

Finally, Quandt's statement that industry benchmarks are needed to establish hardware and software providers as short-list contenders, rather than winners or losers, proves she doesn't get it. "Software providers" isn't the clear-cut term it used to be. Who are Linux's software providers? Several thousand otherwise unrelated people around the globe, most of whom use English as a second language, if at all.

Quandt's approach makes her "no" as useful as an MSCE certification.

Leon Brooks

Programmer/analyst

CyberKnights

Perth, Western Australia

CERTIFIED NOVICES

In her Oct. 29 column, Stacey Quandt puts too much emphasis on Linux certification. What does it prove? I know several certified people who don't know anything but what's in the books.

We converted our servers from NT/2000 and Novell to Linux and OS/2, running a 200-GB-plus database for more than two years with no downtime. We also converted the mainframe VSE to Linux. The desktop arena is the last frontier, and we're almost ready to convert that as well.

Ray Pittigher

Systems specialist

The Credit Index

Mount Arlington, N.J.

Flawed Execution

Regarding Tim Wilson's column "A Federal Case For Web Computing" (Sept. 10), problems with the distributed (client-server) model have been rooted not in technology per se, but in faulty execution. Technology can only compensate so much for boneheadedness and lack of planning.

Troubles arise when we fail to recognize the costs and risks inherent in every so-called solution.

Tom Dean

Assistant chair, computer networking

Texas State Technical College

Waco, Texas