>Did anybody see last week's PC Week? That issue had some redux on some
>recent benchmarking that a third party company did comparing NT and Linux
>which proved rather interesting (even after retesting when the mag thought
>to cry foul)
>
>There sure is a lot of talk of Linux being faster, lighter etc. but the
>benchmarks showed NT running more than twice as fast as Linux across the
>board on all but the lowest end hardware. This includes just plain file
>serving which is definitely not NT's strong point.
>
>I kinda wonder about the logic that is pushing Linux into so many shops
>these days based on faulty assumptions...
Rick,
I don't think Linux is being pushed into shops erroneously. It is being considered, used by some, and dismissed by others...just like any tool. Yes, there is hype, but we can see through that, can't we? The fact is that many shops like the price and stability of Linux, and they are continuing to use it. The proof's in the pudding.
About the benchmark...I don't consider a single-processor machine (where Linux was still slower, but at least not incredibly so) to be the "lowest-end" hardware. Plenty of shops still use that type of hardware, and will continue to do so. (I don't know a lot of web servers that have a 400 MBit line to the web anyway, so is Linux really such a dog in a more realistically configured shop?)
The benchmark did make it abundantly clear that Linux needs _LOTS_ of work on multi-processor utilization...this has been known for some time, and work is being done on the 2.3 kernel as we speak. The benchmark did a very good job of helping isolate several bottlenecks in the Linux/Apache solution, some at as low as the kernel level.
Here's another article I found as a sort of response to the benchmark...not too bad, and not too much in the "Linux-zealot" tone I have begun to get annoyed with on some of the Linux newsgroups...
http://www.kegel.com/mindcraft_redux.htmlI agree that Linux needs lots of work, but I don't think it should be dismissed just because the hype has gotten out of hand with what the product can do.
Thanks,
Joe Kaufman
