Linux for the Enterprise @ CMP 35
Andrew Dvorak wrote to us with Network Computing's take on Linux. Good chronicle of why Linux is moving into the higher end markets, and the huge surge of support that has been seen in the last six months.
Filesystem (Score:1)
One question - He mentions the lack of a 64bit journaling filesystem a couple of times. Even in a quote from "Mad Dog". When will XFS start being mentioned in these articles? I know it's not available yet, but couldn't they mention it?
He just says one is being developed. To some people, that's synonymous with non-existant.
Re:Why bother with Linux, NT or Solaris? (Score:1)
But let's have a look at a few of the assumptions. They compare Netscape's server on Solaris with Apache on OS X. Has the world changed in some sudden way that it's not possible to compile Apache on Solaris any more? Or is it just obvious to everyone but me that, despite using two completely different web servers, the speed problems are all the OS's fault? And have we decided to drop all server applications but web serving? And even drop all web-serving applications that might require, say, eight CPUs and a couple of gig of RAM? And we're no longer worried about things like hardware failover?
cjs
Re:"on the shelf" includes "downloadable" (Score:1)
Vapourware is more an intent to decieve... "Well we don't even have any ideas, but we'll announce it to stifle anyone else trying to do it."
You're arguing that simply because it doesn't have the same name that it is vapour... vapour might include Microsoft Word for Linux. There's nothing even resembling it.. they might announce it to make others rush or drop their plans, but they have no intention of doing it...
A promised peice of software or feature is not vapourware if it ever emerges.
My favorite part... (Score:1)
What a great analogy. My company is slowly seeing this is true (with a little help from a certain employee
----------------
"Great spirits have always encountered violent opposition from mediocre minds." - Albert Einstein
Re:Filesystem (Score:1)
"on the shelf" includes "downloadable" (Score:1)
Linux 2.4 (3.0?) is vaporware. 2.3 is real since it is downloadable.
Same with NT. W2k is vaporware. W2k-BetaX is real since it is downloadable.
The thing to keep in mind is that any individual feature-disparity can be vaporware, too. For instance, a journaling filesystem for Linux is vaporware for any kernel version number. Same with a lot of the promised features of W2k.
--
"Please remember that how you say something is often more important than what you say." - Rob Malda
Re:Why bother with Linux, NT or Solaris? (Score:1)
Why bother with OS X Server? Recent tests have shown NT is slightly faster than Linux. Though these tests have met with some disbelief, the fact remains that NT and Linux are fairly equally matched. The report at NewMedia [newmedia.com] shows that OS X is 7 times faster than NT (& 3 times faster than Solaris too), and therefore its quite reasonable to assume that it's faster than Linux. It also puts the NT/Linux debate into some context.
They ARE synonymous (Score:1)
Unless the product is actually on the shelf, it is vaporware.
--
"Please remember that how you say something is often more important than what you say." - Rob Malda
As long as we're obsessing about Apple... (Score:1)
Of course, Apple's involvement with Linux (MkLinux) goes back years, not months. The author seems to be confusing Darwin with Linux...
Re:Why bother with Linux, NT or Solaris? (Score:1)
http://www.macintouch.com/mxshighavail.html
Amazingly enough, a lot of the problems with MacOS X server are also problems with Linux. But, in my humble opinion, Mac OS X Server is *NOT* a proven solution. Yes, BSD is. Yes, OpenStep mostly was. However, Mac OS X Server has only been out for a few months. One of the major things that I noticed while attempting to use my own MacOS X Server, was the lack of ability to add disks. Maybe i'm on crack. Maybe I could not read. But I could not get it to recognize other partitions on the same disk.
MacOS X Server, as far as scaling is concerned, Mac OS X Server doesn't do it. I don't care how fast a g3 is. It is NOT going to compare to a quad Pentium or quad Sparc machine. Period. EOF. Unfortunatly Apple Hardware can't seem to figure out how to put 2 power supplies in their cases either. I don't know about you, but I wouldn't bet the ranch on anything with a single ps.
Wheee.
Me.
Re:Why bother with Linux, NT or Solaris? (Score:1)
why so defensive?
Re:Why bother with Linux, NT or Solaris? (Score:1)
Philosophical conundrum (Score:1)
"A promised peice of software or feature is not vapourware if it ever emerges."
A promised piece of software is vaporware UNTIL it emerges since there's no way to predict if it ever will emerge until it does.
--
"Please remember that how you say something is often more important than what you say." - Rob Malda
Re:Why bother with Linux, NT or Solaris? (Score:1)
I'm interested in OSX/Apache server plus Samba if the ever port arrives, as a WEB/ NOS solution. It's probably the case that OSX Server spanks NT soundly on standalone webserving. And y'know it could just be that the savings of a free OS (Linux) do not actually outweigh the disadvantage of Intel hw. Could be. (I know about PPC Linux but let's face facts: without Apple's willingness to open up the hardware spec, PPC Linux is likely to always be ~generation behind. Oh yeah, that's FUD too...still it's probably true, isn't it? G4 boards will show up about the time we have PPC linux booting smoothly on the bw G3's) Would like to see more testing on OSX and NT and Linux. I'm kind of surprised at the lack of interest 'round here. And I expect that the original poster is maybe a little more surprised than I --like frustrated, maybe-- with the head in sand attitude some have. No doubt when he includes Linux in his comment he is thinking of the earlier benchmarks from Apple published a week or two after the commercial release of OSX Server which showed the G3-400 server easily outperforming Linux as well as NT on a Dell Poweredge 2300 dual PentiumII. Now aside from a little exaggeration in his attitude, I don't see any serious attempt to desecrate and disparage your favorite OS. The report is based on tests done internally at Intuit corp, and you should know they're no special friends of Apple. In fact, they would have gladly sold their shiny white butts to Microsoft in '95 if the Federal Gov. would have allowed it, and had to be shamed into porting the latest version of Quicken for the Mac. I'd like to see a lot more about this test, the install logs, the procedures, blah blah. But, as is, I do find it interesting, especially as it seems to come from the field, not a ZD symlink to the Redmond mothership, or similar...Anyway. Ease up, you'll get an ulcer or something.
That's not what the report said... (Score:2)
I would suggest that this test shows nothing more than the relative speeds of Apache, IIS and Netscape Enterprise. If one wanted relevant speed comparisons for the OS's, one would most certainly hold the server as a control variable - that is to say that each of the machines would be running Apache.
Further, the Solaris test is an example of very poor science - a dual 200MHz SPARC is not necessarily comparable to a PII 400 MHz Intel or G3. You can't just multiply processor speed by number of processors.
Even worse, the "report" makes no mention whatsoever of configuration of the servers, the setup of the client machines, or any other variables that one would expect any HONEST test would control.
The only thing made clear in the test "report" is that this test says NOTHING about Linux's performance.
Re:Why bother with Linux, NT or Solaris? (Score:1)
Here's a direct quote from the lab report - 'for 100 users, the Mac served over three times more pages than the Sun, and over seven times the NT'
Wonderful (Score:1)
Very well written, researched and presented.
Makes me shiver when I think of what's going to be available for Linux in 4 years.
Let's hear it for momentum. Keep it up, all.
Glad to see the coverage. (Score:1)
Enterprise backup... (Score:1)
Otherwise, a well-written article. Too bad we couldn't see the numbers on the SMB tests - I suppose you have to have permission from Microsoft in order to do that :-(.
Re:They ARE synonymous (Score:1)
In Development may mean vapourware coming from some people... The 2.3 Kernel is "in development" but it's far from vapour. Don't believe me? Go down it...
Re:Why bother with Linux, NT or Solaris? (Score:1)
Why add Linux in with NT and Solaris when the lab report doesn't mention Linux?
Why compare apples to oranges IIS and Netscape to Apache?
Why consider closed source models in relation to open source?
Why consider focusing on just one operating system?
Why do you figure a web test makes OS X better at everything, when all it is measuring is one aspect of a system?
Why accept the report at face value when you don't know what tweaks have been implemented on the system?
Why Post FUD?
Lando
Re:Filesystem (Score:1)
Now there's a true but very interesting statement. If something is in development then that is perceived that it is synonymous with, say, vapourware. Yet, when Microsoft announces something is "in development," magazines, articles, and editors all praise the wonderful product that Microsoft is going to change the world with. And how it is going to be so much better then what their competitor's have out now!
If the same attitude was used when reporting on Linux, I would expect them to be glowing with praise about the up and coming journaling filesystem, and how it was going to blow all other journaling filsystems out of the water.
It's funny, I'm not seeing any press....
However, Linux doesn't need cheesy articles to compete like Microsoft does. All Linux needs is real-life success stories!!!!! Let Microsoft get bloated editorials praising them. I'll go with actual success stories anytime.
Re:Filesystem (Score:1)
If not, then it's fine for him not to mention it other than the vague reference he makes, as it is not on the horizon yet for Enterprise environments (which is, after all the focus of the article).
IMHO.
--
Re:Why bother with Linux, NT or Solaris? (Score:1)
You might be interested to know that there is a version of Samba 2 for OS X Server here [peak.org]
Re:Why bother with Linux, NT or Solaris? (Score:1)
Clustering (Score:2)
The author seems to equate Beowulfs with high availability clusters. Beowulf clusters are for high computational processing, not for high availability processes.