Re: Any Experiences of 11r2 Win2k8R2
Date: Tue, 9 Nov 2010 19:56:27 -0800 (PST)
Message-ID: <05489b98-a8c2-4447-8dd8-69ed6a854493_at_k30g2000vbn.googlegroups.com>
On Nov 10, 8:29 am, Mladen Gogala <n..._at_email.here.invalid> wrote:
> childish act. BTW, it will be interesting to see the "real systems" with
> the write barriers in 2.6.32+:
>
> http://lxr.free-electrons.com/source/Documentation/filesystems/ext4.txthttp://www.spinics.net/lists/xfs/msg00081.html
Yegawds! More madness...
> If nothing else, NTFS is a mature file system which can beat Ext3, Ext4
> ReiserFS and similar toys hands down when it comes to I/O performance.
Why on earth hasn't the Linux development community learned from fullstrength file systems such as xfs - and others - instead of perpetually trying to re-invent the wheel? "Not invented here" is rampant, in those circles!...
> People who have created NTFS must be laughing to tears while looking at
> all of that. The mess is, of course, partially created by Linus Torvalds
> himself and his diatribe against O_DIRECT:
>
> http://lkml.org/lkml/2007/1/10/233
Lovely. He's an expert in data management and file system architecture now, as well? The man must be a God, such is his infallibility and endless knowledge about everything...
Of course dbs don't need to use anything else than the page cache! Next mad idea?...
> As a matter of fact, Linux is in a sorry state these days. Buggy, having
> performance problems and having finally lost the battle for the hearts
> and minds:http://tinyurl.com/26tawax
Indeed. True for desktops. For servers, I still have my doubts and hold hope high.
But fine tuning Linux/Oracle servers still involves a lot of unnecessary "black arts", inadmissible this day and age. Look at the whole rigmarole with hugepages and the heaps of conflicting info out there on why/how/when to use it, most of it against.
This, when IBM, Dijkstra and many others proved beyond any doubt why it'd be needed once we got into 64-bit hardware and very large memory, 35 years ago! It's got nothing to do with OSs and all to do with TLB size and virtual memory translation.
Talk about no clue... Indeed it is true that those who refuse to learn from history are forever comdemned to endlessly repeat the same mistakes.
Sure: it's easy to hire an "expert" to come in and fix things up. Problem is: why should one be needed for something as basic?
Ah well: let's hope they don't learn the lesson too late. Don't hold your breath... Received on Tue Nov 09 2010 - 21:56:27 CST