Re: Disable read/write caching to disk?

From: Scott Long <scottl_at_samsco.org>
Date: Thu, 26 May 2005 22:03:39 -0600
Andre Guibert de Bruet wrote:

> 
> On Thu, 26 May 2005, Scott Long wrote:
> 
>> Chad Leigh -- Shire.Net LLC wrote:
>>
>>> On May 26, 2005, at 8:49 PM, Eric Anderson wrote:
>>>
>>>> So it sounds dangerous, but not disastrous..  Sounds like soft- 
>>>> updates would help this alot, so I'll turn them back on for this  
>>>> filesystem (I typically do use it).
>>>>
>>>> At a minimum, it would be awesome to even have a way to do one host  
>>>> rw and several doing ro.  Think of the case of a web server farm,  
>>>> where it's nearly all reads.
>>>
>>>
>>> use NFS or something. Not ideal but it allows you to have lots of  
>>> clients using the same space without the disasters.
>>
>>
>> NFS and Coda/AFS require that you have an intelligent node, i.e. a
>> computer, in front of each disk.  The whole idea of a Fibre Channel
>> or iSCSI SAN is that you have a network of disks connected to a network
>> of computers, all able to communicate with each other and not have to
>> be fronted by a computer.  This is quite important for high-availability
>> storage networks that want the reliability and scalability of not having
>> a single computer be the choke point or single-point-of-failure for a
>> particular set of data.  Granted it's still somewhat of a niche, but as
>> persistent storage and data mining become more part of the mainstream,
>> it'll start becoming very important.  Right now FreeBSD simply isn't an
>> option, while Solaris, NT, and Linux are.
> 
> 
> Fair enough. So the question becomes: What are the Linux folks doing 
> that we aren't and how?
> 
> Andy
> 

RedHat offers a SAN filesystem called GFS, which they aquired when they
bought Sistina several years ago.  GFS was originally written for IRIX,
but then ported to Linux and mostly kept there.  It would be wonderful 
to get access to the old IRIX code and port it over, since their VFS is
much more similar to ours than Linux.  That's probably just wishful
thinking though.  The Linux source is available under the GPL, so it's
a matter of either porting it over (and living with the GPL), or 
re-implementing the protocol (or something similar) from scratch.  A
few people have suggested modifying UFS to fill this role.  It probably
is just as much work as porting GFS, if not more, since UFS/FFS is 
closely tied to the buffer cache and block layers on BSD, and divorcing
probably would be quite difficult.

Scott

Scott
Received on Fri May 27 2005 - 02:04:59 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:38:35 UTC