Re: [SLUG] Asynchronous Distributed Filesystem

2008-01-16 Thread Alex Samad
On Wed, Jan 16, 2008 at 05:46:35PM +1100, Joel Heenan wrote:
 Hi,
 
[snip]
 
 Sorry I didn't mean to say differences I meant to say distances. The
 DR site is a good 20km away. I have not researched this thoroughly but
 it was my understanding that GFS was designed for fibre connected
 volumes not over large distances with higher latency.

you could possible do it over iSCSI, I believe that is an acceptable protocol 
for gfs



 
  
   unionfs is I think too experimental.
  
   Continous Access, using our SAN to replicate the data, at this point
   has to be discounted because of licensing.
  
   How do other people generally solve this problem?
  
   Thanks
  
   Joel
   --
[snip]
 

-- 
That I'm running on my dad's name...  I'm proud of my dad...  I reconciled my 
love for my dad a long time ago.

- George W. Bush
When asked by Larry King 'What do people misunderstand about you most?'


signature.asc
Description: Digital signature
-- 
SLUG - Sydney Linux User's Group Mailing List - http://slug.org.au/
Subscription info and FAQs: http://slug.org.au/faq/mailinglists.html

[SLUG] Asynchronous Distributed Filesystem

2008-01-15 Thread Joel Heenan
SLUG,

We have a requirement in a new project to have a distributed
filesystem. Files are written to one of 32 * 200MB volumes and we need
to keep them in sync with a DR site. Rsync, I believe, will be just
too slow to replicate changes - unless there is some way to make the
rsync daemon hook into the kernel and know what changes have been
made?

RHEL GFS I think will not work across such differences and won't do it
asynchronously.

unionfs is I think too experimental.

Continous Access, using our SAN to replicate the data, at this point
has to be discounted because of licensing.

How do other people generally solve this problem?

Thanks

Joel
-- 
SLUG - Sydney Linux User's Group Mailing List - http://slug.org.au/
Subscription info and FAQs: http://slug.org.au/faq/mailinglists.html


Re: [SLUG] Asynchronous Distributed Filesystem

2008-01-15 Thread Adrian Chadd
Under Linux? DRDB? http://www.linux-ha.org/DRDB/


On Wed, Jan 16, 2008, Joel Heenan wrote:
 SLUG,
 
 We have a requirement in a new project to have a distributed
 filesystem. Files are written to one of 32 * 200MB volumes and we need
 to keep them in sync with a DR site. Rsync, I believe, will be just
 too slow to replicate changes - unless there is some way to make the
 rsync daemon hook into the kernel and know what changes have been
 made?
 
 RHEL GFS I think will not work across such differences and won't do it
 asynchronously.
 
 unionfs is I think too experimental.
 
 Continous Access, using our SAN to replicate the data, at this point
 has to be discounted because of licensing.
 
 How do other people generally solve this problem?
 
 Thanks
 
 Joel
 -- 
 SLUG - Sydney Linux User's Group Mailing List - http://slug.org.au/
 Subscription info and FAQs: http://slug.org.au/faq/mailinglists.html

-- 
- Xenion - http://www.xenion.com.au/ - VPS Hosting - Commercial Squid Support -
- $25/pm entry-level VPSes w/ capped bandwidth charges available in WA -
-- 
SLUG - Sydney Linux User's Group Mailing List - http://slug.org.au/
Subscription info and FAQs: http://slug.org.au/faq/mailinglists.html


Re: [SLUG] Asynchronous Distributed Filesystem

2008-01-15 Thread Alex Samad
On Wed, Jan 16, 2008 at 05:15:51PM +1100, Joel Heenan wrote:
 SLUG,
 
 We have a requirement in a new project to have a distributed
 filesystem. Files are written to one of 32 * 200MB volumes and we need
 to keep them in sync with a DR site. Rsync, I believe, will be just
 too slow to replicate changes - unless there is some way to make the
 rsync daemon hook into the kernel and know what changes have been
 made?

lustre comes to mind ?  you haven't really expanded on how the striping is 
supposed to be done ?

 
 RHEL GFS I think will not work across such differences and won't do it
 asynchronously.
can you expand on across such differences.

 
 unionfs is I think too experimental.
 
 Continous Access, using our SAN to replicate the data, at this point
 has to be discounted because of licensing.
 
 How do other people generally solve this problem?
 
 Thanks
 
 Joel
 -- 
 SLUG - Sydney Linux User's Group Mailing List - http://slug.org.au/
 Subscription info and FAQs: http://slug.org.au/faq/mailinglists.html
 

-- 
We will stand up for terror --we will stand up for freedom.

- George W. Bush
10/18/2004
Marlton, NJ
in a campaign speech


signature.asc
Description: Digital signature
-- 
SLUG - Sydney Linux User's Group Mailing List - http://slug.org.au/
Subscription info and FAQs: http://slug.org.au/faq/mailinglists.html

Re: [SLUG] Asynchronous Distributed Filesystem

2008-01-15 Thread Joel Heenan
Hi,

In response to Adrian I'm looking for a solution that will work well under RHEL.

Thanks for the suggestions thus far I'll check them out now. Comments below

On 1/16/08, Alex Samad [EMAIL PROTECTED] wrote:
 On Wed, Jan 16, 2008 at 05:15:51PM +1100, Joel Heenan wrote:
  SLUG,
 
  We have a requirement in a new project to have a distributed
  filesystem. Files are written to one of 32 * 200MB volumes and we need
  to keep them in sync with a DR site. Rsync, I believe, will be just
  too slow to replicate changes - unless there is some way to make the
  rsync daemon hook into the kernel and know what changes have been
  made?

 lustre comes to mind ?  you haven't really expanded on how the striping is
 supposed to be done ?

Umm not striping mirroring. Looking at having all the data replicated
out to a DR site so there would be to separate instances.


 
  RHEL GFS I think will not work across such differences and won't do it
  asynchronously.
 can you expand on across such differences.


Sorry I didn't mean to say differences I meant to say distances. The
DR site is a good 20km away. I have not researched this thoroughly but
it was my understanding that GFS was designed for fibre connected
volumes not over large distances with higher latency.

 
  unionfs is I think too experimental.
 
  Continous Access, using our SAN to replicate the data, at this point
  has to be discounted because of licensing.
 
  How do other people generally solve this problem?
 
  Thanks
 
  Joel
  --
  SLUG - Sydney Linux User's Group Mailing List - http://slug.org.au/
  Subscription info and FAQs: http://slug.org.au/faq/mailinglists.html
 

 --
 We will stand up for terror --we will stand up for freedom.

 - George W. Bush
 10/18/2004
 Marlton, NJ
 in a campaign speech

 -BEGIN PGP SIGNATURE-
 Version: GnuPG v1.4.6 (GNU/Linux)

 iD8DBQFHjabdkZz88chpJ2MRAtajAKCyBHYpuJVrDn0/kZxA6Ip2jD7ZOwCgmBL6
 ztaW/YLUVg5UdXS29A7eArg=
 =SVkT
 -END PGP SIGNATURE-

 --
 SLUG - Sydney Linux User's Group Mailing List - http://slug.org.au/
 Subscription info and FAQs: http://slug.org.au/faq/mailinglists.html


Thanks

Joel
-- 
SLUG - Sydney Linux User's Group Mailing List - http://slug.org.au/
Subscription info and FAQs: http://slug.org.au/faq/mailinglists.html


Re: [SLUG] Asynchronous Distributed Filesystem

2008-01-15 Thread Adrian Chadd
On Wed, Jan 16, 2008, Joel Heenan wrote:
 Hi,
 
 In response to Adrian I'm looking for a solution that will work well under 
 RHEL.

I hate to say it, but you bought RHEL, so ask Redhat.

Thats part of the benefit of buying a product with support - you can ask
their engineers what will work well and then go to them when you have
trouble. :)


-- 
- Xenion - http://www.xenion.com.au/ - VPS Hosting - Commercial Squid Support -
- $25/pm entry-level VPSes w/ capped bandwidth charges available in WA -
-- 
SLUG - Sydney Linux User's Group Mailing List - http://slug.org.au/
Subscription info and FAQs: http://slug.org.au/faq/mailinglists.html