Incite into reiser5?

Incite into reiser5?

[lares Moreau]

Are there any forms of documentation that have to deal with reiser5 yet.
It was mentioned on ther reiser4 homepage that reiser5 is projected to
be a distributed extention of reiser4.

I am intruiged by the plugin structure of reiser4, and am intested in
Disributed File Systems.

Is Reiser5 on the make, or is it just a prospect at this point?

Later Days,
Lares

Re: Incite into reiser5?

[michael chang]

On 9/1/05, lares Moreau <lares.moreau@gmail.com> wrote:
> Are there any forms of documentation that have to deal with reiser5 yet.
> It was mentioned on ther reiser4 homepage that reiser5 is projected to
> be a distributed extention of reiser4.
> 
> I am intruiged by the plugin structure of reiser4, and am intested in
> Disributed File Systems.
> 
> Is Reiser5 on the make, or is it just a prospect at this point?

Are you kidding? Reiser4 isn't even done yet!  As much as this sounds
awesome, there are some ... underlying issues that still need to be
fixed in Resier4.  No point in putting a distributed layer if the
underlying filesystem suddenly changes formats and panics the
distributed system...

-- 
~Mike
 - Just my two cents
 - No man is an island, and no man is unable.

Re: Incite into reiser5?

[Lares Moreau]

I'm looking into it because I am looking into doing a masters focusing
on distributed file systems.
There was a mention of reiser5 being distributed.
I'm intruiged by the plugin system of reiser4, and I am looking into
generating a distributed plugin.  

You never know.

On Thu, 2005-09-01 at 22:32 -0400, michael chang wrote:
> On 9/1/05, lares Moreau <lares.moreau@gmail.com> wrote:
> > Are there any forms of documentation that have to deal with reiser5 yet.
> > It was mentioned on ther reiser4 homepage that reiser5 is projected to
> > be a distributed extention of reiser4.
> > 
> > I am intruiged by the plugin structure of reiser4, and am intested in
> > Disributed File Systems.
> > 
> > Is Reiser5 on the make, or is it just a prospect at this point?
> 
> Are you kidding? Reiser4 isn't even done yet!  As much as this sounds
> awesome, there are some ... underlying issues that still need to be
> fixed in Resier4.  No point in putting a distributed layer if the
> underlying filesystem suddenly changes formats and panics the
> distributed system...
> 

Re: Incite into reiser5?

[David Masover]

Lares Moreau wrote:
> I'm looking into it because I am looking into doing a masters focusing
> on distributed file systems.
> There was a mention of reiser5 being distributed.

I think that's the definition of reiser5.  I think that 4, 5, and 6 (at 
least, maybe there's a 7?) are defined not in terms of when they will 
actually be done, but what features they will contain.  If I remember 
correctly, 6 might not need 5 to be implemented first (but both need 4 
for its plugin architecture).

Or, in other words, the version number refers to a major, sweeping 
feature.  4 is plugins, 5 is distributed, and I don't remember 
understanding 6 very well -- but I was young when I read about it.

Re: Incite into reiser5?

[Hubert Chan]

On Fri, 02 Sep 2005 13:34:14 -0500, David Masover <ninja@slaphack.com> said:

> Lares Moreau wrote:
>> I'm looking into it because I am looking into doing a masters
>> focusing on distributed file systems.  There was a mention of reiser5
>> being distributed.
>> I'm intruiged by the plugin system of reiser4, and I am looking into
>> generating a distributed plugin.

That would be great if you could get something implemented.  I think
that Linux needs a good distributed filesystem.

> I think that's the definition of reiser5.  I think that 4, 5, and 6
> (at least, maybe there's a 7?) are defined not in terms of when they
> will actually be done, but what features they will contain.  If I
> remember correctly, 6 might not need 5 to be implemented first (but
> both need 4 for its plugin architecture).

I believe that is true.  6 and 5 are orthogonal in terms of features,
and I remember Hans saying that 6 might come first, depending on what
kind of funding they get.

> Or, in other words, the version number refers to a major, sweeping
> feature.  4 is plugins, 5 is distributed, and I don't remember
> understanding 6 very well -- but I was young when I read about it.

IIRC, 6 is when we get the fancy Future Vision querying.

-- 
Hubert Chan <hubert@uhoreg.ca> - http://www.uhoreg.ca/
PGP/GnuPG key: 1024D/124B61FA
Fingerprint: 96C5 012F 5F74 A5F7 1FF7  5291 AF29 C719 124B 61FA
Key available at wwwkeys.pgp.net.   Encrypted e-mail preferred.

Re: Incite into reiser5?

[David Masover]

Hubert Chan wrote:
> On Fri, 02 Sep 2005 13:34:14 -0500, David Masover <ninja@slaphack.com> said:
> 
> 
>>Lares Moreau wrote:
>>
>>>I'm looking into it because I am looking into doing a masters
>>>focusing on distributed file systems.  There was a mention of reiser5
>>>being distributed.
>>>I'm intruiged by the plugin system of reiser4, and I am looking into
>>>generating a distributed plugin.
> 
> 
> That would be great if you could get something implemented.  I think
> that Linux needs a good distributed filesystem.

It has one.  It's called Lustre.

What Linux needs is good, distributed filesystem that runs on the 2.6 
kernel, without users having to pay licensing fees.

Re: Incite into reiser5?

[Lexington Luthor]

David Masover wrote:
> It has one.  It's called Lustre.
> 
> What Linux needs is good, distributed filesystem that runs on the 2.6 
> kernel, without users having to pay licensing fees.
> 

Lustre is expensive, but the -mm kernel now has ocfs (or is it ocfs2?) 
as well as gfs.

What is wrong with either of those two?

LL

Re: Incite into reiser5?

[David Masover]

Lexington Luthor wrote:
> David Masover wrote:
> 
>> It has one.  It's called Lustre.
>>
>> What Linux needs is good, distributed filesystem that runs on the 2.6
>> kernel, without users having to pay licensing fees.
>>
> 
> Lustre is expensive, but the -mm kernel now has ocfs (or is it ocfs2?)
> as well as gfs.

Don't know about ocfs, I'll look it up later.

But, AFAIK, GFS is for things like ATA-over-ethernet, where you have
multiple machines attached to the same hard drive -- not for situations
where multiple machines each have their own drive, and you want them to
appear as one logical drive.

Re: Incite into reiser5?

[Lexington Luthor]

David Masover wrote:
> Don't know about ocfs, I'll look it up later.
> 
> But, AFAIK, GFS is for things like ATA-over-ethernet, where you have
> multiple machines attached to the same hard drive -- not for situations
> where multiple machines each have their own drive, and you want them to
> appear as one logical drive.
> 

GFS uses shared storage, but that does not mean it needs to be central.

I imagine one can use GFS in a cluster of machines where each machine 
shares its drive as a block device, which can be combined using LVM to 
create one giant logical network drive which can be mounted using GFS on 
each machine.

One could easily set this up with the code currently in the kernel, GFS 
supports redundancy and multiple journals so reliability is not a problem.

LL

Re: Incite into reiser5?

[Lares Moreau]

On Fri, 2005-09-02 at 14:42 -0500, David Masover wrote:
> Hubert Chan wrote:
> > On Fri, 02 Sep 2005 13:34:14 -0500, David Masover <ninja@slaphack.com> said:
> > 
> > 
> >>Lares Moreau wrote:
> >>
> >>>I'm looking into it because I am looking into doing a masters
> >>>focusing on distributed file systems.  There was a mention of reiser5
> >>>being distributed.
> >>>I'm intruiged by the plugin system of reiser4, and I am looking into
> >>>generating a distributed plugin.
> > 
> > 
> > That would be great if you could get something implemented.  I think
> > that Linux needs a good distributed filesystem.
> 
> It has one.  It's called Lustre.
> 
> What Linux needs is good, distributed filesystem that runs on the 2.6 
> kernel, without users having to pay licensing fees.
> 
> 
My Masters is going in a different direction.  My research has to do
with continuous fs operation even with server disconnects and weak
connections.   I Have looked into Coda fs however that system requires
major reworking to impliment my theories.  I like the plugins.

Re: Incite into reiser5?

[David Masover]

Lexington Luthor wrote:
> David Masover wrote:
> 
>> Don't know about ocfs, I'll look it up later.
>>
>> But, AFAIK, GFS is for things like ATA-over-ethernet, where you have
>> multiple machines attached to the same hard drive -- not for situations
>> where multiple machines each have their own drive, and you want them to
>> appear as one logical drive.
>>
> 
> GFS uses shared storage, but that does not mean it needs to be central.
> 
> I imagine one can use GFS in a cluster of machines where each machine
> shares its drive as a block device, which can be combined using LVM to
> create one giant logical network drive which can be mounted using GFS on
> each machine.

Great -- no load balancing or local caching at all, then?

Re: Incite into reiser5?

[Hubert Chan]

On Fri, 02 Sep 2005 16:42:28 -0600, Lares Moreau <lares.moreau@gmail.com> said:

> My Masters is going in a different direction.  My research has to do
> with continuous fs operation even with server disconnects and weak
> connections.  I Have looked into Coda fs however that system requires
> major reworking to impliment my theories.  I like the plugins.

That's good.  IMHO, that's the interesting part, from a research point
of view.  From what I can tell (as someone who isn't in that area),
there are a lot of hard questions, and it is very useful.  I hope that
you can get good results with your theories, and I would be interested
in hearing about what you find out.

-- 
Hubert Chan <hubert@uhoreg.ca> - http://www.uhoreg.ca/
PGP/GnuPG key: 1024D/124B61FA
Fingerprint: 96C5 012F 5F74 A5F7 1FF7  5291 AF29 C719 124B 61FA
Key available at wwwkeys.pgp.net.   Encrypted e-mail preferred.

Re: Incite into reiser5?

[Lexington Luthor]

David Masover wrote:
> Great -- no load balancing or local caching at all, then?
> 

Its actually pretty fast - almost raw disk speed for small clusters, but 
not much slower for larger ones - try it out.

Its not as fast as reiser5 may end up, but its here and working right 
now. It is load balanced, and it does local caching. Use it before you 
start criticizing it.

It has plenty of its own problems (mainly relating to the DLM), but none 
of the things you mentioned are a problem.

For the time being, it beats every other non-commercial distributed file 
system hands down.

I look forward to reiser5, but GFS is just fine for now.

LL

Re: Incite into reiser5?

[Lexington Luthor]

Lares Moreau wrote:
> My Masters is going in a different direction.  My research has to do
> with continuous fs operation even with server disconnects and weak
> connections.   I Have looked into Coda fs however that system requires
> major reworking to impliment my theories.  I like the plugins.
> 

Have you used the Redhat cluster suite? From what I can gather from your 
post, it does everything you mentioned. Perhaps you can use it as a 
subject of your thesis.

LL

Re: Incite into reiser5?

[Lares Moreau]

Thanks for the pointer LL,

My researvch is to do the the mobile client. Connection dependent client
fs caching/hoarding algorithyms.
On Sat, 2005-09-03 at 10:00 +0100, Lexington Luthor wrote:
> Lares Moreau wrote:
> > My Masters is going in a different direction.  My research has to do
> > with continuous fs operation even with server disconnects and weak
> > connections.   I Have looked into Coda fs however that system requires
> > major reworking to impliment my theories.  I like the plugins.
> > 
> 
> Have you used the Redhat cluster suite? From what I can gather from your 
> post, it does everything you mentioned. Perhaps you can use it as a 
> subject of your thesis.
> 
> LL
> 

Re: Incite into reiser5?

[Nikita Danilov]

Lexington Luthor writes:
 > David Masover wrote:
 > > Great -- no load balancing or local caching at all, then?
 > > 
 > 
 > Its actually pretty fast - almost raw disk speed for small clusters, but 
 > not much slower for larger ones - try it out.

Lustre and GFS target quite different environments. GFS is mainly for
small-to-medium clusters with up to hundred nodes, for situations when
one needs fail-over and things like that. Lustre is for the large
clusters with tens thousand nodes, hundreds object stores, with very
fa{,s}t transports both through the network (elan) and to the storage
(FC). This requires completely different architecture of IO and
meta-data management.

 > 
 > Its not as fast as reiser5 may end up, but its here and working right 
 > now. It is load balanced, and it does local caching. Use it before you 
 > start criticizing it.
 > 
 > It has plenty of its own problems (mainly relating to the DLM), but none 
 > of the things you mentioned are a problem.
 > 
 > For the time being, it beats every other non-commercial distributed file 
 > system hands down.
 > 
 > I look forward to reiser5, but GFS is just fine for now.
 > 
 > LL
 > 

Nikita.

 >