Re: TargetAlias in SendTarget response

I concur with Mallikarjun and I would like to add that the major reason 
for my opposition to any inline discovery was that for any installation of 
a nontrivial size having clients machines (initiators) discover targets is 
a bad solution. Clients should know only about "resource managers" (a few 
of them) and those are the ones that keep track  of resources. The reason 
that we have the Sendtargets and the discovery sessions is that some 
people thought that a "resource manager" would be hard to justify in small 
environments - although providing an iSNS or SLP server is not difficult 
even in the tiniest setting. 

Julo



"Mallikarjun C." <[email protected]> 
07/06/06 01:34

To
[email protected]
cc

Subject
Re: [Ips] TargetAlias in SendTarget response






I have some appreciation for the long discovery times
in large SANs.  However, volume-level discovery
consumes the bulk of today's discovery time rather
than target device-level discovery.  So while I see
the proposed optimization for device-level discovery,
I am not sure it is compelling enough for me.

Mallikarjun 



--- Julian Satran <[email protected]> wrote:

> So we are talking milliseconds instead of
> microseconds for very long 
> lasting (hours, days) sessions.
> I am not sure I would support a public key.
> 
> Julo
> 
> 
> 
> David Weibel <[email protected]> 
> Sent by: [email protected]
> 01/06/06 10:23
> 
> To
> Julian Satran/Haifa/IBM@IBMIL
> cc
> [email protected], [email protected],
> [email protected]
> Subject
> Re: [Ips] TargetAlias in SendTarget response
> 
> 
> 
> 
> 
> 
> Julian Satran wrote:
> 
> > The spec authors did not intend to encourage the
> use of discovery 
> > sessions - except for bootstraping a management
> infrastructure.
> > As for the specific TargetAlias you can get it in
> two steps - first a 
> > regular discovery then a "targeted discovery"
> (icluding the target 
> name).
> > Software will take that path to support targets
> that don't understand 
> > the new key so why introduce the new key?
> 
> It is understood this is already supported in a
> multi step way within the
> current iSCSI specification.
> 
> If a target returns an initiator a list of 100
> targets via a 
> SendTargets=All
> and the initiator turns around and logs in and out
> of each target to just
> get the TargetAlias this leads to poor performance. 
> The suggested change
> cuts this process to one step when possible and
> directly improves 
> management
> performance.
> 
> > _______________________________________________
> Ips mailing list
> [email protected]
> https://www1.ietf.org/mailman/listinfo/ips
> 


__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com 

_______________________________________________
Ips mailing list
[email protected]
https://www1.ietf.org/mailman/listinfo/ips

_______________________________________________
Ips mailing list
[email protected]
https://www1.ietf.org/mailman/listinfo/ips