Re: [storm] iSCSI feature removal: SLP ?

["Mark S. Edwards" <[e-mail]>]

David,

Thanks for your input regarding word smithing.

To answer your question on the subject of why iSNS is not a MUST 
implement, iSCSI already has a MUST implement discovery mechanism 
called 'SendTargets' as an explicit part of the iSCSI protocol.

iSNS, SLP and anything else were listed as additional mechanisms that 
could be implemented if the vendor wished to expand their 
implementation's discovery capabilities.  The original RFC wording 
heavily implied that SLP was the best/first mechanism to implement in 
such a case.  The argument way back then was that there was an 
existing RFC for SLP and iSNS was 'yet another discovery mechanism' 
being dreamed up by the WG.

Time has moved on, iSNS became an RFC and running code in the real 
world shunned SLP in favour of iSNS.  This discussion has been about 
stopping any confusion that new implementors may have in the future 
about which additional discovery mechanisms to use.  We have a 
mandatory discovery mechanism in iSCSI, this is all about the text 
for recommending what additional mechanism(s) to use.

Hope that answers your questions.


Mark.


At 00:48 07/07/2010, David Harrington wrote:
>Hi,
>
>yes. you now understand what I was looking for, and the new text is
>much better.
>
>I expect we will be asked why iSNS is not MUST implement, to ensure a
>minimum level of interoperability, since that seems to have achieved
>implementer rough consensus. I support having one mchanism as
>MUST-implement to make interoperability more likely. Then, if desired,
>implementers can support additional mechanisms.
>
>Other than a desire to not hurt the feelings of a few die-hard
>supporters of other mechanisms, why is iSNS not a MUST implement, to
>ensure a minimum level of interoperability, since that seems to have
>achieved implementer rough consensus. I support having one mechanism
>as MUST-implement to ensure interoperability. Then, if desired,
>implementers can support additional mechanisms.
>
>dbh
>
> > -----Original Message-----
> > From: Mallikarjun Chadalapaka [mailto:[e-mail]]
> > Sent: Tuesday, July 06, 2010 7:25 PM
> > To: 'David Harrington'; 'Mark S. Edwards'; [e-mail]
> > Subject: RE: [storm] iSCSI feature removal: SLP ?
> >
> > Hi David,
> >
> > Thanks for the detailed note, I now understand your perspective.
> >
> > Here is the text that I have crafted based on all the inputs
> > so far.  Does that look reasonable?
> >
> > iSCSI equipment that needs discovery functions beyond
> > SendTargets SHOULD implement iSNS (see [RFC4171]) for
> > extended discovery management capabilities and
> > interoperability.  Although [RFC3721] implies an SLP
> > implementation requirement, SLP has not been widely
> > implemented or deployed for use with iSCSI in practice.
> > iSCSI implementations therefore SHOULD NOT rely on SLP-based
> > discovery interoperability.
> >
> > Thanks.
> >
> > Mallikarjun
> >
> > > -----Original Message-----
> > > From: David Harrington [mailto:[e-mail]]
> > > Sent: Tuesday, July 06, 2010 12:06 PM
> > > To: 'Mallikarjun Chadalapaka'; 'Mark S. Edwards'; [e-mail]
> > > Subject: RE: [storm] iSCSI feature removal: SLP ?
> > >
> > > Hi,
> > >
> > > I made a suggested change as a contributor because I didn't want
>to
> > > come down with the force of an AD Evaluation. That will come
>later.
> > > But there are some things I know will not make it through IESG
> > > Evaluation, and the IESG is actually working to document
> > some of the
> > > things for which we commonly reject a document. The use of
> > non-RFC2119
> > > language to specify compliance is a common cause for DISCUSSes.
> > >
> > > The storm WG is something of a cross-SDO effort, and that
> > might lead
> > > to some lack of understanding of IETF process, and IESG common
> > > complaints. Let me be sure you understand my objection.
> > >
> > > The IETF does not support "not almost must" compliance language.
> > > We want our standards to be clear and unambiguous.
> > > Therefore, we have RFC2119 compliance keywords, such as
> > SHOULD NOT and
> > > NOT RECOMMENDED.
> > > What you are proposing sounds to me like a SHOULD NOT, or a NOT
> > > RECOMMENDED:
> > >
> > > 4. SHOULD NOT   This phrase, or the phrase "NOT
> > RECOMMENDED" mean that
> > >    there may exist valid reasons in particular
> > circumstances when the
> > >    particular behavior is acceptable or even useful, but the full
> > >    implications should be understood and the case carefully
>weighed
> > >    before implementing any behavior described with this label.
> > >
> > > The full implications appear to be that nobody seems to
> > implement SLP,
> > > and therefore the SLP part of your implementation will not
> > > interoperate with many other existing implementations. Therefore,
> > > implementing SLP appears to be NOT RECOMMENDED. I am not
> > telling you
> > > that you must say it is NOT RECOMMENDED; I am telling you that you
>
> > > should clarify the compliance requirement that represents
> > WG consensus
> > > using RFC2119 language, so the requirement is clear and
>unambiguous.
> > >
> > > As Responsible AD for the storm WG, I need to champion your
> > documents
> > > during IESG Evaluation. If you submit a document to me that
> > says, "for
> > > compliance, SLP is no longer almost must", then I will almost
> > > certainly tell you to revise it before I will champion the
>document
> > > for you. The WG advice SHOULD be written using RFC2119 language,
> > > because if you don't and I send the document for IESG
> > Evaluation, we
> > > will all waste lots of time debating with the other IESG
> > members when
> > > they raise DISCUSSes questioning why we are not using
> > RFC2119 keywords
> > > to describe the compliance rules.
> > >
> > > Does that explain my objection to non-RFC2119 language better?
> > > This will also apply to other WG documents, especially those on
>the
> > > standards track.
> > >
> > > dbh
> > >
> > >
> > > > -----Original Message-----
> > > > From: Mallikarjun Chadalapaka [mailto:[e-mail]]
> > > > Sent: Tuesday, July 06, 2010 12:27 PM
> > > > To: 'David Harrington'; 'Mark S. Edwards'; [e-mail]
> > > > Subject: RE: [storm] iSCSI feature removal: SLP ?
> > > >
> > > > Hi David and all,
> > > >
> > > > With deployment experience, we now know that the original
>"almost
> > > > must"
> > > > recommendation is not valid (ignoring the fact that it
> > was contained
> > > > in an Informational RFC).
> > > >
> > > > But OTOH, we do not have to say "SHOULD NOT" implement
> > SLP either,
> > > > which IMHO is overly constraining and thus unwarranted.
> > > >
> > > > At this point, I plan to make it clear in the consolidated draft
>
> > > > that SLP is not an "almost must" requirement, along the lines of
>
> > > > what DavidB/Mark had proposed.
> > > >
> > > > Please let me know if that is not agreeable to anyone.
> > > >
> > > > Thanks.
> > > >
> > > > Mallikarjun
> > > >
> > > >
> > > >
> > > > > -----Original Message-----
> > > > > From: [e-mail]
> > > > [mailto:[e-mail]] On Behalf
> > > > > Of David Harrington
> > > > > Sent: Monday, July 05, 2010 11:03 AM
> > > > > To: 'Mark S. Edwards'; [e-mail]
> > > > > Subject: Re: [storm] iSCSI feature removal: SLP ?
> > > > >
> > > > > Hi,
> > > > >
> > > > > I think "recommendation is no longer valid" is
> > equivalent to "NOT
> > > > > RECOMMENDED" which is equivalent to "SHOULD NOT".
> > > > >
> > > > > So I recommend "SHOULD NOT" (or if you prefer,
> > rewording to "NOT
> > > > > RECOMMENDED").
> > > > >
> > > > > dbh (as contributor)
> > > > >
> > > > > > -----Original Message-----
> > > > > > From: [e-mail]
> > [mailto:[e-mail]] On
> > > > > > Behalf Of Mark S. Edwards
> > > > > > Sent: Wednesday, June 23, 2010 6:23 PM
> > > > > > To: [e-mail]
> > > > > > Subject: Re: [storm] iSCSI feature removal: SLP ?
> > > > > >
> > > > > > Looks ok to me.  I was trying to be diplomatic in case there
> > > were
> > > > > > any SLP diehards out there waiting to be offended.
> > > > > >
> > > > > > But since none have appeared over the parapet the hammer
> > > approach
> > > > > > that effectively deprecates SLP as an
> > interoperability mechanism
> > >
> > > > > > suits me.
> > > > > >
> > > > > > I agree that we should not specifically prohibit it's
> > > > use, after all
> > > > > > people are free to create and use any discovery mechanism
> > > > they wish
> > > > > > for their own purposes.  Could be SLP, could be LDAP
> > or "Fred's
> > > > > > Bird-Brained Resource Locator", none should be explicitly
> > > barred.
> > > > > >
> > > > > > Mark.
> > > > > >
> > > > > >
> > > > > > At 21:25 23/06/2010, [e-mail] wrote:
> > > > > > >Mark,
> > > > > > >
> > > > > > > > So if pushed I'd write something like:
> > > > > > > >
> > > > > > > > "iSCSI equipment that need discovery functions beyond
> > > > > SendTargets
> > > > > > > > should implement iSNS for extended discovery management
> > > > > > capabilities
> > > > > > > > and maximum interoperability."
> > > > > > > >
> > > > > > > > Since any unit that implements SLP or even LDAP as a
> > > > discovery
> > > > > > > > mechanism is not going to interoperate with the majority
> > > > > > of systems,
> > > > > > > > I'd hesitate to even mention them.  However, if
> > anyone feels
> > >
> > > > > > > > strongly that SLP deserves some mention, then by all
>means
> > > do
> > > > > so,
> > > > > > > > just please don't assert it too strongly.
> > > > > > > >
> > > > > > > > One final point.  Given that iSNS has won the consensus
> > > > > > argument, is
> > > > > > > > there a case for turning the "should' in the
> > above sentence
> > > to
> > > > > > >"SHOULD" ?
> > > > > > >
> > > > > > >Yes, I would use "SHOULD" plus I would add text to say:
> > > > > > >
> > > > > > >         SLP has not been widely implemented or deployed
>for
> > > use
> > > > > with
> > > > > > >         iSCSI, therefore RFC 3121's recommendation
> > that SLP be
> > > > > > >         implemented is no longer valid.
> > > > > > >
> > > > > > >This is one of those situations in which subtlety
> > does not win
> > > > > style
> > > > > > >points (IMHO) but I don't think we need a recommendation
> > > > that SLP
> > > > > > >"SHOULD NOT" be implemented.  In addition (as noted
> > below), RFC
> > > > > 3721
> > > > > > >would need to be added to the list of RFCs that are
> > > > updated by the
> > > > > > >combined iSCSI Internet-Draft.
> > > > > > >
> > > > > > >Any objections?
> > > > > > >
> > > > > > >Thanks,
> > > > > > >--David
> > > > > > >
> > > > > > >
> > > > > > > > -----Original Message-----
> > > > > > > > From: [e-mail]
> > > > [mailto:[e-mail]] On
> > > > > > > > Behalf
> > > > > > >Of Mark S. Edwards
> > > > > > > > Sent: Monday, June 07, 2010 7:00 AM
> > > > > > > > To: [e-mail]
> > > > > > > > Subject: Re: [storm] iSCSI feature removal: SLP ?
> > > > > > > >
> > > > > > > > Sorry for the delay.
> > > > > > > >
> > > > > > > > As you say, RFC3721 was more about guidance and was
> > > > > > written before
> > > > > > > > there was any deployment experience.  At least the text
> > > > > > says 'should'
> > > > > > > > rather than 'SHOULD'.  It might be nicer to all
> > round if it
> > > > > could
> > > > > > > > just be removed and no extended discovery mechanism be
> > > > > mentioned,
> > > > > > > > but we have realities to consider such as an iSCSI array
> > > > > > won't get
> > > > > > > > Microsoft certification unless it supports at least
> > > > basic iSNS -
> > > > >
> > > > > > > > running code and consensus.
> > > > > > > >
> > > > > > > > The one good thing we have now is that iSNS is
> > indeed now an
> > > > > IETF
> > > > > > > > protocol (RFC 4171) and the original argument for SLP
>was
> > > that
> > > > > it
> > > > > > > > was an existing RFC.  So the argument for using SLP
>ahead
> > > > > > of iSNS is
> > > > > > > > no longer relevant.
> > > > > > > >
> > > > > > > > So if pushed I'd write something like:
> > > > > > > >
> > > > > > > > "iSCSI equipment that need discovery functions beyond
> > > > > SendTargets
> > > > > > > > should implement iSNS for extended discovery management
> > > > > > capabilities
> > > > > > > > and maximum interoperability."
> > > > > > > >
> > > > > > > > Since any unit that implements SLP or even LDAP as a
> > > > discovery
> > > > > > > > mechanism is not going to interoperate with the majority
> > > > > > of systems,
> > > > > > > > I'd hesitate to even mention them.  However, if
> > anyone feels
> > >
> > > > > > > > strongly that SLP deserves some mention, then by all
>means
> > > do
> > > > > so,
> > > > > > > > just please don't assert it too strongly.
> > > > > > > >
> > > > > > > > One final point.  Given that iSNS has won the consensus
> > > > > > argument, is
> > > > > > > > there a case for turning the "should' in the
> > above sentence
> > > to
> > > > > > >"SHOULD" ?
> > > > > > > >
> > > > > > > > Mark.
> > > > > > > >
> > > > > > > >
> > > > > > > > At 23:54 21/05/2010, [e-mail] wrote:
> > > > > > > > > > Regarding the feature removal discussion, can
> > I add SLP
> > > to
> > > > > the
> > > > > > >list ?
> > > > > > > > > >
> > > > > > > > > > The RFC 3721 states
> > > > > > > > > >
> > > > > > > > > > "iSCSI equipment that
> > > > > > > > > >        need discovery functions beyond SendTargets
> > > > > > should at least
> > > > > > > > > >        implement SLP, and then consider iSNS when
> > > extended
> > > > > > >discovery
> > > > > > > > > >        management capabilities are required such as
>in
> > > > > larger
> > > > > > >storage
> > > > > > > > > >        networks.  It should be noted that since iSNS
> > > will
> > > > > > > > > > support
> > > > > > >SLP,
> > > > > > > > > >        iSNS can be used to help manage the discovery
> > > > > > information
> > > > > > >returned
> > > > > > > > > >        by SLP."
> > > > > > > > > >
> > > > > > > > > > The implication is that targets and initiators
>should
> > > > > > expect to
> > > > > > >find
> > > > > > > > > > support for SLP before considering iSNS.
> > > > > > > > > >
> > > > > > > > > > I remember our first iSCSI appliance and we spent
> > > > > > ages trying to
> > > > > > >get
> > > > > > > > > > SLP working because it the above wording
> > > > effectively made it
> > > > >
> > > > > > > > > > mandatory.  SLP turned out to be a complete
> > bust and was
> > > > > > >effectively
> > > > > > > > > > killed off when Microsoft refused to support
> > it in their
> > >
> > > > > > > > > > initiator and in their target logo tests.
> > > > > > > > > >
> > > > > > > > > > The result is that today I doubt you could find a
> > > > target or
> > > > > > >initiator
> > > > > > > > > > out there supporting SLP.
> > > > > > > > > >
> > > > > > > > > > For anybody that does still implement SLP we could
> > > change
> > > > > the
> > > > > > >wording
> > > > > > > > > > for SLP a little to remove the implied hierarchy, or
> > > > > > just admit
> > > > > > >that
> > > > > > > > > > running code has created IETF consensus.
> > > > > > > > >
> > > > > > > > >This is interesting.  RFC 3720 (main iSCSI spec)
> > > > never mentions
> > > > >
> > > > > > > > >SLP.  RFC 3721 is not a Standards Track RFC, but
> > > > that's not a
> > > > > > > > >reason not to update it (if people will excuse the
> > > > > > double negative).
> > > > > > > > >
> > > > > > > > >So, Mark, could you send an email to the list with
> > > > > > proposed text on
> > > > > > > > >SLP that you'd like to see in the new consolidated
>iSCSI
> > > > > > draft (no
> > > > > > > > >more than a few sentences, please)?  If that text (as
> > > > > > revised) is
> > > > > > > > >acceptable to the WG, the result would go into the new
> > > > > > consolidated
> > > > > > > > >iSCSI draft, and RFC 3721 would be added to the list of
> > > > > > RFCs that
> > > > > > > > >are being updated by that draft.
> > > > > > > > >
> > > > > > > > >Thanks,
> > > > > > > > >--David
> > > > > > > > >----------------------------------------------------
> > > > > > > > >David L. Black, Distinguished Engineer EMC Corporation,
> > > > > > 176 South
> > > > > > > > >St., Hopkinton, MA  01748
> > > > > > > > >[phone]             FAX: [phone]
> > > > > > > > >[e-mail]        Mobile: [phone]
> > > > > > > > >----------------------------------------------------
> > > > > > > >
> > > > > > > > _______________________________________________
> > > > > > > > storm mailing list
> > > > > > > > [e-mail]
> > > > > > > > https://www.ietf.org/mailman/listinfo/storm
> > > > > >
> > > > > > _______________________________________________
> > > > > > storm mailing list
> > > > > > [e-mail]
> > > > > > https://www.ietf.org/mailman/listinfo/storm
> > > > > >
> > > > >
> > > > > _______________________________________________
> > > > > storm mailing list
> > > > > [e-mail]
> > > > > https://www.ietf.org/mailman/listinfo/storm
> > > >
> >
> >