Re: iSCSI 07-95 Comments 14 Through 19

["Mallikarjun C." <cbm@rose.hp.com>]

Thomas,

Let me answer your questions.  Also, FYI:
rev08 state diagrams are posted on http://storage-arch.hp.com.

Thomas Dineen wrote:
> 
> Gentle People:
> 
> First of all I would like to commend the author of the
> state machines. I believe that addition of the state machines
> to the document substantially enhances the overall conciseness.

Thanks!

> 
> Several comments regarding 07-95 section 7 starting on page 118:
> 
> 14) In Sections 7.1, 7.2, 7.3 I would suggest enhancing
> the description of each state and each state transition.
> After reading this section carefully I have concluded that
> additional text would be of great benefit. Please provide
> a few sentences or a paragraph describing each state and
> each state transition.

I will attempt, but most possibly not for rev08.

> 
> 15) The state transition descriptions would be improved by
> linking the logical conditions to those of a TCP Service Interface
> such as Sockets. I am looking for a more formal and concise
> definition of each of the referenced logical conditions.

The current text is a conscious attempt to stay away from 
implementation APIs - but still be formal and concise.  What
is described should be equally valid for hardware state machines,
and software building on traditional sockets API.  I believe
it should be fairly straightforward to map the current states 
to sockets.

> 
> 16) In several places I see the term "[OR}" what dose this mean
> exactly a simple "or"?

Yes!  That's an attempt to avoid a "dangling or" :-), used across 
bullets, as opposed to within the bullet.

> 
> 17) In section 7.3 transition N6 a session state time-out is referenced.
> where is this timer and time-out defined?

The timeout value is not defined by the draft.  But what constitutes
the timeout is defined in rev07-99, appendixD, LogoutLoginMaxTime
description.

> 
> 18) In section 7.2:
> 
> "These additional state transitions may be traversed either by using a
> connection in the LOGGED_IN state with an explicit logout (let us
> call it CSM-E), or on a new transport connection in the FREE state
> with an implicit logout (let us call it CSM-I). This recovery state
> diagram hence is applicable only to the instance of the connection in
> recovery, i.e. CSM-R."
> 
> >From what I can tell you are trying to recover a failed connection
> using the communication services of another existing or newly
> established
> connection. Please add a few sentences or a paragraph describing this
> concept fully.

OK.

> 
> 19) In section 7.1 and the following sections:
> 
> "T7: A login redirection/initiator error/target error was
> received, or login timed out. (Initiator only)"
> 
> In this statement the "/" is used repeatedly: Dose "/"
> really mean "or"?

Yes!  Another attempt to avoid dangling references within the
bullet, also to indicate any one of the three return values
would be considered the same subcase (that of login failure).

> 
> So the statement could read:
> "T7: A login redirection, initiator error, or target error was
> received, or the login timed out. (Initiator only)"
> 
> If you mean "or" why not use 'or"?
> 
> 20) Referring to T7 above, where is the Login timer specified?

A "Login timer" per se is not specified in the draft.  It is 
left as an implementation decision (like an I/O timeout value),
since it is a function of network configuration, RTT etc.

Regards.
-- 
Mallikarjun 


Mallikarjun Chadalapaka
Networked Storage Architecture
Network Storage Solutions Organization
MS 5668	Hewlett-Packard, Roseville.
cbm@rose.hp.com