Excellent summary, Jenny. Let me add a couple of things. I'm just
out of the hospital after getting traceroutes of my heart, and it
seems vaguely appropriate to talk about SNA.
SNA is an old technology, possibly dating back to the Garden of Eden.
Is it any coincidence that the first three letters of evil are SNAke?
Seriously, IBM did a great deal of very good work in making SNA
reliable and extremely measurable. It isn't very flexible, but that
wasn't a design goal.
>Darren,
> you are mostly right, but a bit mixed up.
>
>In SNA there are PU types and LU types.
Another way to think of Physical Units (PU's, which don't have to be
physical) and Logical Units (LU's, which seem illogical at times), is
that PUs are network resources that do not directly involve
application communications. While SNA doesn't precisely fit the OSI
model, LU's are more at layer 6/7.
PUs must activate before LU's can do anything. First, the SSCP (PU5)
activates any FEP (PU4). PU4 functionality is built into the AS/400
operating system and is not a separate box.
Important: the Cisco IBM Interface emulates a 3172, which is a PU2,
_not_ a PU4. You can get some, but not all, PU4 functions by adding
IOS features. If you are considering replacing a FEP, ask yourself
first if it could be replaced with an IBM 3172 before asking if it
could be replaced with a CIP in a router.
Anyway, in a mainframe environment, the SSCP, which owns a set of
resources called a domain, activates the PU4's by creating SSCP-PU
sessions. Each PU4, which is owned by the PU5, owns a set of
resources called a subarea. IBM "logical addresses" are generally
subarea/element. All PU and LU types can have IBM SNA addresses.
Via the PU4, the SSCP then activates the various remote PU2s, with a
couple of exceptions -- the SSCP directly activates the 3172 or CIP,
not going through a PU4.
Terminal controllers are PU2. Once they are activated, they can have
one end of an LU on them. Think of each Secondary LU as a
keyboard/display pair that can be assigned to a particular
"application" under a primary LU.
Primary LUs are more application service interfaces than true
applications. An "application monitor" like CICS, IMS, or TSO
actually provides services to applications, just as the OSI
application layer is not itself an application but a set of functions
that provides services to applications.
>
>The PU type describes the function/capabilities of the device within the SNA
>hierarchy:
>PU types:
>5 VTAM (SSCP)
>4 FEP
>3 there is no PU type 3
Correct. There probably was never a PU3. I have heard two
explanations within IBM, one that there was an early proposal for
lines to be PU3, and the other that two different SNA groups started
numbering, respectively, from 5 and from 3. At least Rich McGee, who
was head of the IBM SNA architecture group, and John Aschenbrenner,
one of the principal designers, couldn't give me a better explanation.
>2 3274/3174 cluster controller, 3777 RJE station, other devices (typically
>connected to a FEP by a leased line running SDLC 1974-1980s) . Token ring SNA
>gateway (1990s). SNA gateway on a LAN with a CISC router running
>DLSw to another
>Cisco router on a token ring (1990s-). Any SNA gateway product running on a
>server will be a PU type 2.
>1 I've forgotten what that was.
PU type 1 were really stupid terminals. Reaching back in my memory
archives, I think it was the 3176 that was an SNA-attached Selectric
typewriter. Whatever the number, there was one. Frightening...a fair
number of people on the list probably haven't ever seen a Selectric
typewriter.
>
>
>then when peer to peer networking came into SNA:
>2.1 APPN (advanced program to program networking) supports independant LUs.
>Includes more recent 3174 etc cluster controllers.
>
>
>The type of the LU describes what data stream it uses to communicate. ie the
>format of the data sent from one LU to another.
>
>LU types
>new-fangled ones:
>6.1 application to application CICS - not used much
>6.2 application to application (APPC)
>classic ones:
>3 printer attached to a 3174 cluster controller using 3270 data stream
>2 3270 display terminal (3278, 3179 etc) your basic "green screen", or
>emulation of one on a PC
>1 SNA character set - printer or RJE station
>0 content of data stream not specified (this was used for program to program
>communication before APPC/APPN came in eg Solve:Netmaster INMC, an Aussie
>product)
>
>SLU :
>in classical SNA a session is between two LUs. One of the LUs is the Primary
>(PLU), the other the Secondary (SLU). The primary LU is the application on the
>mainframe, such as CICS, TSO etc. The secondary LU is the user's terminal.
>
>Hope this isn't too overwhelming and it helps. Feel free to ask more
>questions!
>
>Jenny
>
>---------------------------------------------------------------------
>----------
>Question for all:
>
>In the SNA hierarchy there are defined types 5,4,2 and 1 nodes.
>
>From what I've read so far I've assembled the following:
>
>Type 5 IBM Mainframe (Runs VTAM) SSCP
>& PU5
>Type 4 3705/3725/3745/3746 FEP/Communications Controllers PU4
>Type 2 3174 Cluster Controller PU2
>Type 2 Token Ring SNA Gateway PU2
>Type 1 Token Ring Device LU
>Type 1 TN3270 Terminals SLU
>
>Type 5 & 2 Cisco Router acting as DSPU Concentrator PU5
>& PU2
>
>Is this anywhere near close?
>
>Does an ethernet client running a TN3270 session also appear as an SLU
>or LU?
>
>Sorry if this is basic but SNA is the area that I am definately weakest
>and while I can configure up basic DLSW+ services I prefer to understand
>the underlying principles.
>
>Darren
___________________________________
UPDATED Posting Guidelines: http://www.groupstudy.com/list/guide.html
FAQ, list archives, and subscription info: http://www.groupstudy.com
Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
