IBM 3174 C 6.4 Microcode Disks?

[Grant Taylor]

On 2/18/19 1:20 AM, Dave Wade via cctalk wrote:
> So the 3174 does not do this. 3270 terminals don't talk SNA/3270 to the 
> 3174 as defined in the IBM 3270 data streams. They are usually pretty dumb 
> and from what I can gather all keystrokes go to the 3174 just as for an 
> ASCII terminal.  It only becomes a 3270 protocol when it exits the 3174.

Okay.

So I misunderstood ~> misspoke about the protocol on the coax side of 
the 3174.

> You get dumb terminals, either ASCII or 3270 Screens in, and at the 
> other side you connect 3270 over SNA/SDLC, SNA/Token Ring, BiSync, 
> X.25 or whatever.

But it sounds like the 3174 is functioning as an application layer 
gateway in some capacity in that it translates from one thing / protocol 
to another.

> That’s exactly what a 3174 does. IBM calls it a Terminal Controller.

ACK

> I think there is, but to me a gateway has LAN protocols on both sides.

Ah.

To me, a "gateway" can be network layer, application layer, or do other 
things.

I'll give you that a "gateway", as in a network layer gateway or router, 
does have network protocols that it routes / bridges between.  (They may 
be on a single interface, a la one-armed-router.)

> The 3174 NEVER accepts any sort of incoming connections. Just physical 
> terminals.

Um … what do you consider the connections from remote 3174s (physically 
/ logically) connecting to a local 3174 via Token Ring / Ethernet / SDLC 
to be if not connections to the local 3174?

I'm using IBM's definition for "local" and "remote" in this context.

I can see some wiggle room for the connections from the remote 3174 
being to the mainframe via the local 3174 and not actually to the local 
3174.

That being said I still think that the 3270 connection from the RS/6000 
are addressed /to/ the local 3174's Token Ring (MAC) address.  Or is 
this the above wiggle room too?

> When used to connect network traffic to a mainframe the 3174 does not 
> terminate the TCPIP connection., it passes the frames across to the 
> channel. I may be wrong its been a long time since I did this and I 
> don't want to go delving into the VTAM documentation.

The reading that I've done since the start of this thread makes me think 
that the connections from the RS/6000 would be SNA over Token Ring.  As 
I understand it, this means that they are 802.2 LLC SNAP frames carrying 
something other than TCP/IP.

Perhaps the 3174 is receiving those frames and passing them on to the 
mainframe via some form of routing or bridging.

Or perhaps the 3174 is extracting the SNA data off of the Token Ring 
frames and passing just the SNA application layer data to the mainframe.

I suspect that VTAM documentation is in my future if I truly want to 
understand this.  Or maybe I'll get lucky and someone can answer my 
pointed questions.

> Its kind of odd. RS232 (so X.25/SDLC/HDLC/Bi-Sync) connections can only 
> be used to connect to a Mainframe, not another 3174.

That's contrary to what I have been reading this week.

Based on the reading that I've done (I can dig for sources if you want 
me to), a remote 3174 can connect to a local 3174 via Token Ring / 
Ethernet / SDLC.

This implies that the remote 3174 is connecting to another 3174.  (See 
additional comments below.)

> The Token Ring or Ethernet interface can be used to connect traffic to 
> the mainframe But from what I remember the 3174 isn't too involved at 
> this level it is acting as a network router/bridge.

"too involved" is critical.

> Just to confuse things this is an IBM manual where IBM does use it as a 
> "gateway"...

~chuckle~  Very little about IBM is simple.

> http://bitsavers.informatik.uni-stuttgart.de/pdf/ibm/lan/GG24-3366-0_3174_Remote_Token-Ring_Gateway_Feb89.pdf

I have seen virtually identical diagrams to the one on page 15 where the 
NCP was a local 3174 instead of the 3720 / 3725 / 3745.

Notice how the listed 3174 sub models are all the remote variety.

Take a look at page 54 of the following pdf.

http://www.bitsavers.org/pdf/ibm/3174/GG24-4172-0_Using_3174_in_TCP_IP_Networks_Jun94.pdf

The downstream 3174-13R can talk to either the upstream 3174-11L or the 
3172.

Figure 244 on page 259 shows the same.

> so using the Token Ring interface on a remote 3174 to connect SNA traffic 
> to the host via SDLC....  ... again no TCPIP, working at the frame level, 
> and the host end cannot be a 3174...

Figure 244 on page 259 tends to refute that.

This document seems to be from 94 verses the document you linked to 
seeming to be from 89.  Maybe things changed in the intervening years.

> That really muddies the waters because it uses the term "3270" connection 
> in two senses.  It uses it to refer to the co-ax type connection from 
> a work station (CUT or DFT) with with 3270 over Channel/SNA as defined 
> in the 3270 data streams manual and these really are different protocols.

I agree to your prior comment that this traffic between the terminals 
and the 3174 terminal controller is not 3270.

> That’s where you are going wrong. The protocols that the 3174 supports 
> between other 3174s are IBM SNA protocols.  The "other 3174s" do not 
> need to be 3174s and can be any SNA device.

Some of the documents that I've looked at this week have explicitly 
shown a routed TCP/IP network between the upstream / host 3174 and the 
downstream 3174.

> Where does it say that?. In particular on page 39 it says..

I've lost track of what document you're referencing.  I don't see 
anything like the following on (file) page 39 and page (number) 39 is a 
source code listing.

> IEEE 802.2
> • PU 2/LU 2
> • PU 2.1/LU 6.2 (in migration mode)
> 
> In the sort of use Kevin is talking about for connecting to Mainframe 
> channels there is generally no TCPIP on the 3174.  In effect it looks 
> like a Mainframe NIC...
> 
> But the 3174 generally doesn't use TCPIP on the ring...

I'll agree that the RS/6000 may be using SNA on 802.2 LLC SNAP frames 
directly and not using TCP/IP.

> The TN3270 traffic originates from the 3174 and terminates on the 
> Mainframe.

Hum....

> TN3270 (and normal Telnet) traffic NEVER terminates on the 3174...

Normal telnet traffic most certainly does terminate on the 3174 when 
it's being used as a gateway for 3270 / CUT terminals to Telnet sessions 
on other systems, like the RS/6000.

But that's not what we've been discussing.

> IBM describes it as LU6.2.....

LU6.2 doesn't translate to what a network sniffer would show things as, 
which is what I'm trying to determine.

I've gathered that it's 802.2 LLC SNAP frames.  Which work perfectly 
fine on both Ethernet and Token Ring.  As such, they can also be bridged 
between the two.

> See above....
> 
> That’s how it connects, but this is not the normal operating mode of 
> a 3174.

I get the impression that "normal" is highly subjective.  Especially 
with later firmware on 3174s.

> Yes, but a 3270 terminal does not talk 3270 protocol to the 3174....

ACK

> Yes and the waters get muddied because the 3174 has had extra features 
> added along the way that allow it to be used in odd ways....

Agreed.  I think I'm talking about things on the odder end of the spectrum.



-- 
Grant. . . .
unix || die