[Info-vax] What would you miss if DECnet got the chop? Was: "bad select 38" (OpenSSL on VMS)

[Dirk Munk]

Michael Moroney wrote:
> Johnny Billquist <bqt at softjar.se> writes:
>
>> From DECnet point of view, it is just a simple line. Called something
>> like TCP-0-0. And you work with it just like any other line in DECnet.
>> It's a point-to-point line, over which DECnet establish a circuit to the
>> remote machine. No different than if you were to just have a simple
>> RS-232 cable between two machines, using DDCMP, and then have DECnet
>> communicate over that.
>
>> Really, this is identical.
>
> Oh, that is interesting. So the Multinet DECnet over IP is just a
> virtual point-to-point serial line, not a fake Ethernet like I was
> thinking it might be.  Real point-to-point links were common in the
> early days of DECnet before ethernet pretty much took over for
> short distance links.
>
>> Multinet then have a tool to define the TCP/IP connection between the
>> two points, which is totally outside the knowledge of DECnet.
>
> So for a more complicated network, you'd need several of these virtual
> point-to-point links so everyone could talk to everyone else.
>
> I don't know offhand if DECnet V has support for point-to-point links
> at all

Not with the old DDCMP lines if I remember correctly. However you can 
use X.25 lines.

> but if it does, Multinet's virtual p-p links should work with
> it as well.

I would assume Phase V can not use these links.

> But with OSI over IP, there is no need other than to talk
> to a Phase IV node.

No, that is impossible. DECnet Phase V has three transport stacks:

1. DNA, AKA Phase IV. Can be used for communication between DECnet Phase 
V nodes, and with DECnet Phase IV nodes. Uses DECnet Phase IV MAC 
addresses etc.

2. CLNS transport classes. Can be used between DECnet Phase V nodes, and 
non-DECnet OSI nodes. Uses CLNS addresses, can not be used to 
communicate with DECnet Phase IV nodes.

3. IP transport. Can be used between DECnet Phase V nodes, and 
non-DECnet OSI nodes. Uses IP addresses, can not be used to communicate 
with DECnet Phase IV nodes.

>
>> And in the Multinet tool, you define the remote host and port. Multinet
>> will then establish that connection, using TCP or UDP. And once it is
>> up, then from the DECnet point of view, you have a link which transports
>> bytes between the two nodes, using that line.
>
> That makes sense.  A TCP link from A to B with a driver for a virtual
> serial line on each end that puts the stuff over the TCP link.
> (Or UDP. I don't know the advantage of one over the other for something
> like that)
>
>> DECnet can route things just as normal. One more line does not change
>> anything fundamentally. It's just a line.
>
>
>
> As to the argument over the Multinet Phase IV p-p connections and the
> Phase V OSI over IP stuff, there is no reason why both can't coexist,
> as long as they use different TCP/UDP ports. To IP it's just more
> TCP/UDP packets.

I don't think that will be possible. I don't think it is possible to 
combine the DNA and CLNS stacks with the IP stack. With DNA and CLNS a 
machine has an address, with IP an interface has an address. Two very 
different philosophies.

>
> As to VSI, for Phase IV it's just a matter of not breaking anything
> in the Multinet code.

That depends on your point of view. VSI TCP/IP services 10.5 will be 
*based* on Multinet, it is not the same as Multinet. These Multinet 
DECnet over IP lines were never part of TCP/IP services, so from that 
point of view they don't *have* to be included. I'm not saying they 
shouldn't be included, let me be very clear about that.

If they are included, they should be part of the DECnet Phase IV kit, 
after all DECnet Phase IV is the only possible user of these connections.

I assume the IP stack will be a standard part of VMS, if you want DECnet 
you will have to make a choice between Phase IV and Phase V.

> For Phase V it may depend on working on the
> Multinet PWIP driver, or maybe it will just work.
>