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

[Dirk Munk]

Johnny Billquist wrote:
> On 2016-10-06 01:38, Dirk Munk wrote:
>> Johnny Billquist wrote:
>>> On 2016-10-05 16:06, Dirk Munk wrote:
>>>> Unless I'm mistaken, you're writing that DECnet Phase IV over Mulinet
>>>> tunnels should be implemented, but DECnet Phase V + OSI over IP "should
>>>> not make sense".
>>>
>>> Uh! No. What I am writing (please reread it in case you still don't get
>>> it) is that DECnet Phase IV over Multinets are *already* implemented,
>>> and I don't think it make sense to start deleting that code.
>>
>> TCP/IP services 10.5 will not be the same as Multinet, it is *based* on
>> Multinet. Otherwise they could start shipping now.
>
> And read what I wrote one more time.
>
> If this requires work, then I don't think it should be done. But
> assuming that they are pretty much using the existing Multinet stack
> (seems like a reasonable assumption, or else they are writing something
> completely new, in which case there is no point of saying that they are
> taking the Multinet stack in the first place), then there is an existing
> DECnet-over-IP solution, that is probably pretty much no work to bring
> along. In which case I believe it makes sense to include it.
>
>>> I also say that I don't think it makes sense to implement new things for
>>> DECnet, so if Phase V will not work with the new stack, I doubt it makes
>>> sense.
>>
>> So in fact you're saying that if DECnet Phase V doesn't work immediately
>> with the Multinet stuff, forget about DECnet over IP. So everywhere
>> where DECnet Phase V systems are in use on IP only networks, instead of
>> using your antique Phase IV networks, forget about these customers. If
>> they are using OSI with something standard as FTAM, or if they have used
>> the OSI API's to build their own OSI applications, who cares. Mr.
>> Billquist doesn't need anything else then the old DECnet Phase IV, so
>> the rest of all the VMS customers should also return to Phase IV. Are
>> you sure you don't want to return to Phase III?
>
> DECnet-over-IP is not a Phase V specific function. Please be more
> correct in your statements.
>
> And I do believe that it some major work needs to be done to make it
> work, it is very questionable if it should be done. How many places do
> you actually know that use DECnet phase V over IP?
>
> Ignore my personal preferences for a second. They are not relevant here.
> The question is what makes sense for VSI. Does working on DECnet make
> sense? I'd say more or less "no".
>
> TCP/IP is the protocol that everyone is using, and is what is needed
> going forward. Keeping DECnet around as it is makes sense. No point in
> just deleting it. But new work? No.
>
>> Luckily VSI has more sense then you have.
>
> Who knows.
>
>>>>> Are you intentionally just not
>>>>> understanding what I say, or are you in fact just not understanding?
>>>>>
>>>>> I was making a comment that Phase IV with Multinet works just fine
>>>>> talking to other Phase IV multinet hosts, using IP as a carrier. You
>>>>> complain that it's "non-standard", to which I point out that nothing
>>>>> could be less relevant than that comment. If you have two phase IV
>>>>> nodes, they can talk to each other. Would having an RFC (which by the
>>>>> way does not mean it's any more standard) make any difference? No, it
>>>>> would not. The communication works equally fine with or without an
>>>>> RFC.
>>>>> And it does not pretend that it will communicate with anything except
>>>>> another Multinet node.
>>>>
>>>> And that's the point, a phase V system using IP as transport stack can
>>>> talk OSI over IP to non VMS systems that have no DECnet.
>>>
>>> Of which there are none. But sure, in theory...
>>
>> No of course not.
>>
>> Look for a company called Marben. Just one statement from their web site:
>>
>> "Marben has worked on OSI from its first year of standardisation in
>> 1985. Marben has the largest spectrum of customers on OSI for
>> telecommunication from SUN and HP to Adtran, Alcatel Lucent, Fujitsu,
>> NEC, Tellabs, 600 000 stack has been deployed worldwide for SONET/SDH
>> management.
>> Marben has also sold its OSIAM stack products to large industrials such
>> as BAE, Suez in the requirement of OSI for automation."
>
> Do you even know what SDH/SONET is??? This is not something you hook up
> your VMS system to. This is what network operators use on backbones. And
> you then have management protocols for this, which can use OSI. And the
> management are for the SDH/SONET links, and is inband communication on
> those links. But this is not something you can talk with from your Phase
> V DECnet OSI.
>
> OSI is such a wide range of protocols, which many times have no
> commonality that it is pretty much meaningless to just say "OSI", or
> assume that two machines can communicate just because they use "OSI".
>
> Jeez.
>
>> How many DECnet Phase IV stacks are still out there??
>
> No idea. Probably more than Phase V.
>
>     Johnny
>

Well then, let me give you a very good reason to scrap Phase IV. These 
days it is normal to use many Ethernet interfaces on a system, hook up a 
system with 4 Ethernet interfaces to one switch and 4 Ethernet 
interfaces to another switch (redundancy) for instance. Every interface 
can be used for a specific purpose. Datacenters are designed around such 
infrastructure principles.

What happens when you do that with Phase IV, or Phase V in Phase IV 
compatibility mode? Every interface gets the same Phase IV MAC address, 
AA-00-04-00-XX-YY. Three of the four Ethernet interfaces will be 
blocked, and if the switches are connected, seven of the eight will be 
blocked.

So if you use Phase V, never use it in Phase IV compatibility mode 
unless you really have to.