I understand the rationale for phase 5 and years ago when I could remember ncl it was great: lots of information, good for troubleshooting.
But today, 25 years later, I find myself predominantly installing NCP and not ncl.
Phase 5 takes a long time to start up on a VAXstation 3100 class machine (sub 10 VUPS) and is better handled on a VAXstation 4000 90A. Once it runs phase 5 performs on par with phase 4.
What I like best about phase 4 is the SHOW NET(/OLD) command because it tells me the nodes that are up and running. Very useful in a production environment.
I feel that if phase 5 would have had a similar command that it may have been a lot more popular. Why isn't it there ?
Verzonden vanaf mijn draadloze BlackBerry -toestel
On Jul 17, 2011, at 4:39 PM, <hvlems at zonnet.nl> wrote:
Well the term bridge is firmly written in my own memory :-)
Tou mentioned that other company, and I'm trying to remember that name too. Was it something with 'sea" in it?
No... I just found it. Kalpana. Their technology claim to fame was cut through bridging, as opposed to store & forward, so they invented the term "switching" for that. But the cut through stuff is just an optimization, and on half-duplex Ethernet a questionable one at best. It existed somewhat with 100Mb Ethernet, I think, and disappeared in Gigabit Ethernet only to come back with 10G Ethernet. Interesting...
Still, though, it's just a bridge, no matter what you call it.
paul
On 17.7.2011 23:17, Paul Koning wrote:
On Jul 17, 2011, at 2:29 PM,<hvlems at zonnet.nl> wrote:
What is the difference between a bridge and a switch? I don't want to get into a discussion about layer 2 and layer 3 swutching but afaik a bridge is a two port layer 2 switch.
No. "bridge" is the original term, and while the first bridges were 2 port, that never was a limitation.
"Switch" is a term invented by some company whose name escapes me (bought by Cisco way back when) for the purpose of confusing the public into thinking they had something different. Not so, it was just a bridge.
These days "switch" seems to be the common term, but the two words have always been synonyms.
paul
.
Paul answered already what I also was going to say; that bridge was the orginal term.
In fact, a switch is a multiport bridge. A switch can of course have additional features like VLAN support etc but basically it is a bridge with all ports connected together.
Kari
Well the term bridge is firmly written in my own memory :-)
Tou mentioned that other company, and I'm trying to remember that name too. Was it something with 'sea" in it?
------Origineel bericht------
Van: Paul Koning
Afzender: owner-hecnet at Update.UU.SE
Aan: hecnet at Update.UU.SE
Beantwoorden: hecnet at Update.UU.SE
Onderwerp: Re: [HECnet] DECnet et al
Verzonden: 17 juli 2011 22:17
On Jul 17, 2011, at 2:29 PM, <hvlems at zonnet.nl> wrote:
What is the difference between a bridge and a switch? I don't want to get into a discussion about layer 2 and layer 3 swutching but afaik a bridge is a two port layer 2 switch.
No. "bridge" is the original term, and while the first bridges were 2 port, that never was a limitation.
"Switch" is a term invented by some company whose name escapes me (bought by Cisco way back when) for the purpose of confusing the public into thinking they had something different. Not so, it was just a bridge.
These days "switch" seems to be the common term, but the two words have always been synonyms.
paul
Verzonden vanaf mijn draadloze BlackBerry -toestel
On Jul 17, 2011, at 1:38 PM, Johnny Billquist wrote:
On 2011-07-17 16.55, Bob Armstrong wrote:
Paul Koning wrote:
It is legal to have multiple areas on a single Ethernet;
the routing spec covers that case explicitly. If that is done,
end nodes will communicate directly to a destination on the
Ethernet even if off-area ...
Given that Johnny's bridge program effectively creates a big Ethernet,
does that imply that HECnet end nodes in different areas can communicate
without an area router?
Yes, at least partially.
They can comminucate directly with machines sitting on the same ethernet segment, even if those machines are in another area.
However, I think that an endnode will not pick any router in another area as its designated router.
Correct.
The endnode rule is:
1. If the destination is in the cache, send to it.
2. If there is a designated router (in my area), send to it.
3. Send directly.
"Directly" means to the MAC address formed by prefixing aa-00-04-00 onto the little endian node address.
The only difference between Phase IV and IV+ is how the cache works. In phase IV, it's an "on Ethernet" cache. A node sets a bit in the (long format, i.e., Ethernet) packet header when it originates a packet. A router clears that bit if it forwards onto a different circuit, but leaves it alone if it forwards onto the same Ethernet as the packet arrived on. The receiving end node makes a cache entry if the bit it set.
For Phase IV+, the cache is a "previous hop" cache. It remembers the node address (or MAC address, same thing) from the Ethernet source address of incoming packets, and associates that with the node address of the arriving packet. So if you have several routers, and the designated router is not the optimal path to destination X, phase IV will keep sending to the DR, but phase IV+ will send all packets after the first to the router that's best for X, because that's the router which sent the reply back to the endnode.
As for routers on a multi-area Ethernet, yes, the usual rules of the hierarchy apply: all areas must have L2 routers in them. And traffic from an L1 router will visit at least two L2 routers if it goes out of area. So in a multi-area Ethernet, out of area but on-Ethernet traffic from an L1 router will take 3 hops, but if the destination is an end node, the reply will go direct (Phase IV or IV+, either way). And if the source is L2, it will take 2 hops.
paul
On Jul 17, 2011, at 2:48 PM, Johnny Billquist wrote:
On 2011-07-17 20.35, hvlems at zonnet.nl wrote:
DECnet phase 4 is about as old as ethernet and VMS. Ethernet was the main driving force for phase 4 owing to the large number of adjacent nodes and the sheer number.
Not really. Phase IV is about as old as ethernet, yes. VMS is older...
Phase IV was invented for Ethernet support. Originally the plan was to do what became Phase V, but there was a lot of resistance against anything that complicated back in the early 1980s. So "Phase 3E" was invented, a small generalization of Phase 3, and that was subsequently renamed.
Phase 3 does not have areas and only recognizes 255 hosts max. It did have circuit routing (of course)
Yes. About the same thing as L1 routing. I think it might even be partially compatible, within the restrictions on the number of nodes...
Three major differences: Phase IV adds Ethernet support, including "long packet format". That, by the way, was intended to be compatible with the original "phase IV" which used link state routing. But that never saw the light of day; while DECnet Phase V uses link state routing also, and a number of the innovations carried over (like the sequence numbering), the packet format is completely different because of OSI. So the "long packet format" is effectively an orphan, a bit of extra complexity that ended up serving no purpose.
The second difference is that Phase IV adds hierarchical routing.
The third is that it adds partial routing messages. The routing table contains entries for every node in the area (L1) or every area (L2) but the corresponding routing messages can contain just a couple of entries, if only those changed. By contrast, Phase III always sends the entire routing table (up to 255 entries).
paul
On Jul 17, 2011, at 2:29 PM, <hvlems at zonnet.nl> wrote:
What is the difference between a bridge and a switch? I don't want to get into a discussion about layer 2 and layer 3 swutching but afaik a bridge is a two port layer 2 switch.
No. "bridge" is the original term, and while the first bridges were 2 port, that never was a limitation.
"Switch" is a term invented by some company whose name escapes me (bought by Cisco way back when) for the purpose of confusing the public into thinking they had something different. Not so, it was just a bridge.
These days "switch" seems to be the common term, but the two words have always been synonyms.
paul
On 2011-07-17 21.16, Peter Lothberg wrote:
ANF10!
DISCO DUCK
DISCO FEVER
ANF10 is a fond memory... Do I remember right that it was based on PDP-8
machines acting as connection points? Did they talk serial lines between
each other?
=A0 =A0 =A0 =A0Johnny
The PDP8 thing is a DN92, it has a sync interface and a bunch of
terminal lines and a lineprinter interface. It was supposed to talk to
a DS10, sync interface on PDP10 IO bus.
ANF10 is talked by the TOPS10 hosts over DTE (to FE PDP11 on KL), DL10
(DEC10 DMA/IObus to Unibus) and NIA-Ethernet (we got a ether_type for
it..)
There is ANF10 code for PDP11 that run on 40's or 34's.
Ah. I thought the DN92 was somehow related to ANF10.
Did ever anything but TOPS-10 support ANF10?
(Whatever code ran on the FE RSX system don't count... :) )
The PDP11 code can talk DDCMP over ASYNC or SYNC interfaces, DQ11,
DMR11, DMC11. So you can do Wan things. In Sweden we had DEC10's in
Stockholm and Linkoping networked with a 9.6K sync link.
The DEC2020 talks DDCP with a DUP11 and a KMC (I think it was KMC,
unibus card with microdode that basically did DDCMP.)
Yeah, I seem to remember that the KMC was the card with DDCMP on board.
Anf-10 has link-state routing like OSPF/ISIS but only one area.
I was very ignorant at the time I played with this stuff. Exactly what could you do over ANF10? I remember connecting between machines (interactive terminal sessions), but were there file transfer protocols, or network based filesystems? Other features?
Hello!
If that's the case, then how would these machines communicate over
longer distances? For that matter, before Ethernet, and even the
current methods of fast connections we have now, would all of you
believe that we did use either dial-up on leased lines?
Mentioned in Cliff Stoll's book, "Cuckoo's Egg" is how his university
was connected to the Internet.
It depends on what timeframe, the IMP's had 9.6K and later 56K links
between them. There was a pararell and a serial host interface
named 1822, and later (some bozo) did X.25 as access host-imp.
So they either had a IMP at the University, or a 1822 serial link to
an IMP.
-Dept of useless knowledge.. -P
...and Peter don't need to "believe", since he was there and involved... :-)
I remember sitting at the Royal Institute of Technology in the beginning of the 80s, and connecting to MIT to run ITS on their machines. That involved an IMP, if I remember right. And it worked fine, although the speed was probably not impressive by todays standards. But back then you were used to 300 bps modems anyway, so this was no worse.
Johnny
On 2011-07-17 21.06, hvlems at zonnet.nl wrote:
Decnet 3 is compatible with phase 4. DEC was great with upward compatibility.
Yes. I was more thinking of how much a Phase III router could route in a Phsae IV network. I know I've seen the documentation for this somewhere, at some point. I just can't remember the details on exactly how compatible that detail was. Phase III endnodes worked well in Phase IV.
I thought ethernet (10 Mb/s, thickwire) ws from 1977, VMS is about the same age.
The Ethernet as we know and use it (really ethernet II) was published in 1982. The first version, which looked pretty much the same, but which allocated other protocol numbers (before low numbers became reserved for 802.n) was published in 1980. DECnet phase IV came around somewhere around 1982 or 1983 unless my memory fails me. (After Ethernet II, but pretty soon after, I believe). It might have been something like 1984 as well, though.
Phase III was pretty new, if it had even been introduced by 1977.
(Ah, looking at wikipedia, Phase III didn't arrive until 1980, while phase IV was indeed 1982).
Johnny
Verzonden vanaf mijn draadloze BlackBerry -toestel
-----Original Message-----
From: Johnny Billquist<bqt at softjar.se>
Sender: owner-hecnet at Update.UU.SE
Date: Sun, 17 Jul 2011 20:48:42
To:<hecnet at Update.UU.SE>
Reply-To: hecnet at Update.UU.SESubject: Re: [HECnet] DECnet et al
On 2011-07-17 20.35, hvlems at zonnet.nl wrote:
DECnet phase 4 is about as old as ethernet and VMS. Ethernet was the main driving force for phase 4 owing to the large number of adjacent nodes and the sheer number.
Not really. Phase IV is about as old as ethernet, yes. VMS is older...
Phase 3 does not have areas and only recognizes 255 hosts max. It did have circuit routing (of course)
Yes. About the same thing as L1 routing. I think it might even be
partially compatible, within the restrictions on the number of nodes...
Johnny
Verzonden vanaf mijn draadloze BlackBerry -toestel
-----Original Message-----
From: Gregg Levine<gregg.drwho8 at gmail.com>
Sender: owner-hecnet at Update.UU.SE
Date: Sun, 17 Jul 2011 14:26:07
To:<hecnet at update.uu.se>
Reply-To: hecnet at Update.UU.SESubject: Re: [HECnet] DECnet et al
On Sun, Jul 17, 2011 at 2:24 PM, Bob Armstrong<bob at jfcl.com> wrote:
A bridge (hw) maintains a list of mac addresses it sees at each port.
FWIW, that's a "switch" at least in the common usage of the word over
here.
Bob
Hello!
I agree! Now first things first. When did DECNET first appear? And
more importantly, based on what systems and using what means to
connect each system.
-----
Gregg C Levine gregg.drwho8 at gmail.com
"This signature fought the Time Wars, time and again."
ANF10!
DISCO DUCK
DISCO FEVER
ANF10 is a fond memory... Do I remember right that it was based on PDP-8
machines acting as connection points? Did they talk serial lines between
each other?
=A0 =A0 =A0 =A0Johnny
The PDP8 thing is a DN92, it has a sync interface and a bunch of
terminal lines and a lineprinter interface. It was supposed to talk to
a DS10, sync interface on PDP10 IO bus.
ANF10 is talked by the TOPS10 hosts over DTE (to FE PDP11 on KL), DL10
(DEC10 DMA/IObus to Unibus) and NIA-Ethernet (we got a ether_type for
it..)
There is ANF10 code for PDP11 that run on 40's or 34's.
The PDP11 code can talk DDCMP over ASYNC or SYNC interfaces, DQ11,
DMR11, DMC11. So you can do Wan things. In Sweden we had DEC10's in
Stockholm and Linkoping networked with a 9.6K sync link.
The DEC2020 talks DDCP with a DUP11 and a KMC (I think it was KMC,
unibus card with microdode that basically did DDCMP.)
Anf-10 has link-state routing like OSPF/ISIS but only one area.
THere was code to do ANF10 over X.25, but it was to expensive to use
on a public x.25 netowork...
Hello!
If that's the case, then how would these machines communicate over
longer distances? For that matter, before Ethernet, and even the
current methods of fast connections we have now, would all of you
believe that we did use either dial-up on leased lines?
Mentioned in Cliff Stoll's book, "Cuckoo's Egg" is how his university
was connected to the Internet.
It depends on what timeframe, the IMP's had 9.6K and later 56K links
between them. There was a pararell and a serial host interface
named 1822, and later (some bozo) did X.25 as access host-imp.
So they either had a IMP at the University, or a 1822 serial link to
an IMP.
-Dept of useless knowledge.. -P
