Personal computing discussed

Moderators: renee, Steel, notfred

 
Prototyped
Gerbil
Posts: 58
Joined: Wed Dec 26, 2001 7:00 pm
Location: Purgatory

Fri Mar 08, 2002 6:30 pm

Bruce's right, just having the least significant octet of an IPv4 address doesn't make it a network address. It's a perfectly valid host address - just that some older equipment has broken firmware that assumes that anything ending with a .0 is a network address and anything ending with a .255 is a broadcast address.

For one thing, on a network like, say, 192.168.252.208/28 network, the broadcast address is 192.168.252.223 and the network address is 192.168.252.208.
On a network like, say, 10.0.0.0/8, 10.0.0.255 is not a broadcast address; it's just another IP address. Also, a non-broken node on the network can have its IP address set to 10.0.0.0 and will work just fine, as long as other equipment it communicates with isn't broken. Similarly, the router doesn't have to be 10.0.0.1; it could be 10.0.0.172 for all anyone cares.

It sounds like people here haven't gotten used to CIDR/classless networks, and are sticking to /8, /16 and /24 networks, which CIDR supercedes.

(Credentials matter? If they do, I maintain and optimize network stack software for routers.)

<hr />

Squirrel,

169.254.0.0/16 is the APIPA netblock, which Windows 98 and higher systems use IP addresses from if they fail to get one via DHCP, so they can interoperate over an IPv4-based LAN even if they cannot get a DHCP lease.



<font size=-1>[ This Message was edited by: Prototyped on 2002-03-08 17:32 ]</font>
 
Speed
Gerbil Elite
Posts: 702
Joined: Thu Dec 27, 2001 7:00 pm
Location: Chicago, IL USA
Contact:

Fri Mar 08, 2002 6:52 pm

Hokey smoke! Since when does old == broken?
You are false data.
 
Prototyped
Gerbil
Posts: 58
Joined: Wed Dec 26, 2001 7:00 pm
Location: Purgatory

Fri Mar 08, 2002 8:33 pm

Since some older equipment is broken. :wink:
 
SecretSquirrel
Minister of Gerbil Affairs
Posts: 2726
Joined: Tue Jan 01, 2002 7:00 pm
Location: North DFW suburb...
Contact:

Sat Mar 09, 2002 12:40 am


Squirrel,

169.254.0.0/16 is the APIPA netblock, which Windows 98 and higher systems use IP addresses from if they fail to get one via DHCP, so they can interoperate over an IPv4-based LAN even if they cannot get a DHCP lease.



<font size=-1>[ This Message was edited by: Prototyped on 2002-03-08 17:32 ]</font>


Didn't know that.

As far as the classless networks, that's the way I always assume things work now. The only thing that defines what a network address and broadcast address is, is the router controlling the segment and the machines attached to the segment. The outside world can't know.
 
Bruce
Gerbil Elite
Posts: 500
Joined: Thu Dec 27, 2001 7:00 pm
Location: New Zealand

Mon Mar 11, 2002 5:20 am

I'm still trawling through RFC's and text books, but I've come to the same conclusion Prototyped. Speed, I think CIDR/VLSM has obsoleted what you are talking about (esp RFC 1878). I've found a good quote in Halabi's BGP book which seems to be on the right track.

Still, I haven't found a single, definitive mention in the RFC's one way or the other - it may be there, I just haven't found it.

Bruce
 
Speed
Gerbil Elite
Posts: 702
Joined: Thu Dec 27, 2001 7:00 pm
Location: Chicago, IL USA
Contact:

Mon Mar 11, 2002 6:17 am

Bruce, I keep saying that I'm not interested in a pissing contest. If you want to be the authority, then I bow to you.

I will mention that CIDR really has nothing to do with this argument. I never thought I'd say this to someone, but you're looking at the matter through a router person's blinders, and ignoring the big picture. It matters not if you can make a router rule for a number ending in zero. Try configuring a host that way. Try using IP utilities to interact with a zero address. It doesn't work. Sure, you can blame computers and operating systems. But without hosts, routers would be purposeless.

IMHO it's erroneous to assume CIDR when no netmask is given, as was the case here. Since classless IP relies on a specific netmask, rather than the one that can be derived from the number range. So getting back to the original question and my answer, I don't believe there is anything wrong with my conclusion, based on the information presented. I have always maintained that I'm not here to teach a comprehensive course in IP, so I see nothing wrong with a succinct answer.

I do understand your POV, and respect your knowledge of the subject. I hope that you come to understand the realities that systems engineers face every day.
 
Despite
Gerbil XP
Posts: 496
Joined: Thu Dec 27, 2001 7:00 pm
Location: Oklahoma
Contact:

Mon Mar 11, 2002 8:18 am

That's enough of this nonsense. When I get a chance today, I am going to configure 2 hosts with IP addresses that would be considered messed up by Speed's standards. If they can't communicate perfectly, then I will declare Speed "teh winnar". However, if they can communicate okay, I'm going to set up the biggest virtual wet trout bludgeoning this forum has ever seen!
 
Bruce
Gerbil Elite
Posts: 500
Joined: Thu Dec 27, 2001 7:00 pm
Location: New Zealand

Mon Mar 11, 2002 8:47 am

Oh for goodness sake Speed, I'm trying to be gracious and humble and understand what's going on, and why we have a difference of perspective. It's not an "I'm right" argument or any such thing - I treat this forum as an opportunity to learn from others and to share my knowledge with others. In this particular case I think there's a bit of both going on.

Your almost certainly correct that I'm thinking of this from a router person's POV. I tend to think that this is a big picture - you can do (almost) what you like with a network of hosts, but it's when you try to interconnect them that the fun begins.

As for realities of systems engineers: I work as a systems administrator. Similar, but subtley different mindsets.

Despite: which OS/OS's are you using? I'll do a similar test myself at home tonight, probably using FreeBSD.
 
Despite
Gerbil XP
Posts: 496
Joined: Thu Dec 27, 2001 7:00 pm
Location: Oklahoma
Contact:

Mon Mar 11, 2002 9:06 am

well, I just tested with Win2k and linux 2.4 and what did I find? can you guess? I knew that you could.


Squire, bring me my trout!
 
Bruce
Gerbil Elite
Posts: 500
Joined: Thu Dec 27, 2001 7:00 pm
Location: New Zealand

Mon Mar 11, 2002 9:21 am

LOL, well put. I'll post the results for OpenBSD 2.9 and FreeBSD 4.4-RELEASE tomorrow.
 
Speed
Gerbil Elite
Posts: 702
Joined: Thu Dec 27, 2001 7:00 pm
Location: Chicago, IL USA
Contact:

Mon Mar 11, 2002 10:31 am

Well Despite, since I too have tried it on Linux 2.4 and Windows 2000, I know what you found. Couldn't ping nuthin.
You are false data.
 
Speed
Gerbil Elite
Posts: 702
Joined: Thu Dec 27, 2001 7:00 pm
Location: Chicago, IL USA
Contact:

Mon Mar 11, 2002 10:46 am

Bruce, if you look back you will see that I was just trying to be helpful and answer a question. I didn't make you post a hypercritical reply. I didn't fault you for your straw man argument and inconsistencies. And I have stated ad nauseaum that I'm not interested in making this into a competition. You and Despite are so obsessed with winning the argument that you created, that you're the lone smudge on an otherwise excellent forum. Grow up!
You are false data.
 
Despite
Gerbil XP
Posts: 496
Joined: Thu Dec 27, 2001 7:00 pm
Location: Oklahoma
Contact:

Mon Mar 11, 2002 10:54 am

I'm trying really hard here not to blow up and start using inappropriate language, Speed. Your inability to concede on this issue is going to require everybody here to try this experiment for themselves. But that's cool; once they do, they'll know who the jackass is. On the chance that you really did try this experiment and got no communication, may I suggest you go back and make sure you entered all the configuration information correctly?
 
Bruce
Gerbil Elite
Posts: 500
Joined: Thu Dec 27, 2001 7:00 pm
Location: New Zealand

Tue Mar 12, 2002 1:46 am

I had intended to write a long, detailed, reply. I haven't got time, so here's a summary.

Again, my apologies for coming across harsh. I am simply frustrated by your unwillingness to cite a reference for your point of view. You say you haven't got time to write a tutorial, but you have plenty of time to criticise and insult people. People can take from that what they will.

I've been usng IP every day of my working life for the last 6 years. It bothers me that I may have misunderstood something as simple as this. That's the only reason I'm persuing this - nothing to do with a "pissing contest", just a desire to understand.

For the benefit of the people following this thread (you poor, patient sods!) here's what I've found. None of the following documents support what Speed is saying:



Darn, I'm running out of time. I'm not saying that it doesn't exist in RFC 900-something, I just couldn't find it. Other books I've checked are Basam Halabi's "Internet Routing Architectures 2nd Edition", Stevens "TCP/IP Illustrated, Volume 1", Wendell Odom's "Cisco CCNA Exam Certification Guide" and Jeff Doyle's "Routing TCP/IP, Volume 1"

The best explanation I've found is in RFC 1812, section 4.2.2. There, F Baker of Cisco Systems uses the terminology {Network Prefix, -1} to describe a directed broadcast. Ie, all bits of the host portion of the address must be '1' for a directed broadcast. If the host portion of your addressing scheme uses more than 8 bits, then having a '0' or '255' as the last octet does not mean anything - all the host bits must be '1' for a directed broadcast, and all the host bits '0' to denote the network.

Lastly, my trials last night with OpenBSD 2.9 and FreeBSD 4.4-RELEASE agreed with all the above: using a combination of any 2 of 172.24.19.0/16, 172.24.19.255/16, 172.24.20.0/16 and 172.24.20.255 worked as expected - these are all valid IP addresses for use on a private network, and all have the same network address (172.24.0.0) and the same brodcast address (172.24.255.255). The reason for using BSD's is that I believe they have the most mature TCP/IP implementation available today.

OK, I'm done. Hopefully you'll see I spent a lot of time trying to work out what's going on. Again, my apologies if I sound like I'm pissing - I'm simply trying to understand something.

I haven't proof-read the above - will do so at work. I hope there aren't too many clangers in here!

Bruce

Edit: me fixing tags

<font size=-1>[ This Message was edited by: Bruce on 2002-03-12 03:51 ]</font>
 
Speed
Gerbil Elite
Posts: 702
Joined: Thu Dec 27, 2001 7:00 pm
Location: Chicago, IL USA
Contact:

Tue Mar 12, 2002 7:56 pm

On 2002-03-12 00:46, Bruce wrote:
Again, my apologies for coming across harsh. I am simply frustrated by your unwillingness to cite a reference for your point of view.

Bruce, I was about to write my own apology to you. I'm sorry for being snappy. I didn't mean to be evasive, but I did have an important task that needed my full attention. My reasons for posting anything at all were more technical than personal at that point. You see, I was upgrading the webserver software here at TR, and needed to make sure nothing broke after each change. But I had no time to devote to an exhaustive search of a LOT of RFCs. That's not an excuse for my behavior, for which I apologize.

The best explanation I've found is in RFC 1812, section 4.2.2. There, F Baker of Cisco Systems uses the terminology {Network Prefix, -1} to describe a directed broadcast. Ie, all bits of the host portion of the address must be '1' for a directed broadcast. If the host portion of your addressing scheme uses more than 8 bits, then having a '0' or '255' as the last octet does not mean anything - all the host bits must be '1' for a directed broadcast, and all the host bits '0' to denote the network.

Yes, that was my point. Thanks for the information. I admit that I hadn't been thinking of CIDR when I first made the point, but if you substitute "host bits" for "last octet" the logic survives IMHO. If the IP number is identical to the network field, then it is the network, no? I probably could have found a better way of saying it.

So let's put this behind us. Friends, OK?


_________________
Stick a fork() in it!

<font size=-1>[ This Message was edited by: Speed on 2002-03-12 18:57 ]</font>
 
Bruce
Gerbil Elite
Posts: 500
Joined: Thu Dec 27, 2001 7:00 pm
Location: New Zealand

Wed Mar 13, 2002 4:24 am

But I had no time to devote to an exhaustive search of a LOT of RFCs. That's not an excuse for my behavior, for which I apologize.


Apology accepted!

So let's put this behind us. Friends, OK?


Absolutely - there's too many intelligent, knowledgable people using this forum to go around making enemies. Looks like our misunderstanding was primarily terminlogy - not unusual in the networking field.

If the IP number is identical to the network field, then it is the network, no?


If your saying what I think your saying - yes.

Bruce
 
Speed
Gerbil Elite
Posts: 702
Joined: Thu Dec 27, 2001 7:00 pm
Location: Chicago, IL USA
Contact:

Wed Mar 13, 2002 4:44 am

Thanks Bruce!
You are false data.

Who is online

Users browsing this forum: No registered users and 1 guest
GZIP: On