Poison vs Venom vs Toxin

Arcanaville · 09-16-2011 01:28 PM

Quote:

Originally Posted by GuyPerfect

I'm using the OSI definition, which funnily enough happens to say something to the effect of "the Network layer of the network model is about how networks communicate." You may not encounter many people who identify that each interface on a router connects to a network or subnetwork, but then again, does anyone really know what that machine is called that Zamboni makes? (It's an "ice resurfacer," should anyone feel like being clever)

The OSI definition of "network layer" doesn't definitively define "network" to being exclusively the domains between layer 3. Its important to note that OSI was created as a government standard to standardize networking terminology at a time when such language standards didn't exist, and it is necessarily fudged in many ways. Nobody uses it except in academic settings very much.

Quote:

In any case, "network" used to describe chunks of a system separated by routing is quite common among those who design and maintain those systems, and the students I was observing were being fed quite a contradictory definition. To me, it's equivalent to saying food comes in tin cans, when they're in fact usually steel. It makes a significant difference should someone need to decide what to make a can out of at some point.

I'm saying as someone who works in the industry, and explicitly instructs in the industry, that terminology is not adhered to in any appreciable manner except in pockets of jargon. You are far more likely to find people correcting you on it than agreeing with you in it. Specifically because unless your networking experience is restricted to nothing but ethernet, the terminology is blatantly wrong.

You can point to the OSI chart and a book definition somewhere and say "a network is all the systems within a broadcast domain" and I'll pull out my type 3 stuff from the box and say "what's a broadcast domain?"

The definition you are using above "networks are things separated by routing" uses the word "network" in a very specific context: the things requiring layer 3 identifiers. More specifically, they are layer 2 networks. However, layer 2 networks are not the only networks that exist. There are layer 3 networks which cross layer 2 boundaries. And I've never heard of anyone, anywhere, in any professional capacity, not use the term "layer 2 network" in some form as disambiguation, to distinguish from layer 3 networks, rather than assuming all networks were layer 2. You're actually the first in 25 years.

Quote:

It's generally a bridge that acts as a NAT switch with a LAN switch on the other end. But it's definitely not a router; you'll never find a routing table inside one of those unless it actually has multiple WAN interfaces. That's the point I was trying to make.

Not true. I have a cable modem that does NAT, and it has a routing table. Anything that forwards traffic between two different network segments which have different IP networks in use is going to have a routing table of some kind.

There's no such thing as a "NAT switch." NAT implies routing in essentially all cases. That is *definitely* vendor-specific jargon, and far enough outside the norm that I would authoritatively correct anyone using it.

Quote:

If your office complex funnels all of its traffic through a drop run by a local telco to get to the internet, you're not an autonomous system (which is registered with the same people who do IP addresses). The people you're paying for internet access are the people whose autonomous system you're plugging into, even if they don't manage any of your equipment.

If you don't have a unique ASN, as this situation would, you're not an autonomous system. If you do, you are.

[Guide to Defense] [Scrapper Secondaries Comparison] [Archetype Popularity Analysis]

In one little corner of the universe, there's nothing more irritating than a misfile...
(Please support the best webcomic about a cosmic universal realignment by impaired angelic interference resulting in identity crisis angst. Or I release the pigmy water thieves.)

Arcanaville · 09-16-2011 02:25 PM

I should take a step back and mention that in large part, quibbles of this nature are semantic in nature, just as the poison vs venom vs toxin one is largely semantic. Which is to say, calling something by the incorrect term doesn't change the thing and is unlikely to change any aspect of how people treat the thing. The terms provide more contextual information for people familiar with them and use them properly, but most such people also know the average person doesn't have that same command of the technical terms and wouldn't be confused by a lay person calling something the wrong thing. That's why we have poison hotlines and poison control centers, not poison, venom, and toxin hotlines.

However, sometimes use of the wrong term doesn't just blur distinctions, it hints at exposing actual misunderstandings and thats when its important to correct the issue because the problem goes beyond the semantics. For example, the definition of "network" as "data-link broadcast domain" potentially exposes a misunderstanding that can cause problems, which is why I don't let that one slide. Many layer 2 network topologies don't *have* the notion of a broadcast domain which makes that definition non-operative. That's why network *hasn't* been defined that way by professionals.

If you only work with ethernet, you can get away with that definition because the true definition of layer 2 network for ethernet is basically the same. But when you cross over into other topologies, like say ATM, if you actually *believe* your definition of network, you'll be lost: ATM doesn't have the notion of "broadcast" and thus has no "broadcast domains." ATM is a point to point layer 2 topology. Because ATM doesn't have implicit support for broadcast or multicast, there are RFCs that describe how to *simulate* those services in ATM: see RFC 2226 for example.

That's why this is not just a semantic argument. Anyone working in networking that believes a network is "a collection of stuff in a broadcast domain" might just be quirky, or they might be one of those people who believes all networks work like ethernet. The former is not a problem but the latter is. And that's what makes the incorrect definition potentially dangerous.

The same thing is true for whether something is or is not a "router." There are grey areas, but *how* we troubleshoot networks requires we understand where the layer 2 boundaries are, where the layer 3 systems are and which troubleshooting steps work at different layers with different topologies. I don't care what people call things for discussion purposes too much, but in a professional setting that changes: I *demand* people call things the correct things or I presume they don't know what they are doing. If they claim the distinction is semantic in nature only, I warn them that if they make a mistake traceable to a misunderstanding which the incorrect terminology exposes, I will act accordingly.

What we call things really only matters if it breeds confusion, or it hints at a misunderstanding that will come back to haunt us. In this case, for any actual professional I was working in conjunction with, I would consider both to be potentially true until the matter was resolved.

[Guide to Defense] [Scrapper Secondaries Comparison] [Archetype Popularity Analysis]

In one little corner of the universe, there's nothing more irritating than a misfile...
(Please support the best webcomic about a cosmic universal realignment by impaired angelic interference resulting in identity crisis angst. Or I release the pigmy water thieves.)

seebs · 09-16-2011 04:08 PM

Which is to say:

As soon as the game allows us to eat our opponents, and eating some of them damages us, it will become important to distinguish between "poison" and "venom" in the game.