Help me show him the truth

August 21, 2008 12:15 PM

Help me convince my friend of a basic networking truth. He sent me an e-mail saying he was assigning static IPs to all the devices on his home network so the signal would be stronger. More details inside:

My friend has the following set-up: His cable modem goes to his router. His router feeds wall jacks throughout his house. In his basement is a single CAT-5 wall jack, but since he has several devices there he got a switch, so the wall jack goes to the switch, which then goes to his DirectTV and his XBox 360.

This worked fine in this set-up for months. But a couple weeks ago he stopped being able to receive On Demand programming. If he plugs the DirectTV right into the wall jack it works, but through the switch it does not.

He then assigned his DirectTV and his XBox 360 static IPs and the On Demand programming worked through the switch.

He believes the switch is causing weaker signal and the static IP is strengthening the signal to his DirectTV which, as an IT network tech, I know is NOT the case. And I've told him this but I cannot convince him that it's not a signal issue since he has seen the "evidence" that a static IP fixed the On Demand issue that a static IP didn't.

So (a) why would his DirectTV not work through a switch with a dynamic IP and (b) How can I explain in complete laymen's terms the fundamental networking essentials concepts of digital signal transmission through shielded twisted pair cable is NOT effected by the IP addresses being static or dynamic?

posted by arniec to computers & internet (21 answers total)

The setup working with static IPs where it didn't with DHCP points not to a signal strength problem, but a DHCP problem. Check to make sure the DirectTV device can acquire an IP reliably from his DHCP server. It is likely a problem with either the DHCP configuration or the switch itself. Is the switch a cheap one?

Another possibility I've seen is that some switches and network cards/interfaces will only work properly when autonegotiating a link. If the device is hard-set to 100FD then there can be problems (depending on the switch and/or network card).
posted by rhizome at 12:24 PM on August 21, 2008

But would a static IP change the link's autonegotiation? Isn't the IP just assigned by the router but the link still negotiated or hard-set no matter?
posted by arniec at 12:26 PM on August 21, 2008

The switch could be going bad. He could have a bad cable somewhere in there. I thought I had a bad switch recently until I tested a bazillion cables and found the one causing the headaches.

His IP theory works because he probably rebooted the switch or moved the cables some during this process.

Try a different switch. Try different cables.
posted by damn dirty ape at 12:26 PM on August 21, 2008

>Isn't the IP just assigned by the router but the link still negotiated or hard-set no matter?

Yeah but if its a cheap SOHO switch, then just one bad port or one bad cable can screw up the entire device.
posted by damn dirty ape at 12:28 PM on August 21, 2008

The static IP's VS DHCP only pertains to the assignment of addressing on the network, they are not at all related to the transmission characteristics of the cable infrastructure.

It is likely that there is an issue with the device providing DHCP to the network or possibly a problem with the end hosts not acquiring a news lease when their existing lease expires.
posted by iamabot at 12:34 PM on August 21, 2008

He could have a bad cable somewhere in there.

Swapping out the switch and successfully plugging DirectTV straight into the wall obviates this issue.

As for the autoneg, it's just a datapoint to check and configure on the DirectTV if possible.
posted by rhizome at 12:36 PM on August 21, 2008

in answer to a), a couple spring to mind.
1) DHCP timeouts. Since it's a long cable down to the basement, it's possible it's getting either interference or it's out of spec on quality or length, and he's getting packet loss. The
switch adds just sufficient further delay to stop DHCP requests or responses working. Some testing with pings and/or a cable tester may help diagnose this. Is it really shielded twisted pair? Most network cable is UTP.

2) DHCP timeout. His switch is just broken, and causing issues with broadcast packets, like DHCP. Swap the switch, see if it gets better.

3) his DHCP server is broken. Doesn't seem likely, as plugging the devices in sans switch works.

In answer to b), of course fixed IPs aren't 'stronger' than dynamic, but they do require a lot less broadcast packets going backwards and forwards. It sounds like he has a bigger network problem with the switch, and the fixed IPs is a workaround for this problem, but removing the need for that 'back and forth'. Tell him how DHCP works, compared to fixed IPs, in terms of the broadcast back and forth.

The OSI layers might help you; layer 1 is the physical and electrical side of things. It knows nothing about the layers above - as far as electric current is concerned, it doesn't know the difference between one set of binary signals and another set of binary signals.

IP is at layer 3, and TCP at layer 4. Ask if he thinks, when he sets his IP address as fixed, the electrical wiring in his network card cares whether he's using a web-browser or an email program (layer 7) with that IP address.
posted by ArkhanJG at 12:39 PM on August 21, 2008

But would a static IP change the link's autonegotiation? Isn't the IP just assigned by the router but the link still negotiated or hard-set no matter?

Auto negotiation is between the NIC and the switch port; it decides what common speeds that have in common over ethernet (layer 2) - i.e. 100MB half-duplex, 1000MB full-duplex etc.

This is entirely independent as to what is sent over that physical connection in layer 3 and 4 (TCP/IP or UDP) i.e. DHCP requests and all other TCP/IP packets between the router and device.
posted by ArkhanJG at 12:44 PM on August 21, 2008

Hasn't logic already failed to convince him? Why would different a logical explanation work any better?
posted by winston at 12:51 PM on August 21, 2008 [1 favorite]

Winston,

I tried to explain to him that the signal of the devices is what it is, dependent on the cables, interference, etc. and that the IP address does not strengthen the signal. He replied "If static IP addresses provide no benefit, why do businesses use them, it probably has something to do with this stronger signal." So I find myself frustrated...I can't seem to explain clearly (and perhaps to convince him I need citation) that the static IP did not make his connection stronger, faster, etc.

He also believes opening ports on his router will increase his connection strength to XBox Live because he read that he should do that on an XBox 360 web site. I'm trying to help him, and I am even a college instructor of networking, but my students usually BELIEVE me when I tell them something, and he's giving me guff thinking that I don't know my stuff here...so I'm turning to the "hive mind" for different ways to explain the OSI model and these basic networking concepts to him.

(Yeah, this post is more about proving me right than fixing his Direct TV, heh...but if we fix that along the way, so much the better)
posted by arniec at 12:57 PM on August 21, 2008

arniec, if it helps, that issue I mentioned where I thought I had a bad switch, but had a bad cable, worked fine until I tried getting DHCP or any broadcast stuff. I could assign static IPs to two machines and ping each other. His hardware is dying somewhere and the first symptom is broadcast packets timing out.

If you want to convince him, bring a known good switch and cable to his place and swap his stuff out.
posted by damn dirty ape at 1:00 PM on August 21, 2008

nthing OSI layers.

The radio signal is the Physical layer.

The IP setting is the Network layer.

Network layer settings do not affect physical layer characteristics. Case closed.

Ref:
Cisco's Internetworking docs
posted by kalessin at 1:10 PM on August 21, 2008

As we all know, he's loony-nuts when it comes to this "signal strength" issue.

But I have always preferred static IPs on my own (home) networks, where I control all the addresses. It's just so much easier to configure, not to mention manage security, when you can recognize (for example) your PS3 by its IP number, or forward certain ports to certain IPs via NAT.

So I'm just saying there's nothing WRONG with using static IPs, and it comes with benefits... but "signal strength" is unrelated craziness.
posted by rokusan at 1:27 PM on August 21, 2008

OK, by the sounds of it logic simply isn't working. I have to admit, if i was in your shoes I'd throw in the towel. You've explained to him why he's wrong, you can point him towards the OSI layer, you can pull out all the appeals to authority that you like - but he's convinced himself from the evidence that static IPs are 'stronger'.

electrical signals are on, or off. It doesn't matter what sequence of on or off they are in over time to the device making them go on or off. DHCP vs static IP is just a different sequence of electrical binary signals (effectively) down the cable, just as web traffic and email traffic are.
Ports open to XBOX live will simplify the routing, and arguably speed up the connection (slightly), but it won't make the electrical signals in the wire have a higher voltage!

If that's not enough to convince him, then to be honest I'd give up. Some people are so convinced of their theory that no facts, evidence or real theory by guys who've worked in the field for years will convince him. He's already demonstrated he's lacking in pretty basic physics, and he's not prepared to accept the word of experts, so I'm pretty sure you're on a losing ticket here. some users are just happier in their own ignorance.
posted by ArkhanJG at 1:32 PM on August 21, 2008

I'd tell him that not only do static ip address produce stronger signals, a static ip address with an octet over 255 will cause a signal so strong that it could catch the device on fire. (That's why most networking devices don't allow it.)
posted by sexymofo at 1:38 PM on August 21, 2008 [8 favorites]

Arkhan, good point. Sexymofo and rokusan...thank you for making me chuckle for the last several minutes
posted by arniec at 1:40 PM on August 21, 2008

>>He could have a bad cable somewhere in there.

>Swapping out the switch and successfully plugging DirectTV straight into the wall obviates this issue.

That only rules out the switch<>DirecTV cable, not the wall<>switch cable. (or vice versa) Jus' sayin'..

I think kalessin's got it. Apples and Oranges.
posted by cowbellemoo at 1:43 PM on August 21, 2008

He has a chicken-and-egg problem in his theorem. If the amount of electricity on the wire is enough to trigger a "1", then those ones and zeros can send and receive anything, from web requests to mefi, to DHCP requests. If it's not enough for DHCP, it's not enough for web. Ever. There is no middle ground.

Get him to forget about IP altogether. Ask what he thinks would happen if he were using IPX or Lantastic or NetBEUI or Arcnet, something where there is no concept of Internet Protocol or the addresses I.P. uses. His switch is not an IP machine. It would work the same way with those other non-Internet protocols.

He may have a problem with his DHCP server, but it ain't what he thinks it is.
posted by cmiller at 3:54 PM on August 21, 2008

Well he's clearly loony tunes re: signal "strength", but if he arbitrarily equates "working" with being strong and "not working" with being weak, it's pretty easy to see how he gets there.

The only real way to prove your point to him is to get things working with dynamic IPs in such a way that he doesn't know you've done it so he can't come back with "Yeah it works, but it's slower." (despite that "slower" being wholly fabricated/imagined). Then spring the truth on him later after he's been using it for a while in a Folger's crystals way and you might have a chance. Probably not worth the effort though.
posted by juv3nal at 4:12 PM on August 21, 2008

You don't really need to convice him, OP. Facts just get in the way of his opinion, but to him, opinion FEELS like truth.

This scenario repeats so often in life that it makes more sense for you to develop coping strategies than it does to figure out ways to get facts into a closed vessel.

You really can't help someone who doesn't want it. If you command the science, and I mean grok it in its entirety, you might be able to present him with a quandary that his explanation won't explain and eventually, if he's got the mental strength to follow through, he'll discover his own errors and finally get it. He has to GET IT, though. You cannot GIVE IT.

Folks like this can be especially infuriating when they are called managers. Be glad he's not filling that role in your life!
posted by FauxScot at 5:22 PM on August 21, 2008

If you're really hell-bent on having him be logical (which, uh, is almost always a losing game), then have him list the criteria under which he'd be convinced. You're not going to guess what magic incantation will make him come to his senses. But if he provides you a list of "well, if this happens, and then you do this, and it responds that way, then I get it."

What you'll likely find, instead, is that there is nothing you can do to convince him, and he may even freely offer that fact up to you. If that's the case, and he's admitting he's unwilling to use logic to inform his opinions, then all you can do is let him enjoy his TV and focus your efforts on keeping him from having children or voting.
posted by anildash at 1:53 PM on August 22, 2008 [2 favorites]