Comcast: Simulating shitty network connections so you can build better systems

by ot • 11 years ago

330 80 11 years ago

knd775• 11 years ago

Upvote for the name.

ejdyksen• 11 years ago

If you're on Mac OS, you can also install the Network Link Conditioner from the Apple dev tools, which is a prefpane that serves the same purpose:

http://nshipster.com/network-link-conditioner/

athenot• 11 years ago

And on the command line, you can use the dummynet(4) command (can be useful to automate testing under various network conditions).

cordite• 11 years ago

Would something like this be DMCA'd or taken down because of the name? (Not that it is violating copyright by its contents)

bobmagoo• 11 years ago

I had that thought as well. From poking around various IP law sites, DMCA is purely for copyright infringement, and can possibly open you up to more liability if you try to use it for trademark claims (what this would be), see http://www.lexology.com/library/detail.aspx?g=13f9814f-b56e-... for an example.

If Comcast wanted to take this down, it would be through a trademark infringement claim.

(IANAL, etc, etc)

Florin_Andrei2• 11 years ago

So... the project is infringing on Comcast's ownership of shitty networks?

dublinben• 11 years ago

The DMCA doesn't apply to trademarks. This probably wouldn't violate Comcast's trademark anyway, as they operate in different markets.

MatthewWilkes• 11 years ago

Networking and infuriating call centres?

ot2• 11 years ago

I wonder if the author can invoke the freedom of satire.

fooqux• 11 years ago

Technically, no. However, it wouldn't be the first (or last) time that a DMCA takedown was issued for the wrong reasons. I can't remember specifics, but I seem to recall DMCAs issued for name conflicts before.

troymc• 11 years ago

I was confused if this was something from the real Comcast, so it seems to be a clear case of trademark infringement (but it's not my call, and I may be wrong).

If GitHub agrees, then it would be a violation of their Terms of Service [1], section A.8:

"You may not use the Service for any illegal or unauthorized purpose. You must not, in the use of the Service, violate any laws in your jurisdiction (including but not limited to copyright or trademark laws)."

[1] https://help.github.com/articles/github-terms-of-service/

kozhevnikov• 11 years ago

There is also clumsy with what possibly is the best demo gif ever made.

https://jagt.github.io/clumsy/

digi_owl• 11 years ago

Tamper mode was downright hilarious.

14113• 11 years ago

Regardless of how good the tool is - that was an excellent demonstration clip! Within ~30 seconds I knew exactly what the tool did, why it was cool, and what I might want to use it for.

Startup landing pages could learn a lot from that single gif.

azdle• 11 years ago

I'm curious if anyone knows of tools like this that are more deterministic?

I'm in the process of writing a library at work that is basically implementing the CoAP protocol from the ground up and having something that I can script to force the same packets to be lost or delayed is something that I would find very useful for testing. I've been using Apple's network link conditioner, but when I do discover a problem, it can take a long time for the same scenario to happen again which makes testing the fix quite difficult.

I'm currently trying to think through how I'd write this to make the filters easy to setup [0], but if anyone knows of something that already exists, please let me know.

[0] https://github.com/Azdle/dolos

blueskin_• 11 years ago

On linux, tc can do similar things, such as dropping a percentage of packets or adding random delays.

http://linux.die.net/man/8/tc

lucb1e• 11 years ago

As said in the readme:

you can use tc which supports some additional options.

     $ tc qdisc add dev eth0 root netem delay 50ms 20ms distribution normal
     $ tc qdisc change dev eth0 root netem reorder 0.02 duplicate 0.05 corrupt 0.01

> To reset:

    $ tc qdisc del dev eth0 root netem

Florin_Andrei2• 11 years ago

Yeah, I used to do stuff like this with tc directly back in the day, but I could never remember the options. This wrapper seems to make it a bit more obvious what goes where.

scott_karana2• 11 years ago

From TFA :)

On Linux, we use iptables and tc. Comcast is merely a thin wrapper around these controls.

poor_too_• 11 years ago

yep, https://github.com/tylertreat/Comcast/blob/master/throttler/...

api7• 11 years ago

Nice.

It'd be neat to include options to simulate other pathological conditions often encountered in the field like:

- Multiple layers of NAT - overlapping un-synchronized timeouts are one thing ... but the REAL fun comes in when intermediate layers have the same IP ranges as IPs you are trying to reach on the "outside" of the NAT sandwich. All kinds of "interesting" things can happen, like the "software laser": http://catb.org/jargon/html/S/software-laser.html

- Stateful NATs/firewalls with bizarrely short connection / UDP association timeouts or that randomly forget connection state

- Shitty NATs/routers that go into bizarre failure modes when there are too many open connections, like forgetting early ones because they remember connections via a small ring buffer

- People who block all ICMP "because security."

- NAT + short DHCP leases = musical external IP addresses

- Small MTUs (<1500) like those imposed by PPPoE and other nasty encapsulation protocols, sometimes without ICMP errors for big packets because some jackhole blocked ICMP "because security."

All these things are common in the field. WHYWHYWHYWHYWHY

Spooky233• 11 years ago

My favorite enterprise network pathologies:

- Randomly block specific ports "because security"

- Limit RPC port ranges to some random small number.

- Strange network optimizations. Example: Optimize TCP Window Size to support NT4 clients on 56k frame relay circuits.

- Very Slow DNS response

- Optimize core switching rules to fully utilize switch CPU. Avoid configurations that take place in an asic.

- Long DHCP Leases + Undersized DHCP scopes.

api7• 11 years ago

I also forgot TCP through TCP tunnels, which causes strange timing and flow control behaviors due to never actually dropping packets on congestion and double-ACKing.

wallflower• 11 years ago

It makes me wonder at how Skype manages to tunnel though almost anything.

akshatpradhan• 11 years ago

What do you mean randomly block specific ports because "security"? All ports should be denied access except ports which have a justified business reason. Got a web app? The only thing open should be 80/443. There's no reason for SMTP to be open on the web server. Anything doing mail should be on its own MTA server. Least functionality per server. That's not even security. That's just good system administration.

steventhedev• 11 years ago

Don't forget jitter, and simulating an ISP that will shut off your DNS traffic at random (literally anything DNS, not just their own servers).

api7• 11 years ago

I forgot about DNS munging shenanigans. Not only do some people selectively block DNS, but there are also cases where middle boxes and proxies pretend to be remote DNS servers but really aren't... or rewrite DNS traffic, insert fields, etc.

angersock2• 11 years ago

I take it you've seen core switches on fire off the Jupiter gate?

Freaky• 11 years ago

Juniper gate.

cdelsolar• 11 years ago

Tannhauser gate

superuser2• 11 years ago

WHYWHYWHYWHYWHY

Are there any network administrators who don't consider NAT traversal to be a security breach? Making it difficult would seem to be a feature, not a bug, in most enterprises.

api7• 11 years ago

NAT traversal is a security breach that became a de-facto standard and is part of the SIP, STUN, and IPSec (extended) standards among other RFCs. It exists because NAT itself is an abomination.

chatmasta• 11 years ago

People talk about "fuck you money," meaning the amount of money you need to make before you can never work again. Personally, my "fuck you money" isn't money per se, but the fall of Comcast. When I build enough value in the world to displace Comcast, I will retire.

We need to see more work in the areas of mesh networking, layer-3 routing, and consumer networking in general. This tool is a good step. Personally I'm hacking on some openwrt routers right now -- I recommend everyone try it. The documentation is dense, with a friendly community writing it.

Fuxy• 11 years ago

You know your company sucks when its name has become synonymous with high latency, dropped packets and a all round shitty network.

Unfortunately i suspect they will just send a trademark claim to Github to get it removed.

paulrosenzweig• 11 years ago

This seems like satire to me. Could that save it like "Dumb Starbucks"?

https://en.wikipedia.org/wiki/Dumb_Starbucks

shanemhansen• 11 years ago

I hate comcast customer service with a passion, and their pricing models are super hostile to existing customers.

But their internet service has been very reliable and very fast for several years for me. I rarely have internet outages.

btreecat• 11 years ago

You just reminded me that I have been dragging my feet upgrading my two OpenWRT routers for the last few weeks. I guess I know what I am spending my lunch break on!

yincrash• 11 years ago

Was this made in response for more real version of https://theonion.github.io/comcastifyjs/ ?

nwilkens• 11 years ago

SmartOS has this nice utility - ipdadm which does something similar.

ipdadm - administer the Internet packet disturber

https://smartos.org/man/ipdadm

spiritplumber• 11 years ago

I wonder if this can be used to simulate speed-of-light delay for a Mars mission.

exabrial• 11 years ago

In the words of the great poet Macklemore:

"This is fucking awesome"

bas• 11 years ago

This is a beautifully-named project.

xtrumanx• 11 years ago

Google Chrome's device mode has a subset of this functionality where you can make Chrome throttle the connection simulating different conditions (GPRS, DSL, offline, etc.).

Access device mode by clicking the smartphone icon on the top left of the Developer Tools (F12) right next to the Elements tab.

bduerst• 11 years ago

It's very useful. I discovered this the other day when testing some client side scripts. I was able to simulate a 3G mobile connection and determine which parts were causing the intolerable load times.

SEJeff• 11 years ago

You really shouldn't do something like this without at the very least mentioning Kyle Kingsbery and the Jepesen test suite he wrote to validate how distributed datastores actually follow CAP or not.

For any devops / sysadmin / systems engineer: I highly suggest reading this if not only to understand failure conditions better:

https://aphyr.com/tags/jepsen

The entire "Call Me Maybe" blog series is because that is Carly Rae Jepsen's stupid pop song, for which he named his test suite after. That and Kyle's posts are absolutely hilarious to read. Seriously, read them.

mpu• 11 years ago

Written in Go for nothing, you might as well hack a cheap shell script.

scott_karana2• 11 years ago

Who cares? It also could have been written in Perl, csh, C, APL, Pascal, or Fortran, and it would work exactly the same.

Can we stop shitting on cool things just because the technology being used is "trendy", or you're sick of hearing about it?

nexneo• 11 years ago

`--device=eth0 --latency=250 --bandwidth=1000 --packet-loss=0.1`

This could have been named `--i-am-in-india`

GrinningFool• 11 years ago

Nice, and the detailed README is appreciated.

I do have to wonder if the name won't cause you trouble though...

copsarebastards• 11 years ago

I don't need to simulate this, I have actual Comcast service.

farresito2• 11 years ago

I have a feeling that we will see an incoming DMCA soon.

farresito2• 11 years ago

Getting downvoted. The lack of humor in Hacker News is astonishing.

jpetersonmn• 11 years ago

Id change the name, not because of Comcast coming after you, but I guess it's just not very clever IMO. Good luck on the project though. :)

jvehent• 11 years ago

I don't need this, I have comcast.

    $ sudo ping -i 0.02 8.8.8.8
    [...]
    64 bytes from 8.8.8.8: icmp_seq=578 ttl=53 time=15.5 ms
    ^C
    --- 8.8.8.8 ping statistics ---
    579 packets transmitted, 364 received, 37% packet loss, time 13854ms
    rtt min/avg/max/mdev = 15.009/19.308/50.884/6.152 ms, pipe 2

YuriNiyazov• 11 years ago

It's so weird to hear so much complaining about Comcast here. My stats with that same command:

     --- 8.8.8.8 ping statistics ---
     615 packets transmitted, 612 packets received, 0.5% packet loss
     round-trip min/avg/max/stddev = 21.332/35.875/95.576/11.248 ms

I have never seen it be anything different.

sarciszewski• 11 years ago

I have never seen it be anything different.

If you parse this statement literally, it has amusing consequences.

FireBeyond• 11 years ago

Comcast Business connection in Olympia WA:

``` --- 8.8.8.8 ping statistics --- 538 packets transmitted, 530 packets received, 1.5% packet loss round-trip min/avg/max/stddev = 13.706/27.511/101.208/11.207 ms

mchristoff• 11 years ago

This is great. I would love mobile support.

alexcasalboni• 11 years ago

Very nice.

How's this better/different than Crapify?

https://www.npmjs.com/package/crapify

mml• 11 years ago

Did this with dummynet on FreeBSD around 1999. Made it a bridge though, so everyone on the test network could "enjoy" 28.8kbaud dialup experience! Great fun.

Called it "the molasses network", which is what they should change the name to before Comcast smokes them.

pc2g4d• 11 years ago

Prediction: name of this project will be different within 12 months due to trademark infringement.

dangerboysteve• 11 years ago

The project name is pure gold!

Comcast: Simulating shitty network connections so you can build better systems

Related Stories