Sign in with
Sign up | Sign in
Your question

Sort routers by family

Tags:
  • Routers
  • Linksys
  • Performance
Last response: in Toms Network
Share
April 20, 2006 2:21:07 PM

Hi,

I think router performance charts under heavy loads are very useful, but I (humbly) think you could improve your testing methodology:

Sort the routers by family:
As different brands use common components, could you make a reference for routers that use the same components, despite differences in functions, because performance should (theoretically) be comparable.
Example: I own a brand new Linksys WAG354G router, and I have no idea with wich router in your test I should compare it to.
1) If your test has a Linksys "cousin" of my router, I'd like to see some sort of mention like "comparable to Linksys WAG354G".
2) If your test had no Linksys "cousin" of my router, but suppose a D-Link was comparable to mine, it would be nice to see some sort of mention like "comparable to Linksys WAG354G".
I know this would take a lot of work, as gathering information about the processors of each router is certainly a painful job...

Stress test:
I know time is money, but stress tests are VERY important for users that put heavy loads on routers (which is your readers' target).
My personal experience with a Linksys WAG54G v2 was a decrease of performance over time, until the modem froze. That took around 36 hours in real life situation, which was a complete nightmare (I trashed the damn thing).
If Tom's lab could produce some sort of torture test that would artificially simulate as many connectionsas the router can handle, that would be nice. You could (for pratical reasons) even cap the test at X hours (example: over 5 hours, all routers that would survive would be given the torture test survival label ;-)

Cheers,

More about : sort routers family

April 20, 2006 7:08:08 PM

Quote:
As different brands use common components, could you make a reference for routers that use the same components, despite differences in functions, because performance should (theoretically) be comparable.

Good point. However, firmware can make an appreciable difference in routing performance, even with the same hardware. At any rate, we hope to add processor info (and maybe RAM and Flash capacity) in a future chart edition, because of the reasons you state.

Quote:
If Tom's lab could produce some sort of torture test that would artificially simulate as many connectionsas the router can handle, that would be nice. You could (for pratical reasons) even cap the test at X hours (example: over 5 hours, all routers that would survive would be given the torture test survival label

Another good suggestion. I know I sound repetitive, but ISP network settings and the cable modem used might be just as much the culprit. Still might be worth a run with a few routers. What application were you using in your 36 hour test?
April 20, 2006 9:15:56 PM

Quote:
As different brands use common components, could you make a reference for routers that use the same components, despite differences in functions, because performance should (theoretically) be comparable.

Good point. However, firmware can make an appreciable difference in routing performance, even with the same hardware. At any rate, we hope to add processor info (and maybe RAM and Flash capacity) in a future chart edition, because of the reasons you state.

Quote:
If Tom's lab could produce some sort of torture test that would artificially simulate as many connectionsas the router can handle, that would be nice. You could (for pratical reasons) even cap the test at X hours (example: over 5 hours, all routers that would survive would be given the torture test survival label

Another good suggestion. I know I sound repetitive, but ISP network settings and the cable modem used might be just as much the culprit. Still might be worth a run with a few routers. What application were you using in your 36 hour test?

First comment:
My personal feeling is that for me the most important information is to which router I can make a (sort of) comparison with mine (I think it's really the big issue), and if there's a better option for me out there.
I'd even put some sort of button in the chart on each of the routers, when pressed it would deploy a list of similar routers.
Processor name is only relevant if I feel like drilling down my router and search for additional info (mods, custom BIOS, whatever). I'd rather put that in a sort of "extra"-like paragraph (altough this kind of information is surely appreciated by readers).

As for BIOS version, given Tom's target, I'd just assume that readers update firmware to the latest version, and present directly results accordingly. Only if there's I huge performance gap, I'd make a comment about it (not need to talk about the bugfix from a version to another I guess).

I know that reviewers get hardware at the begining of the introduction cycle, and therefore when final consumers like myself get the same model, I get to update to a firmware that wasn't available at the time of the review, but that's another story...



Second comment:
I was using real life applications: heavy LAN transfers (full PC backups and more), and also Wireless between computers.
The killer was when on top of that I turned on eMule; even with its speed capped (to not take all Internet bandwidth) the router would overheat and freeze.

I only speak as an complete amateur, but I'd suggest the following torture test:
take a router, and a pc with 2 wired LAN ports and also a wireless port, and a "fake" DSLAM side modem.
Now plug in both wired cables to the PC and the router, and simulate with your specially crafted program a "storm", with full stress (simulate multiple clients, that keep connecting/disconnecting, I don't know, just do your geek magic!).
I'd do my best to torture all the router's functions, for instance if its got a built in modem, I'd try to simulate a modem connection with a "fake" DSLAM up to its full speed, and put also the wireless to send/receive packets to the "fake" wired ports.

Five hours at that stress would surely give accurate results.

I'd like to stress that my router could survive on me for weeks with casual use.
Only with simultaneous use of all its functions (wireless, DSL modem, LAN, firewall), the generated heat made it crash.

Cheers,

PS: I hope I'm not making a fool out of myself, because I'm just a manager talking about stuff that are absolutely NOT in my domain... you guys are the real pros!
!