Calguns.net  

Home My iTrader Join the NRA Donate to CGSSA Sponsors CGN Google Search
CA Semiauto Ban(AW)ID Flowchart CA Handgun Ban ID Flowchart CA Shotgun Ban ID Flowchart
Go Back   Calguns.net > GENERAL DISCUSSION > Technology and Internet
Register FAQ Members List Calendar Mark Forums Read

Technology and Internet Emerging and current tech related issues. Internet, DRM, IP, and other technology related discussions.

Reply
 
Thread Tools Display Modes
  #1  
Old 02-15-2019, 9:30 PM
high_revs's Avatar
high_revs high_revs is offline
Calguns Addict
 
Join Date: Feb 2006
Location: Silicon Valley
Posts: 6,210
iTrader: 12 / 100%
Default my 5 week docsis3.1 cable modem no good anymore?

couldn't get upload speed past 10mpbs w/o a gigabit package, so i opted for it. I did have to get a new cable modem - motorola mb8600 to replace the motorola/arris docsis 3.0 that served me for maybe 5.5 yrs. I did get to mid 300mpbs with the arris. I only do the tests via ethernet connection.

I wasn't expecting full gigabit speed but below was close enough. I don't really monitor it. oddly the last 2 weeks, the speed slowed to half. on avg, high 300mpbs. very occasionally, i'd get 500'ish mpbs. no firmware update on router. i test direct connect to the modem (no router in middle). comcast came today and checked the signal strength and they're all good. after the guy's test, i got to 750'ish mpbs only to go back down to high 400 mpbs after.

did the modem get bad? i used the same coax cable the cable modem came in. though i'm not really using it, i'm still paying for it.

I also use speedtest.net and usually I get much better speeds connecting to ISPs in San Jose also. Oddly it's been like that for years now (faster speed using sites in San Jose).

P.S. I did read somewhere that the router's QOS sometimes mess up d/l speeds. I reset the router also and did fresh QOS speedtests. same results. But I do test direct to the modem and I really doubt the issue is the router also (no firmware upgrades recently).

test below is done using router.




Last edited by high_revs; 02-15-2019 at 9:33 PM..
Reply With Quote
  #2  
Old 02-15-2019, 9:44 PM
bigmike82 bigmike82 is offline
Bit Pusher
CGN Contributor
 
Join Date: Jan 2008
Location: W. Los Angeles
Posts: 3,215
iTrader: 64 / 100%
Default

Have you tried swapping out the cable from the modem to the wall? And verifying that every is tight and not loose?

Coax can be very susceptible to wonky/loose connections. That would be my first test. Make sure you're solid from port on the modem all the way to the wall.

OH. Any splitters involved? Those will down your signal strength a bit.

My second one would be buying another modem from amazon and testing it. If the results are the same, return it; otherwise, there's your culprit.
__________________
-- 09 F9 11 02 9D 74 E3 5B D8 41 56 C5 63 56 88 C0
Reply With Quote
  #3  
Old 02-15-2019, 9:55 PM
SkyHawk's Avatar
SkyHawk SkyHawk is offline
Front Toward Enemy 🔫
CGN Contributor
 
Join Date: Sep 2012
Location: Nakatomi Plaza - 30th floor
Posts: 12,398
iTrader: 138 / 100%
Default

Who is your ISP?

If you just want to confirm the modem and the line speed, you have to eliminate every possible variable. That means only using your ISP speed test platform, not some random speed test target that requires hops through peers. No matter how geographically close. So don’t use Speedtest.net, use your ISPs designated speed test target.

You also have to eliminate any NAT/firewall/inspection devices(s). And ideally you also eliminate a PC or other general purpose OS/compute device, or at least test using a bare bones, purpose tuned general purpose device/OS.

IMO, Gb to the home is sort of a marketing gimmick. I am not saying it is not real and your own tests confirm you can get close, but doing it correctly and reliably and with repeatable results, for multiple client devices concurrently, is not cheap for a home user. IMO the best you can count on repeatedly is 300-400Mb and that also means you can have reasonably priced NAT/firewall/inspection devices that can keep up with that rate. But if you want to serve a home full of devices and have aggregate access for them to 1Gb - you are looking at some serious coin for a router and/or firewall that can do 1Gb. And even then, once your traffic leaves your ISP network, all bets are off.
__________________
.


Last edited by SkyHawk; 02-15-2019 at 10:01 PM..
Reply With Quote
  #4  
Old 02-16-2019, 8:22 AM
high_revs's Avatar
high_revs high_revs is offline
Calguns Addict
 
Join Date: Feb 2006
Location: Silicon Valley
Posts: 6,210
iTrader: 12 / 100%
Default

bikemike,
i'll try swapping a new coax. both sides are tight. no splitter. i only have internet so the jack is by itself from the outside.

skyhawk,
it's comcast. and i'm using comcast's own speed test. even in speedtest.net, i normally selects comcast. my router can handle it. i got those speeds before. i also use a gigabit switch (not that matters for speedtest since i'm direct connected at times to modem).
Reply With Quote
  #5  
Old 02-16-2019, 10:25 AM
Robotron2k84 Robotron2k84 is offline
Senior Member
 
Join Date: Sep 2017
Posts: 708
iTrader: 2 / 100%
Default

There are a couple of things to remember.

1) If your service uses an ONT (the box outside that transitions fiber to coax), then you are on Comcast's old gig deployment strategy. Get them out to diagnose the ONT.

2) If you do NOT have an ONT, then you are on Comcast's recent residential coax-only gig service that simply aggregates more channels via bonding. It's important to note that this tier of service works identically to classical cable-internet and all nodes on the neighborhood uplink or head-end port are aggregate. Tl;dr, your neighbors still affect your throughput and time-of-day can be an issue.

3) Even though you own your router/modem, two things can affect your performance: Comcast-delivered CPE firmware updates and routing table updates. Get in touch with their level-II ISP techs and see if your modem is provisioned properly.

4) For multi-gig firewall and packet inspection, routing, you don't need a high-end appliance. Most i5-based SFF PCs can have any number of router-firewall OSs installed that will exceed gig-throughput for home use. All you really need is a NUC with 2 Gbe ports running pfSense or similar. With Nbase-T starting its rollout 10Gbe will be commonplace next year.
Reply With Quote
  #6  
Old 02-16-2019, 11:17 AM
SkyHawk's Avatar
SkyHawk SkyHawk is offline
Front Toward Enemy 🔫
CGN Contributor
 
Join Date: Sep 2012
Location: Nakatomi Plaza - 30th floor
Posts: 12,398
iTrader: 138 / 100%
Default

pfSense, yep free is always cheaper If OP can setup, install and run an open source firewall, then that is not a bad way to go. If he wants to use Cisco or Meraki with 1Gb speed, bust out another thousand (or two or three).

OP, I would be using this speedtest for Comcast just to make sure your test stays on their network http://speedtest.xfinity.com
__________________
.

Reply With Quote
  #7  
Old 02-17-2019, 8:28 PM
MrFancyPants's Avatar
MrFancyPants MrFancyPants is offline
Member
 
Join Date: Jun 2017
Location: Salt Lake City
Posts: 363
iTrader: 3 / 100%
Default

The problem with gigabit over coax is always the inconsistent download speeds and horrendous upload speeds. Channel bonding is not as reliable as a single fat pipe. Fiber to the home is the only way to go. I believe the ISP provided router is the reason my download speed is less than optimal, but I just haven't had time to set up my Cisco ASA yet. I pay $65 per month with no contract for gigabit fiber, love it. Costs me less than Comcast charged for 250 Mbit.

Sent from my SM-N960U using Tapatalk
Reply With Quote
  #8  
Old 02-17-2019, 9:20 PM
Robotron2k84 Robotron2k84 is offline
Senior Member
 
Join Date: Sep 2017
Posts: 708
iTrader: 2 / 100%
Default

Generally with cable internet, upload consistency is not as variable as download, save for the case of insufficient broadcast power via faulty wiring.

For download speed tests, I rarely use the Speedtest GUI as I don't want to use a speed test server that everyone else is hammering on, thereby affecting my results. The 500 MB ZIP is also not a test that is exposed in the GUI.

Usually, I just do this on the router or firewall, whichever is the edge device:

Code:
curl -o /dev/null http://speedtest.sjc01.softlayer.com/downloads/test500.zip
and monitor the instantaneous and aggregate throughput. I also do it via http so that my results aren't skewed by SSL overhead. I'm not sure what Speedtest does, I haven't run a packet snooper on it for a long time. The ZIP link obviates any cache or acceleration, and Comcast does not do turbo or whatever burst prioritization on their business side.

Just to prove the point, running the GUI now, yields a speed of 24.7 Mbps on my bottom-tier Comcast business connection that is rated 25/5. When actually doing the test from the firewall, the speed is 28.4/6.2 via curl, which I trust a heck of a lot more than Speedtest's GUI. I have found that on Comcast's business network they generally give you 10-15% more bandwidth than advertised so they don't get sued as it's a business connection with an SLA.

Last edited by Robotron2k84; 02-17-2019 at 9:49 PM..
Reply With Quote
  #9  
Old 02-17-2019, 9:39 PM
high_revs's Avatar
high_revs high_revs is offline
Calguns Addict
 
Join Date: Feb 2006
Location: Silicon Valley
Posts: 6,210
iTrader: 12 / 100%
Default

robotron,

very deep insight. thanks. i had a tech come out yesterday and they only based the analysis on signal strength. i only see coax connection from the post so i don't believe it's fiber to house. i don't see any kind of converter box. supposedly they have fiber in my neighborhood but i'll need to verify that again. and will ask for L2.

skyhawk, i normally use speedtest.xfinity.com and just compare against speedtest.net. i do fully see testing would be on their network, vs. someone else. i managed to get their speeds before but with mimosa or speedtest or sonic in san jose.


belwo are all my results to date. it dropped again to just mid 300's right now. wish i was tracking it regularly before when i got 900's so i can pinpiont when the drop started.

grrrr


http://speedtest.xfinity.com/results/JS1E28D62Q2L45Q
http://speedtest.xfinity.com/results/JS1E7WEZIYHIUWO
http://speedtest.xfinity.com/results/JS2KYX4281SQ7WT
http://speedtest.xfinity.com/results/JS2NYO2NUXT4ZWQ
http://speedtest.xfinity.com/results/JS4Q8G9TVQ0ISLG
http://speedtest.xfinity.com/results/JS4QR99GFBLN7TK
http://speedtest.xfinity.com/results/JS5F9W5AT9X55B4
http://speedtest.xfinity.com/results/JS6JEU1OTM3CDI3
http://speedtest.xfinity.com/results/JS6JFL3M5MWLCYG
http://speedtest.xfinity.com/results/JS6JYUTF3TUBLH0
http://speedtest.xfinity.com/results/JS6L5PGMYWJHXBB
http://speedtest.xfinity.com/results/JS6Y3ZFBI72J1AZ
http://speedtest.xfinity.com/results/JS9O88S6ACE678R
http://speedtest.xfinity.com/results/JS9OFVOJKA6VT3Y
http://speedtest.xfinity.com/results/JS9PDTCNXFAJL3K
http://speedtest.xfinity.com/results/JS9YTMQG1HH43SU
Reply With Quote
  #10  
Old 02-19-2019, 9:20 AM
high_revs's Avatar
high_revs high_revs is offline
Calguns Addict
 
Join Date: Feb 2006
Location: Silicon Valley
Posts: 6,210
iTrader: 12 / 100%
Default

Quote:
Originally Posted by Robotron2k84 View Post
Generally with cable internet, upload consistency is not as variable as download, save for the case of insufficient broadcast power via faulty wiring.

For download speed tests, I rarely use the Speedtest GUI as I don't want to use a speed test server that everyone else is hammering on, thereby affecting my results. The 500 MB ZIP is also not a test that is exposed in the GUI.

Usually, I just do this on the router or firewall, whichever is the edge device:

Code:
curl -o /dev/null http://speedtest.sjc01.softlayer.com/downloads/test500.zip
and monitor the instantaneous and aggregate throughput. I also do it via http so that my results aren't skewed by SSL overhead. I'm not sure what Speedtest does, I haven't run a packet snooper on it for a long time. The ZIP link obviates any cache or acceleration, and Comcast does not do turbo or whatever burst prioritization on their business side.

Just to prove the point, running the GUI now, yields a speed of 24.7 Mbps on my bottom-tier Comcast business connection that is rated 25/5. When actually doing the test from the firewall, the speed is 28.4/6.2 via curl, which I trust a heck of a lot more than Speedtest's GUI. I have found that on Comcast's business network they generally give you 10-15% more bandwidth than advertised so they don't get sued as it's a business connection with an SLA.
I"ll have to figure this out how to do a command line with my orbi...I think there is a way but last i recall reading, it's a little tricky.

I called them again last night, they won't let me go to L2 w/o yet another tech visit to my home. i (nicely) challenged what the tech would do any different this time other than going thru a formal process of a repeat issue. Guess I do have to go thru it before they get me to L2. grrrr

Doing this before i call the customer loyalty group to say "hey, you're not meeting your service delivery.. what do we do now?"
Reply With Quote
  #11  
Old 02-19-2019, 9:39 AM
Robotron2k84 Robotron2k84 is offline
Senior Member
 
Join Date: Sep 2017
Posts: 708
iTrader: 2 / 100%
Default

Another reason to get their business package, direct connect to the ISP support folks.

In all seriousness. When the tech comes out, have them call the L2 folks and they have you talk to them. Open a support case so you can continue to work on it with them without the line-puller there.
Reply With Quote
  #12  
Old 02-19-2019, 3:33 PM
Oddball_One's Avatar
Oddball_One Oddball_One is offline
Member
 
Join Date: Mar 2011
Location: Butte County
Posts: 119
iTrader: 3 / 100%
Default

Quote:
Originally Posted by Robotron2k84 View Post
Another reason to get their business package, direct connect to the ISP support folks.

In all seriousness. When the tech comes out, have them call the L2 folks and they have you talk to them. Open a support case so you can continue to work on it with them without the line-puller there.
Good advice. Sorry SkyChicken I disagree with PFSense for the average user. Too much configuration. Yes, most off the shelf routers cannot handle 1 Gb service. Look into the untangled appliance. Basically a branch of PFSence dumbed down.

Another thing to consider is buffer bloat.

https://www.bufferbloat.net/projects.../Introduction/

Easy handled by good routers
__________________
"Don't hit me with them negative waves so early in the morning."
Reply With Quote
  #13  
Old 02-19-2019, 4:33 PM
Robotron2k84 Robotron2k84 is offline
Senior Member
 
Join Date: Sep 2017
Posts: 708
iTrader: 2 / 100%
Default

I'm not a huge fan of pfSense either, not for any reason except I prefer Linux over *BSD for appliances. But, it tends to run on the widest swath of hardware compared to other firewall-specific distros. It's not a terrible pain to get up to speed and learn as you go.

But, I'm a hard case and roll my own firmware, forking various OpenWRT / Tomato bits.
Reply With Quote
  #14  
Old 02-22-2019, 10:21 AM
high_revs's Avatar
high_revs high_revs is offline
Calguns Addict
 
Join Date: Feb 2006
Location: Silicon Valley
Posts: 6,210
iTrader: 12 / 100%
Default

well hell... second appointment today... tech's device is showing 975mpbs. the tech actually said that's the highest he's seen on gigabit residential! he used a device called cioeco and i took pics of it below. i'll look into purchasing it if cheap enough.

My Dell L521x laptop is supposed to be a 100/1000 ethernet iirc. i used my kid's 6 month HP old laptop also (thinking it's gigabit ethernet) and it's not getting that gig speed, but i did get that speed once using my Dell laptop it once using comcast speedtest as noted above screenshot, and via a router at that!

i'm just befuddled at this point. i'll move the tower to my office and try again there using that ethernet NIC. I guess best thing is really just either leave it alone (can't fully leverage from one machine anyway, and upload is really what I was after) or get a usb-type gigabit nic for my Dell - https://smile.amazon.com/USB-Gigabit...gateway&sr=8-7 . (i'm really the only one who uses ethernet, and the dell xps8700 2yr old tower mainly as a media server).

below is the tech's device.



Last edited by high_revs; 02-22-2019 at 10:24 AM..
Reply With Quote
  #15  
Old 02-22-2019, 12:28 PM
Robotron2k84 Robotron2k84 is offline
Senior Member
 
Join Date: Sep 2017
Posts: 708
iTrader: 2 / 100%
Default

Realize that non-bus master NICs that utilize software checksumming have a rough theoretical maximum throughout of 330Mbps.

The NICS you are using for testing, are they Intel Gbe, or Realtek by chance? Realtek onboard is widely known to not be able to achieve gig throughput and only certain Intel ones can do all the work in silicon to sustain gig.
Reply With Quote
  #16  
Old 02-22-2019, 12:42 PM
Oddball_One's Avatar
Oddball_One Oddball_One is offline
Member
 
Join Date: Mar 2011
Location: Butte County
Posts: 119
iTrader: 3 / 100%
Default

https://www.speedtest.net/insights/b...-tips-to-help/
__________________
"Don't hit me with them negative waves so early in the morning."
Reply With Quote
  #17  
Old 02-22-2019, 3:33 PM
high_revs's Avatar
high_revs high_revs is offline
Calguns Addict
 
Join Date: Feb 2006
Location: Silicon Valley
Posts: 6,210
iTrader: 12 / 100%
Default

Quote:
Originally Posted by Robotron2k84 View Post
Realize that non-bus master NICs that utilize software checksumming have a rough theoretical maximum throughout of 330Mbps.

The NICS you are using for testing, are they Intel Gbe, or Realtek by chance? Realtek onboard is widely known to not be able to achieve gig throughput and only certain Intel ones can do all the work in silicon to sustain gig.
The Dell laptop is Realtek. My 4 yr old Lenovo worklaptop is Intel, which I'll use to test tonight.

I can't recall the 6 month old HP laptop at home of my 2 yr old dell 8700 desktop. will check tonight.

still strange or was it by pure luck I got the speed captured at least one time?
Reply With Quote
  #18  
Old 02-25-2019, 8:26 PM
high_revs's Avatar
high_revs high_revs is offline
Calguns Addict
 
Join Date: Feb 2006
Location: Silicon Valley
Posts: 6,210
iTrader: 12 / 100%
Default

Quote:
Originally Posted by Robotron2k84 View Post
Realize that non-bus master NICs that utilize software checksumming have a rough theoretical maximum throughout of 330Mbps.

The NICS you are using for testing, are they Intel Gbe, or Realtek by chance? Realtek onboard is widely known to not be able to achieve gig throughput and only certain Intel ones can do all the work in silicon to sustain gig.
I bought a startech.com usb based NIC... I'm getting high speeds now.

Really strange i got these speeds early unless a driver update from microsoft changed things. i downloaded the dell driver to get it back to original state, and also a realtek driver off their site. in the meantime, i guess i am getting the speeds.

sucks no easy way to switch the NIC on a laptop w/o using one of the usb ports.

Reply With Quote
  #19  
Old 02-25-2019, 8:41 PM
Robotron2k84 Robotron2k84 is offline
Senior Member
 
Join Date: Sep 2017
Posts: 708
iTrader: 2 / 100%
Default

Glad it's finally a solved issue at least. Now you know what to look for in future PC purchases.
Reply With Quote
Reply

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump



All times are GMT -8. The time now is 6:21 PM.




Powered by vBulletin® Version 3.8.9
Copyright ©2000 - 2019, vBulletin Solutions, Inc.
Proudly hosted by GeoVario the Premier 2A host.
Calguns.net, the 'Calguns' name and all associated variants and logos are ® Trademark and © Copyright 2002-2018, Calguns.net an Incorporated Company All Rights Reserved.
Calguns.net and The Calguns Foundation have no affiliation and are in no way related to each other.
All opinions, statements and remarks made by Calguns.net on this web site and elsewhere are solely attributable to Calguns.net.