Welcome to the Spectrum Community

Get answers, share solutions, help your community.

Start here

High Latency in Florida

BillGeisBillGeis Posts: 16 ✭✭✭
edited August 21 in Connectivity Aug 09, 2020

I live in St.Petersburg, Florida and around 4 months ago my ping spiked to 75-100ms and hasn't gone down since. I have no other ISPs in my area to switch to so I really need this to get fixed.


Here is traceroute to Google's public DNS:


f1 <1 ms <1 ms <1 ms www.routerlogin.com [192.168.1.1]


2 11 ms 10 ms 20 ms 072-031-128-049.res.spectrum.com [72.31.128.49]


3 31 ms 10 ms 10 ms 071-046-023-111.res.spectrum.com [71.46.23.111]


4 13 ms 15 ms 15 ms bundle-ether25.tamp05-car2.bhn.net [72.31.117.146]


5 21 ms 19 ms 17 ms 072-031-003-170.res.spectrum.com [72.31.3.170]


6 23 ms 18 ms 18 ms bu-14-orld71-car2.bhn.net [71.44.1.215]


7 18 ms 15 ms 17 ms 072-031-188-176.res.spectrum.com [72.31.188.176]


8 89 ms 87 ms 87 ms 10.bu-ether15.orldfljo00w-bcr00.tbone.rr.com [66.109.6.98]


9 82 ms 86 ms 87 ms bu-ether18.atlngamq47w-bcr01.tbone.rr.com [66.109.1.72]


10 85 ms 87 ms 88 ms 209-18-36-61.tbone.rr.com [209.18.36.61]


11 89 ms 85 ms 89 ms 108.170.225.166


12 86 ms 91 ms 86 ms 108.170.225.107


13 86 ms 89 ms 88 ms dns.google [8.8.8.8]


Here is a traceroute to a closer server in Miami, FL:


1 <1 ms <1 ms <1 ms www.routerlogin.com [192.168.1.1]


2 10 ms 14 ms 9 ms 072-031-128-049.res.spectrum.com [72.31.128.49]


3 10 ms 13 ms 10 ms 071-046-022-111.res.spectrum.com [71.46.22.111]


4 15 ms 11 ms 11 ms bundle-ether25.tamp05-car2.bhn.net [72.31.117.146]


5 13 ms 15 ms 11 ms 072-031-003-170.res.spectrum.com [72.31.3.170]


6 12 ms 15 ms 12 ms hun0-1-0-5-tamp20-cbr1.bhn.net [72.31.3.97]


7 82 ms 78 ms 79 ms 10.bu-ether15.tamsflde20w-bcr00.tbone.rr.com [66.109.6.96]


8 81 ms 79 ms 87 ms bu-ether17.hstqtx0209w-bcr00.tbone.rr.com [66.109.1.70]


9 76 ms 75 ms 77 ms ge-3-1-0.cr1.sea20.tbone.rr.com [66.109.6.90]


10 83 ms 76 ms 77 ms 66.109.5.121


11 86 ms 78 ms 79 ms 66.109.9.158


12 107 ms 110 ms 103 ms 94.142.98.35


13 128 ms 125 ms 126 ms 94.142.119.188


14 105 ms 102 ms 106 ms 84.16.8.37


15 * * 108 ms 173.44.32.250


16 107 ms 105 ms 104 ms 173.44.32.10




I was wondering what caused this and if there is any easy way to switch the way spectrum routes me through. I switched my DNS as well and that didn't change anything.


I have tried using a VPN or ExitLag to change my route and it did not help.

«1

Comments

  • Julia_RJulia_R Posts: 4,287 Lead Mod
    Aug 09, 2020

    @BillGeis Welcome to the Community!


    If you will trace to some actual websites, not DNS servers, it would help to look deeper. Just pick

    three different sites. Any 3 websites/game servers or what-have you that you normally would use.


    Thanks!.


    Julia_R

  • BillGeisBillGeis Posts: 16 ✭✭✭
    Aug 09, 2020 Accepted Answer

    I did 3 different tests. One to Google. One to an AWS server and one to Twitch.tv.


  • BillGeisBillGeis Posts: 16 ✭✭✭
    Aug 11, 2020

    @Julia_R

    I did 3 different tests. One to Google. One to an AWS server and one to Twitch.tv.


  • BillGeisBillGeis Posts: 16 ✭✭✭
    Aug 13, 2020

    @James_M For that there was one spike but it still doesn't answer why there is a constant high ping on the RoadRunner path, and if it is possible to fix that.

  • BillGeisBillGeis Posts: 16 ✭✭✭
    Aug 13, 2020

    @James_M I am talking about the hop that is constant across all of the traces, where it goes from around 10-20ms to 75 -90 which is more than double what is was a few months ago. On twitch.tv it was the 7th hop, and aws it was the 7th hop aswell.

  • James_MJames_M Posts: 2,272 ADMIN
    Aug 13, 2020

    Understood. Network congestion is likely caused by the increased number of customers that have been working from home in the past 4+ months. Results that are around 78 -79 ms simply suggests increased demand.

  • BillGeisBillGeis Posts: 16 ✭✭✭
    edited August 13 Aug 13, 2020

    @James_M From what I have seen it doesn't look like it's from increased demand because it is still bad at 1-2am and I get pretty much get the same ms when connecting to servers on the other side of the country. Suggesting that I am going through a path I should not be taking.

  • James_MJames_M Posts: 2,272 ADMIN
    Aug 13, 2020

    Thanks. Two follow up questions. Are you connected wireless? Also, you identified the ping times, but what issue is it creating for you?

  • BillGeisBillGeis Posts: 16 ✭✭✭
    edited August 13 Aug 13, 2020

    @James_M I am using Ethernet and it is making it very hard to game and almost impossible to compete against people with normal 30-40ms latency.

    I am a competitive gamer and it is affecting my income.

  • James_MJames_M Posts: 2,272 ADMIN
    Aug 13, 2020

    Thanks. I was able to locate your account based on your registration information. It looks like the modem has been online for 18+ days. Can you restart the modem and repeat and post the tracerts?

  • BillGeisBillGeis Posts: 16 ✭✭✭
    Aug 13, 2020

    @James_M I restarted the modem and here is what I got.


  • RAIST5150RAIST5150 Posts: 797 ✭✭✭✭
    edited August 13 Aug 13, 2020

    @BillGeis

    Unless you are extremely close to the host, you are on the wrong medium to use 30-40ms as the metric for end to end latency. That is something that always worked against cable internet. Pull up your rate card to see the typical latency for within your market and perhaps you will see what I mean by that.

    Remember also, these are milliseconds. That metric falls awfully close to the blink of an eye... right around 2 frames at 60fps.

    Also keep in mind your tracert data is tracking the time to receive a response for an ICMP ECHO request... which can simply get delayed or ignored under high times of utilisation... it does not automatically follow that forwarding is getting delayed. Not uncommon to see jumps in ECHO response mid route, but the end point response is considerably lower... as in it may break 150+ in the middle, but still fall under 100 at the end. Have to be mindful of WHERE you are requesting that response from, as well as the path taken to get there.


    So if you are having latency issues to a specific service, best to provide traces relevant to that particular service. ie, if it is in New York, need to be hitting servers relevant to that location.If it is hosted on Cogentco out of Montreal... than you should be tracing to those servers, or at least Cogentco nodes in Montreal.

  • BillGeisBillGeis Posts: 16 ✭✭✭
    Aug 13, 2020

    @RAIST5150

    I am normally connecting to AWS East which is located in Virginia. I also used to get the 30-40ms latency to the servers before late February 2020. Also, the big jumps to 150+ is not what I am referring to, it is the constant 80ms. Even though it may be only 2 frames on 60fps(which no one uses) in competitive gaming that will lose you the game or win it.

  • RAIST5150RAIST5150 Posts: 797 ✭✭✭✭
    edited August 13 Aug 13, 2020

    Median latency in my SC market is 28ms. And that is just within my market. Once I hit the IXP in GA or NC, all bets are off. And ever since sars2 hit us, those exchanges have been under load around the clock. Not to mention hurricane season... major storm brewing here right now in fact.

    That higher localised latency range is a problem that has always persisted with cable internet. Though, to be fair, it has actually improved a lot... I remember a time when it was 38ms, and we got ragged on a lot by people using other more dedicated digital options.

    Just saying... may need to frame things in the proper context. Both the expectation under current conditions and the method of testing.

    Since you mention in particular AWS East, have you tried using any of the cloud ping type tools to see how things are holding up? Might find a bit of a pattern when looking at things in a larger scale.

  • BillGeisBillGeis Posts: 16 ✭✭✭
    Aug 13, 2020

    @RAIST5150

    I have not, what website/tool should I use?

  • RAIST5150RAIST5150 Posts: 797 ✭✭✭✭
    edited August 13 Aug 13, 2020

    Just ran cloudping and AWS East reported 81ms from Florence, SC... on an AT&T business class line...

    grrr... forgot I turned off remote access to my router again, so can't run a direct ping from my Spectrum account. But typically, it is around a 10-20ms variance at this range... with ATT having the better response time.

  • BillGeisBillGeis Posts: 16 ✭✭✭
    edited August 13 Aug 13, 2020

    @RAIST5150

    I just did that one and I have lower latency on the west server than east, which leads me to believe it is not an AWS problem and that I am being routed weirdly. I also have noticed this in games so it isn't just showing me the latency wrong.

    I also want to point out that I know other people with Spectrum cable that dont have these same issues.

  • RAIST5150RAIST5150 Posts: 797 ✭✭✭✭
    edited August 13 Aug 13, 2020

    I typically go through Atlanta to go to the west. So Level3 is usually the culprit when things go bad westward... has to get above about 180 or so before it flips to an alternate peer... and that is usually Cogentco or TATA to the north first. Not much more Spectrum can do in those scenarios.

    You may also be getting shunted to Level3 for westbound traffic.

  • BillGeisBillGeis Posts: 16 ✭✭✭
    Aug 13, 2020

    So is there anything I can do or Spectrum can do or am I just screwed over until it somehow fixes?

  • RAIST5150RAIST5150 Posts: 797 ✭✭✭✭
    edited August 13 Aug 13, 2020

    Hard to say... a lot of variables in play. If it is just the usual congestion... wait for it to clear, they may acquire more upstream bandwidth/processing, or try to get some traffic rerouted--provided better alternate pathing exists in the tables, if not then they may need to wait for new announcements. Any of those scenarios will take time regardless of the path taken.


    In the short term, since you appear to be on PC and not console (your comment about 60fps), could try your hand at the VPN route if the game(s) you play are compatible. Can be hit and miss though. Many offer some form of free or limited use so you can test them before you commit any cash... some may even provide enough free usage to cover a weekend here and there.


    If you do try a VPN and it helps, may want to grab some trace data while on it... it might provide insight on some changes they could look into.

  • BillGeisBillGeis Posts: 16 ✭✭✭
    Aug 13, 2020

    Ok, a lot of the games I play don't allow VPN for TOS reasons so I guess im just screwed.

  • RAIST5150RAIST5150 Posts: 797 ✭✭✭✭
    edited August 13 Aug 13, 2020

    @BillGeis

    Out of curiosity, see what your pings/traces look like to this address:

                   52.119.234.196

    It is one of the targets from an AWS test batch. Was seeing some weirdness on the other side of Atlanta's exchange, like it was hunting to find a better route. Each time it tried to reach this destination it seemed to hunt a different path out of Atlanta, with an occasional spike and time out here and there. There was no other peer involved either... no Level3 and such. Just a direct charter/spectrum hop into AWS ASN territory.

    Makes me wonder if there isn't something up at the peer exchanges between Spectrum and AWS?

  • BillGeisBillGeis Posts: 16 ✭✭✭
    Aug 13, 2020

    @RAIST5150

    Here it is but it keeps timing out on all of the AWS tests


  • RAIST5150RAIST5150 Posts: 797 ✭✭✭✭
    edited August 14 Aug 14, 2020

    @James_M , @BillGeis

    Yeaaahh.... idk know if there is something there or not... could just be coincidence with all the old assets that got gobbled up in the merger...

    But I am seeing weird things recurring around hops designated as "Legacy Time Warner Cable IP Assets" in their registration data... most notably around hops in the 66.109.xxx.xxx territory. Even more curious that these are also the final Spectrum hops into AWS networks.

    May just purely be a side effect of elevated congestion, or stale metrics in the tables for available pathing... no idea. But things are getting screwy right around these exchanges, resulting in recurring latency spikes and intermittent rejected echo requests. Seems like some hops are spinning their wheels trying to manage the traffic or something.


    May be purely anecdotal... but it is quite curious indeed considering I saw a similar pathing pattern from both SC and FL to get to the same IP addresses in different subnets (similar pathing pattern was also in your earlier trace to that AWS compute IP the other day), even though the addresses were different, they appeared to be in the same subnet on the Spectrum side.


    Surely makes me wonder if maybe something is up in relation to peering with AWS that needs revisiting, be that shoring up resources on our side of the equation, or Amazon's... but something may be amiss at that juncture. Perhaps not enough to throw any alarms, but surely enough to be a nuisance for some.


    IDK if there is anything actionable that may come of this if @James_M were to pass this along, but when I dig a little in a third party's looking glass going from Charlotte, Atlanta, and Tampa to different addresses for AWS East-1 EC2 and such, I seem to be getting more stable results when the next hops chosen are in the CIDR ranges bound to AS36086, which appears to be some legacy Telx registration of Amazon's. BGP data shows some potentially useful IX points for use... Ashburn, Atlanta, Dallas, Pheonix, Chicago, L.A., San Fran... Again...don't know if it amounts to anything useful, but the end to end latencies in this LG were pretty tight... talking a range of 15-28, scaling appropriately by distance. Even if we stepped a little out of our "up to 28ms median" latency and tacked that on there, this might have some potential considering we are seeing 70-100 with what is currently in our tables. IDK if there is any way to review metrics as if this ASN was put into play somehow with the peering arrangement or not... just was something interesting to see happening so thought I would toss it out there just in case.

  • BillGeisBillGeis Posts: 16 ✭✭✭
    Aug 14, 2020

    @RAIST5150

    Hopefully this gets brought up to Spectrum.

    I just wanted to say thank you for what you have done so far to help me though.

  • RAIST5150RAIST5150 Posts: 797 ✭✭✭✭
    edited August 14 Aug 14, 2020

    No biggie.

    Trying to find/address routing issues is a pain. So much of it just "happens" automatically now... and the games/services get so locked down it makes it near impossible for people to gather good data. Best we can do sometimes is like playing with horseshoes or hand grenades... chuck somethimg out there and hope it hits the mark.

    Was just bugging me I couldn't find anything concrete... could see something wonky, but just couldn't pin anything down... had to try poking around. AWS is tricky without the proper resources. Was so much easier in my time.... the more traditional client/server designs.

    Perhaps someone who still looks at this crud for a living can spot something useful.

  • karlbeckmankarlbeckman Posts: 2,223 ✭✭✭✭
    Aug 14, 2020

    Maybe this user @BillGeis can explain why he needs to go outside Spectrum to use a Google DNS server when Spectrum already has several DNS server sites strategically located inside their transport network. They all receive the same data updates from the same sources. If you use the Spectrum DNS sites you can avoid the data traffic congestion at every IXP handoff and have MUCH lower ping times as a result.

  • BillGeisBillGeis Posts: 16 ✭✭✭
    Aug 14, 2020

    @karlbeckman

    I don't know what you mean. I was only using a Google DNS as a test to show that my ping is high and the path it took. I just tested a Spectrum DNS and it came back with the exact same result. I have also tried to use many different DNS to try to lower it in my router configuration.

Sign In or Register to comment.