dslreports logo
site
 
    All Forums Hot Topics Gallery
spc

spacer




how-to block ads


Search Topic:
uniqs
19
share rss forum feed

guppy_fish
Premium
join:2003-12-09
Lakeland, FL
kudos:2
Reviews:
·Verizon FiOS
reply to Vamp

Re: [Northeast] High pings / bad routing...

I would say the folks in western NY are all being routed thru NYC, which being the one of the busiest data hub in the country its not surprising that pings are not great. That doesn't mean much as routers are programed to only respond and hence forward IMCP pings as the lowest priority traffic.

Ping is NOT a reliable metric for finding latency issues for network paths, people see ping number and think they have proof of an issue, and sometimes in gross mis-managed networks it can show issues, but most the time as in ALL of your three tracert, nothing shows as an issue.

Ping and Speed tests are not a smoking gun, the first due to router priorities the second, due to server load as all unknown. You would need the ability to do end to end large data-transfers to get hard statistics of what the network delays are.

Most of what people think is there network slow down or gaming lag is due to the content server, as the internet grows, content company's will do all they can to maximize there income, buying more bandwidth for peak usage isn't a good return on investment

All the posters about Verizon this, Verizon that network issues are for the most part misdirected in there ire.

Lastly, there is NO larger backbone than Verizons, they are about 30% of the US internet backbone, the amount of traffic that FIOS adds is noise compared to the amount of data flowing on alternet ( the original uunet , which was MCI, now Verzon business )


serge87

join:2009-11-29
Reviews:
·Verizon FiOS

said by guppy_fish:

I would say the folks in western NY are all being routed thru NYC, which being the one of the busiest data hub in the country its not surprising that pings are not great. That doesn't mean much as routers are programed to only respond and hence forward IMCP pings as the lowest priority traffic.

Ping is NOT a reliable metric for finding latency issues for network paths, people see ping number and think they have proof of an issue, and sometimes in gross mis-managed networks it can show issues, but most the time as in ALL of your three tracert, nothing shows as an issue.

Ping and Speed tests are not a smoking gun, the first due to router priorities the second, due to server load as all unknown. You would need the ability to do end to end large data-transfers to get hard statistics of what the network delays are.

During the time I took those trace routes, I was only having successful webpage loads 50-70% of the time(not just limited to those 3 sites!). For a large ISP with the advanced network infrastructure it has that is a sad state of affairs. People experiencing up to 10ms latency to their local VZ gateway router(2nd hop! /wtf) and timeouts on the 3rd and 4th hop(note: no traffic has left Verizon's network at that point, please don't blame third-party internet entities!) is clear that Verizon can't shrug and throw their hands up and say "not my problem" when it can be shown reliably from numerous sources.

said by guppy_fish:

Most of what people think is there network slow down or gaming lag is due to the content server, as the internet grows, content company's will do all they can to maximize there income, buying more bandwidth for peak usage isn't a good return on investment

All the posters about Verizon this, Verizon that network issues are for the most part misdirected in there ire.

Lastly, there is NO larger backbone than Verizons, they are about 30% of the US internet backbone, the amount of traffic that FIOS adds is noise compared to the amount of data flowing on alternet ( the original uunet , which was MCI, now Verzon business )

And I'm sure hardware never, ever fails or functions abnormally. I experience major slowdowns or timeouts to various content from all types of sources on the web. Apparently I didn't get the memo where the whole internet is now stored on 1 content server with 1 pipe to Verizon...

I find it hard to believe that internet content has grown so much in the last month or two that such a stark contrast in FIOS performance has developed from years before.

guppy_fish
Premium
join:2003-12-09
Lakeland, FL
kudos:2
Reviews:
·Verizon FiOS

said by serge87:

During the time I took those trace routes, I was only having successful webpage loads 50-70% of the time(not just limited to those 3 sites!).

Latency has nothing to do with failed page loads, if that's your problem, it would have to be some form of packet loss

said by serge87:

And I'm sure hardware never, ever fails or functions abnormally. I experience major slowdowns or timeouts to various content from all types of sources on the web.

All teir one ISP's have very complete auto fail-over systems in place, that's the whole point of many of the used routing algorithms which the internet runs on

said by Xtreme2damax:

Routing with isp's is strange to say the least, really no reason why I should be routed to VA when connecting to a server here in NY.

This is likely the core issue, but its just as much as where the servers are and on whom's network. Preferred routing comes from the hosts, and here lies a common problem. Host A will broadcast the preferred path ( usually on where they get the lowest cost peering ) and these get into the Verizon routers tables, but that may not be the best path from a Verizon customers perspective. So Verizon has to deal with this and things can get bounced around.

Its a global issue, not a Verizon specific issue and for sure there are times a core router tables get foo-bar'ed , as all the possible providers in the chain from you to the host, they can infect other providers performance to to table updates.

To be clear, I'm not suggesting if something is not working for you that is OK, I'm trying to in a friendly way provide information on how this stuff works, and that our ISP Verizon may have nothing to do with the performance issue but could just be a tier 3 network spewing bad BGP table updates

Summary, the HOST provides BGP updates for the preferred routing, which is to the advantage of the Host, when things aren't at capacity, works great, when peering points get overloaded, thing can get horribly bounced around and cause issues. Yes, a network provider can do things to cause issues, but peak performance problems, which is what is being described here starts at the Host. Hope this helps your understanding of how things work in this maze of inter-tubes

»en.wikipedia.org/wiki/Border_Gat···Protocol