any tips to reduce ping?

[farzher]

there's not a lot of information out there about ping.
i want low ping, even if it's at the cost of low bandwidth.
is there anything i can do??

switching from wifi to ethernet is obvious.
but can i do anything else?
does my router matter?
is it possible to have a slow router when it comes to ping? or do they only affect bandwidth?
is there anything my ISP can do if i call and complain? unfortunately there's no concept of an internet plan optimized for ping / latency.

i've heard there's weird vpn-like services that you can connect to that claim to reduce your ping in specific games by having optimal paths to the game servers. is that even possible?

my friend lives less than 1 ms light disance away from me.
yet we have a ping of 45ms when directly connected p2p... who's fault is that?
(i have the fastest internet plan available ...... 50 megabits at&t) wtb fiber.
 

[Anghammarad]

before going "OMFG!!! MY PING IS HIGH WUUUUTAFOCK??!??" 

 

Try finding out where the latency spikes happen.

 

For example get the free version of Pingplotter to explore the connection.

 

Ping isn't the same as Ping. It depends on the roundtrip from and to, and which IP on what net you are pinging. 

 

So 45 ms could very well be a really good ping. 

 

Just as a little information, when I ping my company router over vpn I get 14 ms, when I ping google I get 56 ms... It really depends.

 

 

[Alex Atkin UK]

5 hours ago, farzher said:

my friend lives less than 1 ms light disance away from me.
yet we have a ping of 45ms when directly connected p2p... who's fault is that?
(i have the fastest internet plan available ...... 50 megabits at&t) wtb fiber.

Are you both on the same ISP?  If you are, it depends how far into their network you need to go before routing back to your friend.  I believe cable often does it at the street level so you'd get really low latency between houses there, other networks can vary.  If for example you are using PPPoE as the connection method, it probably happens in your ISPs core network, much further away.

 

If you are on different ISPs then you're likely still traveling a fair bit across the Internet before getting to each other as your ISPs probably don't directly peer with each other.

 

I'd suspect 50Mbit on ATT will be DSL based (they incorrectly call it fiber) as their true fiber products are MUCH faster than that.

[farzher]

we're not on the same ISP. either way 45ms is ridiculous.
you can travel across the whole country in 45ms.

so you think there's nothing i can do besides move to a different ISP or a different house and just hope it's better?

 

[Falcon1986]

1 hour ago, farzher said:

we're not on the same ISP. either way 45ms is ridiculous.
you can travel across the whole country in 45ms.

so you think there's nothing i can do besides move to a different ISP or a different house and just hope it's better?

Have you ever run a traceroute to see where the ping starts to increase?

 

But out of curiosity, what kind of problem are you experiencing with a ping of 45ms? Are you sure your internet connection is 100% fiber? I normally see 2-5ms on speed tests with my 100/50Mbps fiber.

[farzher]

45 minutes ago, Falcon1986 said:

Have you ever run a traceroute to see where the ping starts to increase?

 

But our of curiosity, what kind of problem are you experiencing with a ping of 45ms? Are you sure your internet connection is 100% fiber? I normally see 2-5ms on speed tests with my 100/50Mbps fiber.

i've used traceroute but i don't know how to read anything meaningful from it.
i don't have fiber. i used confusing wording. "wtb fiber" means "i want to buy fiber". but of course i can't. it's only available in a few cities. i have cable probably.

i'm trying to play games over parsec. and just competitive games in general

[jec6613]

Background on how residential internet works:

 

When you route out of your home, you will end up on the node (cable) or at your local CO (DSL).  Functionally, fiber optic is a DSL connection, FYI.  Transmission speed is 70% or so the speed of light over copper, somewhat higher over fiber, and the cable node shares bandwidth among all subscribers on the node - so you have a 1-2ms ping to the next hop out.

 

From there, your packet will travel over your ISP's fiber optic backbone network.  Routers on that network determine the lowest cost route to your destination based on the BGP map your ISP has, and exits their fiber network over one of the peering connections.  Good ISPs have many peering connections, but for large, sprawling networks, that doesn't mean that one is necessary close to you.  Connecting AT&T to Comcast, for example, there isn't likely a direct connection.  So your packet out of AT&T will travel, often hundreds of miles, to the nearest peering point to somebody like Level3, who will then travel hundreds of miles again to the Comcast peering point, then across the Comcast network to your friend's house.  On the other hand, when connecting to Facebook, AT&T may have a peering connection to them just a few dozen miles away in a large datacenter, in which case the latency can be single digit milliseconds.

 

Most of that is out of your control, but, yes, home routers can and do introduce latency, especially if you're using WiFi.  The lowest latency is to bypass the router entirely, but that's not necessarily practical - you can expect 3-5ms just from your typical home router, and 1-2 from a good wired router.  One other thing to consider: copper Ethernet adds latency as well, so you want to connect directly to your router and not through a switch.  Each switch hop can add another 1-2ms of latency over copper (fiber or DAC switch to switch connections are much faster, microsecond latency).

 

So, one test, if you connect directly to your modem and ping your friend's public IP, what is the latency?  That will give you a good idea how much is caused internally versus externally - and your friend can easily do the same to determine how much is internal to their network versus the external latency you can do nothing about.

[Alex Atkin UK]

3 hours ago, jec6613 said:

Each switch hop can add another 1-2ms of latency over copper (fiber or DAC switch to switch connections are much faster, microsecond latency).

A switch adds nowhere near that much.

 

eg:
PING  (192.168.1.253) 56(84) bytes of data.
64 bytes from Server.lan (192.168.1.253): icmp_seq=1 ttl=64 time=0.058 ms
64 bytes from Server.lan (192.168.1.253): icmp_seq=2 ttl=64 time=0.077 ms
64 bytes from Server.lan (192.168.1.253): icmp_seq=3 ttl=64 time=0.071 ms
64 bytes from Server.lan (192.168.1.253): icmp_seq=4 ttl=64 time=0.074 ms
64 bytes from Server.lan (192.168.1.253): icmp_seq=5 ttl=64 time=0.082 ms
64 bytes from Server.lan (192.168.1.253): icmp_seq=6 ttl=64 time=0.073 ms
64 bytes from Server.lan (192.168.1.253): icmp_seq=7 ttl=64 time=0.078 ms
64 bytes from Server.lan (192.168.1.253): icmp_seq=8 ttl=64 time=0.070 ms
---  ping statistics ---
8 packets transmitted, 8 received, 0% packet loss, time 7179ms
rtt min/avg/max/mdev = 0.058/0.072/0.082/0.006 ms

PING god.lan (192.168.1.1) 56(84) bytes of data.
64 bytes from God.lan (192.168.1.1): icmp_seq=1 ttl=128 time=0.185 ms
64 bytes from God.lan (192.168.1.1): icmp_seq=2 ttl=128 time=0.281 ms
64 bytes from God.lan (192.168.1.1): icmp_seq=3 ttl=128 time=0.241 ms
64 bytes from God.lan (192.168.1.1): icmp_seq=4 ttl=128 time=0.194 ms
64 bytes from God.lan (192.168.1.1): icmp_seq=5 ttl=128 time=0.232 ms
64 bytes from God.lan (192.168.1.1): icmp_seq=6 ttl=128 time=0.212 ms
64 bytes from God.lan (192.168.1.1): icmp_seq=7 ttl=128 time=0.485 ms
64 bytes from God.lan (192.168.1.1): icmp_seq=8 ttl=128 time=0.483 ms
^C
--- god.lan ping statistics ---
8 packets transmitted, 8 received, 0% packet loss, time 7185ms
rtt min/avg/max/mdev = 0.185/0.289/0.485/0.115 ms

Bearing in mind some of the above is because the first test is 10Gbit wired to 10Gbit DAC on the same switch and the second is 10Gbit wired to 5Gbit wired uplink to a 5Gbit wired client on another switch.  The difference is much lower when going Gigabit to Gigabit.

A Gigabit switch as I recall adds about 0.1ms which is completely negligible.

[Jarsky]

21 hours ago, farzher said:

there's not a lot of information out there about ping.

ping is a colloquialism, you're talking about latency (network latency, packet latency). 

21 hours ago, farzher said:

i want low ping, even if it's at the cost of low bandwidth.
is there anything i can do??

ping / latency, is not an absolute number. the latency refers to the time to transit from 1 point to another. 

You could have 0.2s ping to your router, 8ms ping to your ISP, if youre connecting to an Amazon server that could be 50ms ping away..if you connected to me that could be 140ms ping away.

 

Ping is completely dependant on the points you're referring to, and the path in between. 

 

21 hours ago, farzher said:

does my router matter?

Not really, your typical router will be at most 1ms. 

 

21 hours ago, farzher said:

is it possible to have a slow router when it comes to ping? or do they only affect bandwidth?

Its really only going to affect features and maximum bandwidth.

 

21 hours ago, farzher said:

is there anything my ISP can do if i call and complain? unfortunately there's no concept of an internet plan optimized for ping / latency.

They can really only do something if they have an issue (e.g packet storms / faulty ports / etc...causing errors and retransmission), or changing a router to a specific location you're experiencing high latency too if its possible and within their cost to do so (pairing agreements become a thing). 

 

21 hours ago, farzher said:

i've heard there's weird vpn-like services that you can connect to that claim to reduce your ping in specific games by having optimal paths to the game servers. is that even possible?

That only helps, if it gives you a shorter route. For example, if youre trying to connect from MIchigan to LA, and lets say for some reason they route via Washington. If your VPN had a node near you in Michigan, and went via Nevada, then you'd have a lower latency. (In practice these type of services have more use for international connectivity)

 

21 hours ago, farzher said:

my friend lives less than 1 ms light disance away from me.
yet we have a ping of 45ms when directly connected p2p... who's fault is that?
(i have the fastest internet plan available ...... 50 megabits at&t) wtb fiber.

Thats not how internet works. Your internet goes out of your house to your local node, back to your ISP's point of presence. If your friend is on a different ISP it then has to go to a Peering Exchange, and then all the way back again. 

 

I assume 50Mbit, is probably coaxial? You then have to deal with signal interpolation. Data travels significantly slower through copper as well which adds additional latency in your "last mile" connection(s). Additionally light in fiber travels closer to about 2/3 of c, the speed of light doesnt apply to glass, it only applies to a vacuum like space.

 

Honestly 45ms doesnt sound that bad. It all depends on what technology is being used to provide your connections and peering & routing agreements.

 

 

 

 

 

[jec6613]

10 hours ago, Alex Atkin UK said:

A switch adds nowhere near that much.

...

A Gigabit switch as I recall adds about 0.1ms which is completely negligible.

A Cisco 2960X adds about 80 microseconds copper to copper, your topical off the shelf consumer switch has upwards of 200 microseconds of latency just from the copper port to the switch plane in each direction, and the lowest bidder switch chip, the cheap ones easily add 1-2 ms of latency.  You're not exactly using consumer equipment when you have NBaseT and are using SFP+ connections.  

[farzher]

my ping seems to be higher than it should be, no matter who i'm connecting to.
i've now tested a bunch of pings to a bunch of places using online free ping tests i can get random servers to ping me and my friends. my ping is always worse.
is at&t as an ISP just garbage? i never even knew they did internet, i thought they were only phones. maybe my internet is going over their wireless phone network? that's definitely what it feels like

[Alex Atkin UK]

8 hours ago, jec6613 said:

A Cisco 2960X adds about 80 microseconds copper to copper, your topical off the shelf consumer switch has upwards of 200 microseconds of latency just from the copper port to the switch plane in each direction, and the lowest bidder switch chip, the cheap ones easily add 1-2 ms of latency.  You're not exactly using consumer equipment when you have NBaseT and are using SFP+ connections.  

But even 1-2ms is negligible vs the path over the ISP.  You just know I'm going to have to dig out my random unmanaged cheapo switches now, as I don't remember them being that high.  I used to have them daisy-chained to save cable clutter.

 

Also, a lot of people use old routers - I only recently removed the last one in favour of a bridged 8 port appliance I was going to use pfSense on but decided to make it my dedicated torrent box doubling as a switch.