Gaming & speed

Why Is My Ping So High on Fast Internet? Find the Slow Hop

Why is my ping so high on a fast connection? Latency is not bandwidth. Read a traceroute hop by hop to spot Wi-Fi, bufferbloat, CGNAT or routing delay.

In this guide

Ping is high on a fast connection because latency and bandwidth measure different things: speed is how much data moves per second, ping is how long one packet takes to go and return. The delay is added at one point, usually Wi-Fi, a full router buffer, a long route, or a distant server. A traceroute shows which hop adds it.

Test it now

Cloudflare edge
IP websites see216.73.216.115
LocationColumbus, United States
Internet providerAnthropic, PBC

Read at the Cloudflare edge from this request. The homepage adds your provider, network type and the VPN, DNS and WebRTC checks.

Show details

Why is my ping so high when my internet is fast?

Bandwidth is a rate and latency is a duration, and a connection can score well on one while scoring badly on the other. A gigabit line moves a 1 GB file in seconds. A single 60-byte game update still has to cross every cable, radio link and router queue between you and the server, wait its turn at each one, and come back. None of those waits gets shorter because the pipe is wider.

The two also fail for different reasons. Bandwidth drops when a link is saturated. Latency rises when a packet is held somewhere: a Wi-Fi retransmission, a buffer behind someone else's upload, a route that detours through another city, or a distant server. The fix depends on where the hold-up is. For the list of fixes once you know the cause, see the "How to reduce ping" section of what ping is and what a good ping looks like.

How to read a traceroute hop by hop

A traceroute lists every router between you and a destination with the round-trip time to each one, so you can see at which position the time jumps. It sends probes with an increasing time-to-live. RFC 1812 requires that when a router reduces the TTL to zero it "MUST send an ICMP Time Exceeded message, Code 0 (TTL Exceeded in Transit) message to the source." The first probe expires at the first router, the second at the second, and each reply reveals one hop and the time it took.

Read the positions, not just the final number.

  1. Hop 1 is your own router. The time here is your Wi-Fi or Ethernet link alone. On a cable it should be a millisecond or two. If hop 1 shows tens of milliseconds, or varies wildly between probes, the problem is inside your home.
  2. Hop 2 or 3 is your ISP's edge. This is the ISP's first router (CMTS, BNG/BRAS); the cable node or fiber terminal in between does not decrement the TTL, so it never appears. The difference between hop 1 and this hop is your access link, usually the largest fixed cost.
  3. The middle hops are transit. These are backbone routers inside your ISP and the networks it hands off to. Times should climb gradually with distance; a jump of tens of milliseconds between adjacent hops means a long physical link, an ocean crossing for example, or congestion there.
  4. The last hop is the destination. If the final time is high but every hop before it was fine, the server or its network is far away or overloaded; only a different server region changes that.

Two cautions keep traceroute honest. First, the time shown for a middle hop is the time for that router to generate an error reply, a low-priority chore. RFC 1812 says a router "SHOULD also be able to limit the rate at which it sends other sorts of ICMP error messages (Destination Unreachable, Redirect, Time Exceeded, Parameter Problem)." So one hop that reads slow, or loses probes, while later hops look normal is not a bottleneck; it is a router that deprioritizes answering you while forwarding your traffic without delay. Only a slowdown that persists on every later hop is real. Second, the path back may differ from the path out, and traceroute only shows the outbound side.

What the asterisks mean

An asterisk means no reply arrived for that probe within the wait time. Microsoft's tracert reference explains that "some routers don't return time Exceeded messages for packets with expired TTL values and are invisible to the tracert command"; a row of asterisks is printed and the trace continues. The Linux traceroute manual adds: "Routers can throttle the rate of icmp responses, and some of replies can be lost." One starred row followed by normal rows is a silent router and harmless. Asterisks from some hop to the end mean a firewall is dropping probes or the path is broken, and the last answering hop tells you where.

Running it in both directions

On Windows 10 and 11 open Command Prompt and run tracert -d <server>; per Microsoft's reference, -d skips reverse-DNS lookups and the default is 30 hops with a 4-second wait per probe. On macOS and Linux run traceroute <server>, which per the traceroute(8) manual sends three probes per hop, stops at 30 hops and waits up to 5 seconds per probe. Use the address of the server you are lagging against.

The CheckMyIP traceroute runs the opposite direction: from the site's probe server in Frankfurt toward your public IP, one probe per hop, up to 30 hops, with the hostname and network of each router. If the probe reaches your ISP quickly and the last reachable hop is slow, the delay is in the access link. If the last hop never answers, your router is not replying to ICMP, which most home routers do not.

Where the delay hides, by hop position

Position in the traceWhat it usually isWhat a jump here points to
Hop 1Your router, reached over Wi-Fi or EthernetWi-Fi interference, distance, a congested 2.4 GHz band, or bufferbloat if it only happens under load
Hops 2–3The ISP's first router (CMTS, BNG/BRAS) or a CGNAT gatewayThe access technology's inherent latency, an oversubscribed node, or a CGNAT box
Middle hopsTransit inside the ISP and peering partnersA long physical link, a detour through a distant exchange point, or a congested peering link
Last hopThe game or application serverWrong region, a server far from you, or a busy host; only a different server fixes it

Jitter and bufferbloat: ping spikes under load

Jitter is the variation in ping from one packet to the next, and it is usually what players mean by ping spikes. Riot's connection troubleshooting guide defines it: "Jitter is the change in latency over time." A steady 60 ms feels better than a ping that swings between 25 and 150 ms, because the game cannot predict when your packets will land. Riot asks for jitter and packet loss as close to zero as possible, without naming a ping number.

The commonest cause of jitter on an otherwise good connection is bufferbloat. The Bufferbloat project defines it as "the undesirable latency that comes from a router or other network equipment buffering too much data." When one device starts a large transfer, the queue in your router or modem fills and every other packet, the game's included, waits behind it. The bandwidth number looks fine, because the queue keeps the link full; the ping climbs by tens or hundreds of milliseconds, because the queue is deep.

RFC 8289 describes the CoDel algorithm as one that "controls bufferbloat-generated excess delay in modern networking environments," and RFC 8290 (FQ-CoDel) separates flows so a bulk transfer cannot hold up small interactive packets; its default queue delay target is 5 ms against a 100 ms interval. Router firmware offering "SQM," "CAKE," "FQ-CoDel" or adaptive QoS implements this idea.

You can measure it directly. The connection quality test records idle latency to a Frankfurt server and the nearest Cloudflare edge, measures it again while downloading and uploading, and grades the rise from A to F (A is under 30 ms of added delay, F is over 200 ms). A good idle number with a poor under-load grade is bufferbloat; a poor idle number that improves on cable is the radio link.

Wi-Fi versus cable

Wi-Fi adds latency in a way a speed test hides: a collided frame is retransmitted, and a busy channel makes your device wait before it can send. Each event costs a few milliseconds, nothing to a download and everything to a 20 ms ping. Riot's guide says that "Wireless connections tend to be difficult to troubleshoot due to potential interference from the environment and other radio signals" and recommends switching to Ethernet whenever possible. NVIDIA's GeForce NOW requirements likewise recommend wired Ethernet or 5 GHz Wi-Fi.

The test is the ping tool, not the speed test. Open the ping test on the device you game on and run the browser-to-Frankfurt and browser-to-Cloudflare measurements: ten samples each, with median, minimum, maximum, jitter and loss. Plug in a cable and run it again. If the minimum barely moves but the maximum and jitter collapse, Wi-Fi was the cause; if both runs match, the delay is beyond your router. The page's third measurement sends five ICMP pings from Frankfurt to your public IP, which many home routers ignore; no reply there is normal.

CGNAT and routing distance

Carrier-grade NAT adds a translation step and often a detour. An ISP short of IPv4 addresses puts many customers behind one shared public address at a gateway inside its network. RFC 6598 reserves 100.64.0.0/10 for this, so an address in that range on your router's WAN side means you are behind CGNAT. The gateway may sit far from your city, so traffic detours there first; in a traceroute that is an early hop with a larger-than-expected jump. The NAT type test reports whether a CGNAT range is in the path and what that does to console NAT types.

Routing distance is the other cost a speed test cannot show. Packets follow the agreements between networks, not the map, and two ISPs in the same city may exchange traffic in another country. Middle hops carrying airport codes or city names from a distant exchange point show that detour; no setting changes it.

Distance applies on the server side too. Sony's Remote Play page, which otherwise specifies only bandwidth (5 Mbps minimum, 15 Mbps recommended), notes that "Increased distance between the PS5 or PS4 console and the remote TV may increase latency." That footnote concerns streaming from your own console; it is our inference, not Sony's claim, that the same distance cost applies to any game server. Pick the region whose traceroute ends soonest, not the one the matchmaker chose.

What "good" ping is, per use

Good ping depends on what the application does with each packet; these figures come from vendors that publish them.

  • Competitive games. Checkmyip.com's ping guide treats under 20 ms as excellent and 20 to 50 ms as good for any online game, with shooters and fighting games feeling unresponsive above 100 ms. Riot publishes no ping number.
  • Cloud gaming. NVIDIA's GeForce NOW requirements demand less than 80 ms of latency from an NVIDIA data center, next to a bandwidth minimum of 15 Mbps for 720p at 60 FPS. Neither number substitutes for the other.
  • Voice and video calls. Microsoft's Teams network requirements target under 100 ms round trip, under 30 ms jitter and under 1 percent packet loss from the client to Microsoft's edge over the whole path including Wi-Fi; from the network edge the targets tighten to 60 ms, 15 ms and 0.1 percent.

Ping is a cost, not a quality. Downloads and streaming tolerate several hundred milliseconds; anything interactive does not.

Troubleshooting order

Work from the hop you control toward the one you do not; each step finds the cause or rules out a section of the path.

  1. Run the ping test on the affected device while the problem is happening. Record median, maximum and jitter.
  2. Connect the same device by Ethernet and run it again. If jitter and maximum fall sharply, the cause was Wi-Fi: move the device or router, prefer 5 GHz, or stay wired.
  3. Run the connection quality test wired. A poor bufferbloat grade with good idle latency means your router queue is the problem: enable SQM, CAKE, FQ-CoDel or adaptive QoS if the firmware offers it.
  4. Run tracert -d (Windows) or traceroute (macOS, Linux) to the server you lag against. Find the first hop where the time rises and stays risen on every hop after it.
  5. If that hop is 1, go back to steps 2 and 3. If it is 2 or 3, the delay is your access link or a CGNAT gateway; check the NAT type test for a 100.64.x address and repeat the trace at a quiet hour.
  6. If the jump is in the middle hops, the route is the cause. Run the CheckMyIP traceroute to see the inbound path; if both directions detour, change server region or time of day, and report the hop to your ISP if it is inside their network.
  7. If only the last hop is slow, the server is far away or busy. Select a closer region and confirm with another traceroute.

Keep the before-and-after numbers; a fix that cuts jitter from 40 ms to 5 ms has solved the problem even if the median barely moved.

Sources

Questions

What causes high ping?

Distance to the server, Wi-Fi retransmissions and interference, a router buffer that fills while something uploads or downloads (bufferbloat), an ISP route that goes the long way around, carrier-grade NAT, or a busy game server. A traceroute shows the hop where the round-trip time jumps, which points to the cause.

Is high ping good or bad?

Bad for anything interactive. Ping is delay, so lower is always better. Checkmyip.com’s ping guide treats under 20 ms as excellent and 20 to 50 ms as good for games; Microsoft’s Teams targets ask for under 100 ms round trip. Downloads and streaming barely notice ping; shooters, fighting games and calls do.

How do I fix high ping?

Find the hop first. If the first hop is slow, use Ethernet or move closer to the router. If ping rises only while someone streams or uploads, enable SQM or QoS on the router. If a middle hop or the server is slow, pick a closer game region or test another time; the ISP route is not yours to change.

Why does my ping spike?

Spikes are jitter: latency that changes from packet to packet. The usual causes are Wi-Fi retransmissions, a router buffer that fills when another device transfers data, and background updates on your own machine. Run a ping test while the spikes happen and compare wired to wireless; the one that stays flat tells you which it was.