Internet Latency: What Is It, How to Measure It, and How to Improve It

Internet latency, the often-overlooked delay between sending and receiving data, can mean the difference between a flawless video conference and a frustrating, glitchy mess. Measured in milliseconds (ms), these microscopic delays accumulate, creating tangible performance issues across all online activities.

For businesses and individuals, Internet latency affects:

• Video calls & VoIP – High latency causes choppy audio and lag.
• Online gaming – Real-time responsiveness depends on low latency.
• Cloud applications – Slow response times hurt productivity.
• Web browsing & streaming – High latency means longer load times.

Let's begin by understanding what's really happening when data travels across the Internet — and why distance isn't just geographic, but digital.

At its core, latency is the time it takes for data to travel from one point to another in a network. However, not all latency is the same — it varies depending on whether data moves within a private network or over the public Internet.

While people often mix them up, Internet latency and network latency aren’t quite the same thing. Think of network latency like traffic delays on any kind of road, whether it’s your local streets, a highway, or a private driveway. Internet latency, on the other hand, is just the delay on the highways, the public Internet. So all Internet latency is network latency, but not all network latency happens on the Internet.

Network latency refers to delays that occur across any type of network, including:

• Local Area Networks (LANs) – Office networks, home networks.
• Wide Area Networks (WANs) – Connections between different office locations.
• Data Center Networks – Communication between servers.

What Contributes to Network Latency?

1. Hardware Delays – Routers, switches, and firewalls add processing time.
2. Transmission Medium – Fiber optics are faster than copper cables.
3. Internal Traffic Congestion – Too many devices sharing bandwidth.
4. Protocol Overhead – Encryption (VPNs, TLS) and packet processing add delay.

Example:

If you transfer a large file between two computers in the same office, the delay is network latency, which is affected by your internal switches, cables, and local traffic.

What is Latency: The Hitchhiker’s Guide

We asked a supercomputer “What is latency”, its impact on network performance, and strategies for minimizing it and created this comprehensive guide.

Learn more

Internet latency is specifically the delay introduced when data travels over the public Internet. Unlike a controlled LAN environment, the Internet introduces extra variables:

Key Factors Affecting Internet Latency:

1. Physical Distance – Data travels at about ⅔ the speed of light in fiber optics, so New York to Sydney will always have higher latency than New York to Chicago.
2. ISP Routing Efficiency – Some ISPs take longer paths due to peering agreements.
3. Public Internet Congestion – Peak hours slow down shared networks.
4. Last-Mile Connection Quality – Poor Wi-Fi or old copper lines increase latency.
5. Peering Points & Interconnections – Traffic bottlenecks at ISP handoff points.

An Example of Internet Latency:

If you’re on a Zoom call between New York and London, the lag you experience is Internet latency, shaped by ISPs, undersea cables, and global routing.

• Troubleshooting: If your corporate VPN is slow, is it your office network (internal latency) or the Internet (external latency)?
• Optimization: Reducing LAN latency may involve upgrading switches, while improving Internet latency might require a better ISP or CDN.
• Performance Expectations: A game server hosted in your city may have 5ms latency, while an overseas server could have 150ms+ due to Internet routing.

Internet latency is measured by calculating the time it takes for data to travel from your device to a destination server and back. This is commonly referred to as Round-Trip Time (RTT), which we discuss in this section. The result is expressed in milliseconds (ms), and lower numbers indicate faster, more responsive connections.

That being said, there are a variety of other metrics that are included in this calculation, and each plays a pivotal role in understanding how fast (or slow) data travels through the Internet.

As we mentioned above, the most common way to test Internet latency is Round-Trip Time (RTT), which tracks how long it takes for a data packet to travel from your device to a server and back. This is what tools like ping measure, and it’s expressed in milliseconds (ms).

While RTT is easy to test, it only gives you the total delay, not where the slowdown happens. For example, if your RTT to a cloud server is 150ms, that could mean a fast connection to your ISP but a slow route beyond it, or vice versa. Businesses often use RTT to check general connectivity, but deeper analysis is needed for precise troubleshooting.

Unlike RTT, One-Way Latency measures the delay in just one direction, from sender to receiver. This is crucial for applications like VoIP and video streaming, where upload and download paths can behave differently.

However, measuring one-way latency accurately requires synchronized clocks between endpoints, making it more complex than a simple ping test. For instance, a company using a cloud-based phone system might find that calls sound clear in one direction but choppy in the other, indicating an asymmetrical latency issue that RTT alone wouldn’t reveal.

Even if your average latency looks good, inconsistent delays, known as jitter, can ruin real-time applications. Jitter measures the variation in latency over time, and high jitter leads to problems like robotic VoIP audio, frozen video calls, and lag in online gaming.

For businesses, this is especially critical for unified communications platforms (e.g., Microsoft Teams, Zoom), where a stable connection matters more than raw speed. A network might have an average latency of 50ms but jitter spikes of 200ms, causing intermittent disruptions that frustrate users.

When it comes to web applications and APIs, Time-to-First-Byte (TTFB) is a key metric. It measures how long it takes for a server to start sending data after receiving a request. A high TTFB (e.g., over 500ms) can slow down entire web applications, even if bandwidth is plentiful.

For e-commerce sites, a delay here directly impacts revenue, studies show that every 100ms increase in load time can reduce conversions by 7%. TTFB is influenced by server processing speed, database queries, and network latency, so fixing it often requires backend optimizations.

A single speed test or ping check only gives you a snapshot of your network’s performance, like checking the weather at one moment and assuming it won’t change. In reality, latency fluctuates constantly due to:

• Network congestion (peak business hours, large file transfers)
• ISP throttling (artificial slowdowns during high-traffic periods)
• Background processes (backups, updates, cloud syncs)
• Routing changes (ISP traffic shifts, undersea cable issues)

Without continuous monitoring, intermittent latency spikes go unnoticed until users complain, costing businesses productivity, revenue, and reputation.

Unlike traditional tools that rely on sporadic ping checks or synthetic traffic, Obkio’s Latency Monitoring tool provides real-time, end-to-end visibility using lightweight agents deployed across your network. Here’s how it differs:

1. Continuous Synthetic Traffic (No Guesswork)

• Simulates real application traffic (TCP/UDP) between endpoints
• Measures actual latency, jitter, and packet loss — not just ICMP pings
• Tracks performance every 500ms to catch micro-spikes

2. Pinpoint Latency Spikes to the Exact Network Segment

• ISP vs. internal issues: Instantly see if latency originates from your LAN, ISP, or cloud provider
• Historical baselining: Compare current performance against normal behaviour

3. Proactive Alerts Before Users Notice

• Get notified when latency exceeds thresholds, before calls drop or apps lag
• Correlate latency with business events (e.g., "Zoom lags every day at 2 PM during backups")

4. Enterprise-Grade Without the Complexity

• No hardware required: Deploy software agents in minutes
• Works across hybrid networks: Offices, data centers, clouds, home networks

Every network administrator knows the frustration of latency issues - those mysterious delays that make VoIP calls choppy, cloud applications sluggish, and video conferences unbearable. Unlike bandwidth problems that are easily measured, latency issues often hide in plain sight, only revealing themselves during critical operations.

Let's examine the most common causes through real-world scenarios that every IT professional will recognize.

Data travels at about two-thirds the speed of light in fibre optic cables, making physical distance an unavoidable factor. That Australian branch office connecting to your Frankfurt data center will always face higher latency than your Chicago location.

We've all seen the tickets: "SAP responds slowly for APAC users" or "Teams calls lag between our Asian and European offices." While we can't break the laws of physics, solutions like edge computing deployments or latency-optimized cloud services (AWS Global Accelerator, Azure Front Door) can help minimize the impact.

Just like city traffic at 5 PM, network congestion creates frustrating delays. Consider the classic scenario: your morning latency checks show a healthy 30ms to Office 365, but by mid-afternoon, VoIP calls start breaking up. Investigation reveals that scheduled backups are flooding the WAN link.

This is where Quality of Service (QoS) policies become essential - prioritizing voice and video traffic while scheduling large transfers for off-peak hours can make all the difference in user experience.

Each network hop typically adds 1-10ms of latency, and poor ISP routing decisions can send traffic on unnecessary detours. Remember that time when traceroute showed your traffic taking 8 hops to reach Azure East while a competitor's path used just 4? The culprit was often a congested peering point that your ISP insisted on using.

This is where multi-ISP strategies and SD-WAN solutions prove their worth, allowing traffic to take the most efficient path available.

We've all encountered those "it still works" pieces of equipment that somehow become sacred cows in the network. The 8-year-old firewall that adds 80ms of latency to VPN connections or the "gaming" router in Accounting that chokes under the load of 50+ Zoom calls.

Enterprise-grade hardware refreshes and regular performance testing (using tools that provide hop-by-hop analysis) are often the only solutions to these stealthy latency creators.

Wi-Fi presents unique latency challenges that wired networks don't face. Interference from microwaves, Bluetooth devices, and neighbouring networks can wreak havoc, as can the retransmission of lost packets.

Who hasn't dealt with the CEO complaining about daily Zoom freezes at lunchtime, only to discover the break room microwave was murdering the 2.4GHz band? Or warehouse barcode scanners missing updates because cheap access points couldn't handle the device load? The solution often lies in strategic use of wired connections for stationary devices and proper enterprise Wi-Fi deployments with band steering to prefer 5GHz connections.

Last-mile issues and ISP problems represent the most frustrating category because they're often outside our direct control.

That remote office where latency spikes to 500ms every evening? The ISP finally admits to node congestion. The rural teleworkers struggling with 10% packet loss on aging DSL lines? There's only so much we can do internally.

This is where business-class Internet with SLAs and continuous monitoring to hold providers accountable become essential.

Our security measures sometimes become our own worst enemies. That new next-gen firewall added 120ms of latency to cloud applications, and enabling TLS inspection broke a legacy ERP system's timeouts. We walk a constant tightrope between protection and performance, often needing to create careful exceptions for latency-sensitive traffic like VoIP or financial trading applications.

The reality of network administration is that high latency rarely stems from a single cause. It's typically two or three of these factors combining in unexpected ways. This is why establishing performance baselines, continuous monitoring, and methodical testing are so critical.

Before jumping to conclusions about any latency issue, we need to ask: Is this wired or wireless? Internal or ISP-related? Constant or intermittent? Only with proper data can we implement the right solutions.

As any seasoned network administrator knows, properly measuring latency requires more than just running a quick speed test. Different tools serve different purposes, and choosing the right one can mean the difference between spotting a real issue and chasing false leads. Let's walk through the essential toolkit for comprehensive latency measurement.

Every network professional's first line of defence is the classic command-line tools we've all used countless times.

How to Measure Internet Latency with Ping:

The ping command sends a small packet of data (called an ICMP echo request) to a target IP address or domain and waits for a response. The time it takes to receive the response is the latency.

The humble ping command gives you that immediate round-trip time (RTT) measurement that's so useful for basic connectivity checks. But as we've all learned the hard way, while ping is great for "is it up?" checks, it doesn't tell the whole story.

How to Measure Internet Latency with Traceroute (or tracert):

That's where traceroute (or tracert on Windows) comes in. Traceroutes breaks down the latency between each "hop" (router or server) between you and the destination. Useful for identifying where delays are occurring along the path.

Remember that time when all your European offices were experiencing terrible latency to your cloud provider? Traceroute showed the traffic taking a bizarre detour through an overloaded peering point halfway across the world. These tools are like the stethoscope in a doctor's toolkit - basic but essential.

For more advanced path analysis, Linux users often turn to mtr, which combines ping and traceroute functionality with statistical analysis. It's particularly useful for identifying intermittent packet loss that might be causing latency spikes.

We've all had users send us Speedtest.net results as "proof" that their connection is fine. While these tools can provide a quick snapshot of your connection to a nearby test server, they have significant limitations that every IT pro should understand:

• They typically measure only the last mile to your ISP
• Results vary wildly based on which test server you select
• They don't reflect real-world application performance
• They miss intermittent issues that occur between tests

That said, when used properly (comparing multiple servers over time), they can help identify last-mile issues, like when we discovered a branch office's "slow Internet" was actually due to a failing DSL line that only showed problems during peak hours.

How to Measure Latency

Learn how to measure latency with Obkio’s Network & Latency Monitoring tool. Check for latency in your network & analyze latency measurements.

Learn more

For enterprise-grade latency monitoring, tools like Obkio provide the continuous, end-to-end visibility that ping and speed tests can't match. Unlike basic tools that give you a single point-in-time measurement, Obkio's approach mirrors how your actual applications experience the network:

• Lightweight agents deployed across your network (offices, data centers, cloud)
• Continuous synthetic traffic that behaves like real application data
• Every 500ms measurements to catch those elusive micro-spikes
• Hop-by-hop analysis to pinpoint exactly where delays occur

Remember that VoIP quality issue that kept happening every afternoon? Ping tests showed nothing unusual, but Obkio's monitoring revealed jitter spikes correlating with backup jobs - something we'd never have caught with manual testing.

As with any troubleshooting, the right tool depends on what you're trying to accomplish:

• Quick connectivity check? Ping is your friend
• Routing issues? Traceroute or mtr
• Last-mile problems? Speedtest can help
• Enterprise monitoring? You need Obkio's continuous visibility

The key is understanding that latency isn't a single number - it's a complex behaviour that changes throughout the day, across different paths, and under varying network conditions. Proper measurement means looking at both the big picture and the fine details.

For a deeper dive into measurement techniques, our comprehensive guide on how to measure latency covers advanced methodologies and real-world case studies that every network admin will find valuable.

Every network administrator has faced that moment when a user asks, "Is my latency good?" - and we need to give an answer that's both technically accurate and practically useful. But what is a good Internet latency benchmark? The truth is, "good" latency depends entirely on what you're trying to accomplish.

When testing what is a good Internet latency for your needs, consider:

Online Gaming (Sub-50ms)

For esports and fast-paced multiplayer games, every millisecond counts. Professional gamers look for low latency Internet under 20ms, while casual players can tolerate up to 50ms. Beyond this threshold, you start seeing noticeable disadvantages in reaction times. Remember that tournament where players complained about "unfair lag"? That was likely a case of some players connecting at 15ms while others struggled with 80ms connections.

Voice and Video Conferencing (Under 100ms)

In our remote work era, VoIP quality makes or breaks meetings. The ITU recommends under 150ms for acceptable voice quality, but we've found that keeping it under 100ms prevents those awkward "you go first" moments in conversations. That executive who keeps complaining about robotic audio during important client calls? They're probably experiencing latency spikes right at that 100ms threshold.

General Web Browsing and Streaming (Up to 150ms)

While streaming services buffer content, latency still affects startup times and quality switching. Google's research shows users start abandoning sites at 150ms delays. Remember when the marketing team complained their demo videos took too long to start playing? That was likely a latency issue rather than bandwidth.

Cloud Applications and SaaS (Under 100ms)

Modern cloud apps expect snappy responses. Latency above 100ms makes CRMs, ERPs, and collaboration tools feel sluggish. That sales team complaining about Salesforce being "slow today"? Check their latency to the cloud instance before blaming the application.

Here's what many teams miss: average latency tells only part of the story. We've all seen networks with "good" average latency (say 45ms) that still have terrible user experiences because of jitter (latency variation). Consider these real scenarios:

1. A VoIP system shows 80ms average latency but has 300ms spikes every few minutes, causing dropped words and frustration
2. A trading platform maintains 10ms latency 95% of the time, but those 5% spikes to 150ms result in failed transactions
3. A video conferencing system works perfectly until the 3 PM backups start, introducing unpredictable delays

This is why tools like Obkio that track latency consistency over time are so valuable. They help identify these fluctuation patterns that single-point tests miss completely.

When assessing your network's latency health:

• Establish baselines for each critical application
• Monitor consistency, not just averages
• Correlate latency with business activities (backups, peak hours, etc.)
• Compare paths - different ISPs, wired vs wireless, VPN vs direct

For a deeper technical dive into optimal latency thresholds and troubleshooting approaches, our comprehensive guides on what is good latency and the causes of high latency provide network teams with detailed benchmarks and resolution methodologies.

Remember - in the world of network performance, context is everything. The same 100ms latency that's unacceptable for competitive gaming might be perfectly fine for email. The key is matching your latency profile to your actual business requirements.

After diagnosing latency issues, the real work begins - implementing effective solutions. Based on countless network optimizations we've performed, here are the most impactful strategies that actually move the needle on latency reduction.

Every network engineer knows the first question to ask: "Are they on Wi-Fi?" Wireless connections inherently introduce more latency than wired ones due to:

• Signal interference (microwaves, Bluetooth, neighboring networks)
• Medium contention (multiple devices sharing airtime)
• Retransmissions from packet loss

Actionable Fix:

• Run Ethernet to all stationary workstations (especially for VoIP phones and trading desks)

• For unavoidable Wi-Fi:

  1. Prefer 5GHz over 2.4GHz (less interference)
  2. Implement band steering
  3. Reduce transmit power to minimize co-channel interference

That "perfectly good" 7-year-old router might be your latency culprit. Common equipment issues include:

• Underpowered CPUs struggling with encryption/VPN traffic
• Small packet buffers cause drops during congestion
• Outdated firmware with inefficient traffic handling

Pro Tip:

Use monitoring tools to measure latency hop-by-hop. We once found a "fully functional" switch adding 80ms latency to VoIP traffic due to a faulty GBIC module.

Background processes often sabotage latency without anyone realizing it:

• Cloud backups running during business hours
• Windows updates are consuming bandwidth
• Personal streaming on work networks

Solution:

Implement application-aware traffic shaping:

1. Identify latency-sensitive apps (VoIP, trading platforms)

2. Create QoS policies to prioritize them

3. Schedule large transfers for off-hours

DNS lookups add latency before any data transfer begins. Problems compound when:

• Using default ISP DNS servers (often overloaded)
• Having geographically distant resolvers
• Lacking proper caching

Quick Wins:

• Deploy local caching resolvers (like Unbound)
• Use performant public DNS (Cloudflare 1.1.1.1, Google 8.8.8.8)
• Implement DNS-over-HTTPS for security without latency penalty

Not all Internet connections are created equal. For latency-sensitive operations:

• Business-class fibre beats consumer cable/DSL
• Some ISPs have better peering arrangements
• Multi-WAN setups provide failover and optimal routing

Real-World Example:

A trading firm reduced latency to Chicago markets by 40ms simply by switching from a national ISP to a low-latency specialized provider.

A Guide to Troubleshooting and Improving Network Latency

In this guide, learn how to troubleshoot and improve network latency with fun analogies, step-by-step instructions, and tips for both users and businesses.

Learn more

Proper Quality of Service configuration acts like a traffic cop:

• VoIP/VTC: Highest priority (DSCP EF/CS5)
• Trading apps: Second tier (AF41/CS4)
• Web browsing: Best effort
• Backups: Lowest priority

Critical Note:

QoS only helps when there's congestion - it doesn't reduce base latency.

Content Delivery Networks bring resources closer to users:

• For public websites: Cloudflare, Akamai, Fastly
• For internal apps: Enterprise CDNs like Azure Front Door
• For video: Specialized providers (Limelight, AWS MediaConnect)

You can't optimize what you can't measure. Effective latency reduction requires:

• Continuous baseline monitoring
• Alerting on threshold breaches
• Historical data for trend analysis

Obkio's Approach:

1. Lightweight agents deployed across network edges
2. Synthetic traffic mimicking real applications
3. Hop-by-hop latency visualization
4. Proactive alerting before users notice

For teams serious about Internet latency reduction:

Remember: Latency optimization is an ongoing process, not a one-time fix. The Internet changes constantly - new routes, new congestion patterns, new applications. What works today may need adjustment tomorrow, which is why continuous monitoring is so critical.

At its core, Obkio is a network performance monitoring solution designed by network engineers for network engineers. Unlike traditional tools that simply tell you "there's a problem," Obkio provides the end-to-end visibility needed to actually understand and fix latency issues. The platform takes a unique approach by combining continuous monitoring with synthetic testing - giving you the complete picture of how your network truly performs.

• 14-day free trial of all premium features
• Deploy in just 10 minutes
• Monitor performance in all key network locations
• Measure real-time network metrics
• Identify and troubleshoot live network problems

Obkio deploys lightweight agents across your network infrastructure - in branch offices, data centers, cloud environments, and even employee home networks. These agents don't just ping occasionally; they maintain constant communication, measuring performance every 500 milliseconds.

This granular approach catches those fleeting latency spikes that other tools miss - the ones that cause VoIP dropouts or trading platform timeouts but disappear before you can troubleshoot them.

The synthetic traffic generated between agents behaves exactly like your real business applications. Whether it's mimicking VoIP packets, replicating SaaS application traffic, or testing video conference streams, Obkio measures how your actual services experience the network - not just how ICMP packets perform. This is crucial because, as every network pro knows, different applications face different latency profiles.

When latency issues occur, Obkio doesn't just tell you "there's a problem" - it shows you exactly where the problem is happening. Obkio Vision, a visual traceroute tool, hop-by-hop visualization maps the entire path between any two points in your network, identifying:

• Which specific network segment is introducing the delay
• Whether the issue is in your LAN, WAN, ISP network, or cloud provider
• How do different routes compare in real-time performance

This eliminates the endless back-and-forth between teams trying to determine where latency originates. We've seen cases where:

• A faulty switch in a branch office was adding 80ms latency to all cloud traffic
• An ISP's routing change introduced 200ms spikes during peak hours
• A cloud provider's regional interconnect became congested every afternoon

Obkio's smart alerting system notifies you about latency issues before they impact users. Customizable thresholds allow you to:

• Get immediate alerts when latency exceeds acceptable levels for specific applications
• Receive warnings about developing trends before they become critical
• Correlate latency spikes with network changes or business events

The platform maintains historical data that's invaluable for:

• Demonstrating SLA compliance to management
• Proving to ISPs that latency issues exist on their network
• Planning capacity upgrades based on actual usage patterns

For network teams serious about latency optimization, Obkio provides the continuous, application-aware monitoring needed to:

1. Establish performance baselines
2. Identify issues before users notice
3. Pinpoint exactly where problems occur
4. Verify the effectiveness of fixes
5. Maintain optimal performance over time

In today's hybrid work environment, where latency directly impacts productivity and revenue, having this level of visibility isn't just helpful - it's essential for any organization running latency-sensitive applications across distributed networks.

Throughout this guide, we've explored how latency impacts everything from VoIP calls to cloud applications, and more importantly, how to measure, analyze, and ultimately reduce it.

1. Latency varies by application - While sub-50ms is ideal for gaming, under 100ms makes the difference between smooth and frustrating video conferences

2. Consistency matters as much as speed - A network with 80ms average latency but 300ms spikes creates more problems than a stable 100ms connection

3. Troubleshooting requires the right tools - Basic ping tests miss the micro-spikes that disrupt real-time applications

4. Optimization is ongoing - Network conditions change constantly, requiring continuous monitoring rather than one-time fixes

Waiting for user complaints is no longer an acceptable strategy. The most effective network teams

1. Establish baselines for normal performance across all critical applications

2. Monitor continuously to catch issues before they impact operations

3. Validate improvements with data rather than assumptions

1. Test your current latency using both basic tools (ping, traceroute) and professional monitoring solutions (We’ll let you discover the best solution. Hint: It’s not just another ping tool.)

2. Identify your most latency-sensitive applications and their performance requirements

3. Implement at least one optimization from this guide, whether it's QoS policies, DNS changes, or equipment upgrades

4. Make monitoring part of your routine, because in networking, what gets measured gets improved