Confused by buzzwords? You’re not alone.
EXPLAINED is a new explainer series for people who work with technology – but don’t want marketing fluff or academic theory.
We break down complex digital topics clearly and practically: what they mean, why they matter, and how they work in the real world.
Latency has a reputation problem.
In technology conversations, “latency” often sounds like a polite word for “slow”. Mention that a piece of software runs on a server and not directly in the browser, and many immediately imagine keystrokes making long and slow journeys across the internet, before even trying out the piece of software in question.
But before deciding whether something is “fast” or “slow”, it helps to ask a better question: fast or slow compared with what?
What is latency?
Latency is the time it takes for something to happen after you do something else. It applies everywhere – the stopping distance of a car, the time it takes your brain to respond to a stimulus, or the fundamental cause and effect speed limit of the universe ~300,000 km/s. Your speech has latency, but we don’t tend to notice the speed of sound in conversations. It is something you worry about if you’re designing a concert hall however.
Similarly with computers, when you press a key, move a mouse, tap a touch screen, or send an “instant” message, there is some delay before the result appears. The question is whether it’s fast (or slow) enough to matter.
Fast and slow
A Formula 1 driver may react in around 0.2 seconds – that’s 200 milliseconds (ms). The average human reaction time is closer to 250ms. (Interestingly in athletics, a reaction quicker than 100ms after the starting signal is treated as a false start!)
Human-computer interactions have similar thresholds. Studies have found computer system response times of around 100 milliseconds generally feel instantaneous to users. That’s about the same time, or slightly quicker than a human blink, roughly 100-150ms. So that’s pretty fast! Or is it actually quite slow?!
Displays introduce their own timing too. At the cinema, a 24 fps film delivers you a new frame every 42ms, a modern 60Hz monitor refreshes every 17ms and a 120Hz gaming monitor refreshes every 8.3ms. If you’re frantically shooting zombies those tens of milliseconds might matter, although monitor sales suggest many gamers aren’t convinced.
Your keyboard has latency too. A Bluetooth keyboard can add 20-30 ms before a keystroke reaches the computer, something most of us have never considered. Again – some gamers buy wired or lower-latency wireless peripherals to mitigate this, many aren’t aware of the apparent lag.
The internet in contrast, is blindingly fast. Current WonderNetwork ping data shows London-to-Rome round trip takes about 20ms, with London-to-Frankfurt taking closer to or less than 10ms. In other words, sending data from London to another European city and back can be faster than going from your keyboard to your PC, and 10x faster than you can blink.
40ms Worst Case Scenario?
To test this theory, we wrote a quick script to measure the response time from a keystroke being sent to the letter appearing on the screen. We then asked one of our developers in India to test it with a Europe-based server.
They measured a 42.8ms average round-trip, with the median of 29.6ms – essentially the same as plugging in a Bluetooth keyboard.
The worst response time recorded in the test at 166ms was still much faster than the gap between keystrokes at normal typing speed, and well within even the F1-driver response time mentioned above.
From Rajkot, India, to Europe and back, a total of ~14,000 km averaging 43ms. Not bad right?
Document editing vs e-sports
Some activities are extremely sensitive to latency. Competitive gaming is the obvious example: if you’re aiming, reacting, and moving in real time, tens of milliseconds might make a difference. Professional audio is another. If a musician hears themselves delayed, even slightly, it can become distracting very quickly.
Document editing is a little different.
A typical typing speed of 40 words per minute (wpm) works out at roughly 200 characters per minute, or one character every 300ms. A professional 60 wpm worker is roughly one character every 200ms. Clearly, typing and the speed of ‘the internet’ are in completely different categories.
When you type a sentence, adjust a table, change a heading style, or update a spreadsheet cell, the interaction still needs to feel responsive – but it does not require the same latency budget as a first-person shooter or a live performance.
And even many of the latency-critical cases are moving to the server. Cloud gaming services run the game in a data centre, streaming the frames to the player, and send the controller input back the other way. It isn’t the choice of professional e-sports players, but the fact that the model works at all – for first-person twitch-reaction shooters no less, says a lot about what modern networks can carry.
If a remote server can run a first-person shooter, it can handle your quarterly report.
What about editing documents on a server?
This is where Collabora Online sometimes surprises people.
Unlike less security-conscious editors that do document processing directly in the browser, Collabora Online keeps the document on the server and renders it there. That might sound like a slower solution, and if you include the additional 15ms round trip time to update the document, it is, but in practice, the round-trip time between user and server barely factors into the combined latency of the document editing system.
The advantages on the other hand, are that there is just one document that every works on at the same time, users get the same powerful document engine wherever they work, independent of their device or browser, integrators get a consistent platform to build around, IT teams get a centralised deployment model that is easier to manage, scale, update, and secure, and organisations keep documents where they belong – under their chosen governance, infrastructure, and compliance controls.
Keeping things close
Of course, lower latency is still better. That’s why we generally recommend deploying document servers close to users where practical. If most users are in London, there is no need to make them edit documents in Australia.
For larger international installations, the Collabora Online Kubernetes Controller can route users intelligently to the most appropriate server location. This allows organisations to operate multiple regional deployments while keeping editing responsive and efficient.
But the bigger point is this: server location should be treated as a sysadmin choice, not a reason to dismiss server-based editing altogether.
Latency is something to design for, measure, and optimise – just like capacity, storage, security, or uptime. With the right architecture, users connect to the right server, documents open quickly, and the infrastructure does its job quietly in the background.
And even when the server isn’t next door, modern networks are often fast enough that they don’t become the limiting factor people imagine.
So what’s the point?
Latency is everywhere, but it is not automatically a problem.
With good infrastructure, sensible server locations, and intelligent routing, Collabora Online delivers responsive editing while keeping document processing on the server. That gives organisations the benefits of centralised, consistent document rendering.
So why does the web feel slow?
It is faintly scandalous that an average web page still takes two seconds to appear when a packet can cross Europe in ten milliseconds. But that delay isn’t the network’s fault: it’s hundreds of round trips, layers of handshakes, and megabytes of JavaScript, each dependency politely waiting its turn. Two seconds is a software problem, not a physics problem.
A live editing session in Collabora Online is a different animal. The WebSocket connection is opened once and stays open – no repeated handshakes, no page loads, just messages travelling in both directions at whatever speed the network can manage.
While our own wet-ware hasn’t improved its latency in millennia, networks have, and continue to do so. More bandwidth and lower latency have been a reliable bet for thirty years, and there’s little sign of that changing.
Try lightning-fast server-based editing
The fastest way to understand the difference is to try it.
Collabora Online combines secure, server-based document processing with a responsive and powerful browser editing experience, giving users fast, consistent access to documents across devices and environments.
Try Collabora Online today and experience lightning-fast server-based document editing for yourself.
