Bandwidth vs latency: why a fast connection can still feel slow
The two network measures that are most often confused, and how to tell from the symptom which one is the problem.
What are bandwidth, latency, throughput and jitter?
- Bandwidth: the maximum rate a link can carry, in bits per second (Mbps, Gbps). A wider pipe.
- Throughput: the rate actually achieved, which is lower than the bandwidth because of overhead, congestion, errors and the application itself.
- Latency: the time a piece of data takes to reach the other end, in milliseconds. Often given as round-trip time (RTT): there and back, which is what a request and its answer cost. A longer pipe.
- Jitter: the variation in latency from one packet to the next. It matters most for live voice and video.
- Packet loss: data that never arrives; TCP traffic sends it again, which adds delay, while live voice and video simply lose it (gaps, frozen frames).
Latency comes from distance (signals travel at a finite speed: a geostationary satellite link alone adds roughly half a second to a round trip), from every router and device on the way, and from queues when the link is busy.
How do bandwidth and latency differ?
| Bandwidth | Latency | |
|---|---|---|
| Measures | Capacity: how much per second | Delay: how long one trip takes |
| Unit | Bits per second (Mbps, Gbps) | Milliseconds (ms) |
| Hurts most | Large transfers: images, backups, video, software updates | Many small back-and-forth exchanges: chatty client-server applications, database queries, voice |
| Symptom | A big download takes long; slows when many users share the link | Every click waits, even when little data moves and few people are online |
| What improves it | A bigger link, compression, fewer competing transfers | Shorter distance, fewer hops, a better link type, fewer round trips |
How do they look on two examples?
Made-up figures, for illustration only.
- A CT study of 400 megabytes sent over a 100 Mbps link: 400 × 8 = 3,200 megabits, ÷ 100 = about 32 seconds at best. Doubling the bandwidth roughly halves that. Latency barely matters here, because the time goes into moving a lot of data.
- A remote clinic using the hospital's client-server EHR through a link of the same 100 Mbps, but with a 300 ms round trip: if opening a patient's chart needs 40 round trips between the screen and the server, that is 40 × 0.3 = about 12 seconds of waiting, while the data itself would take a fraction of a second. Doubling the bandwidth changes almost nothing.
Mind the units: network speeds are in bits (Mb), file sizes in bytes (MB), and one byte is eight bits.
How can the problem be told from the symptom?
- Slow at busy hours, fine at night, and worst for big files: the link is full; bandwidth (or congestion).
- Every screen pauses even with one user and little data, at a distant site or over satellite: latency.
- Voice breaks up or video freezes on calls although the average speed is good: jitter or packet loss.
- Slow only in one application while others are fine: often the application (chatty design, an overloaded server or database), not the network.
A speed test measures bandwidth; a ping or traceroute shows the round-trip time and where on the path it grows.
What helps with each?
- For bandwidth: a larger link, compression, quality of service (QoS) that gives clinical traffic priority over less urgent traffic, scheduling backups and updates out of hours, and caching large files near the users (for example images pre-fetched to the site before the clinic day).
- For latency: a terrestrial or low-orbit link instead of a geostationary satellite where possible, servers or a cloud region closer to the users, a virtual desktop or thin-client session that keeps the chatty traffic inside the data centre and sends only screen updates, and applications that make fewer round trips.
- For jitter: QoS for voice and video, buffering in the application, and a less congested path.
Which fix fits comes from measuring first: buying bandwidth for a latency problem is the classic waste.
What do candidates get wrong?
- Treating latency as low bandwidth. A big pipe with a long delay is still slow for interactive work.
- Mixing bits and bytes. 100 Mbps moves about 12.5 megabytes per second, not 100.
- Taking bandwidth for throughput. The rate achieved can never exceed the link's capacity and is usually below it.
- Blaming the network for every slowdown. If one application is slow while others are fine, the cause is often the application or its server.
- Forgetting jitter for telehealth. Live video needs a steady delay, not only enough capacity.
Summary for revision
- Bandwidth = capacity (Mbps); latency = delay (ms); throughput = what is actually achieved.
- Large files need bandwidth; chatty, interactive applications need low latency.
- Jitter is variation in delay and breaks voice and video.
- Distance and satellites add latency that no extra bandwidth removes.
- Measure first: a speed test for bandwidth, ping or traceroute for latency.
Sources
- IETF RFC 2544, Benchmarking Methodology for Network Interconnect Devices (throughput and latency).
- IETF RFC 3393, IP Packet Delay Variation Metric (jitter).
- ITU-T Recommendation G.114, One-way transmission time (delay limits for voice).
Related free practice questions
The site files this concept under the domain Healthcare and Technology Environments.
Try exam-style questions
A 20-question trial exam with an explanation for every option and a score per domain, drawn from 40 free questions. No payment method needed.
Start the trial examQuestions about the exam: CPHIMS guide · What is CPHIMS? · Exam cost · Eligibility · Passing score · Renewal · Exam day · How long to study · Is it worth it? · What is CAHIMS?
Concepts explained: HL7 v2 vs FHIR · Capitation vs fee-for-service · Safe Harbor vs Expert Determination · Incident vs problem vs change · Backup types, RTO and RPO · Privacy Rule vs Security Rule · CDS five rights · Lewin vs Kotter · SDLC and go-live strategies · Testing types · Earned value (SPI, CPI) · Structure, process, outcome · HIE models · Six Sigma vs Lean · Levels of interoperability · EMTALA · Risk responses · RACI matrix · Build vs buy · Sunk vs opportunity cost · Data governance
The four domains: Healthcare and Technology Environments · Clinical Informatics · Healthcare Information and Systems Management · Management and Leadership
Written by Adel Moustafa, About us
Written and checked against the sources above on 22 October 2026. An independent site, not affiliated with or endorsed by HIMSS. CPHIMS® is a registered trademark of HIMSS.