Network tools
Looking Glass.
Public network tests are currently unavailable. Before choosing a network, ask for an authorized target on the same network as your proposed server.
How to compare two server routes
Choose a target and window
Ask for a test address or file on the proposed network and permission to use it. Agree the maximum rate and duration first. Record the source ISP, destination network, direction and times in UTC.
Measure the route
Collect round-trip and destination loss samples from the audience network during its busy hour. Compare with a quieter window using the same target. A hop that answers few probes is not proof of loss at the destination.
Separate aggregate from single-client delivery
Using the agreed file and rate, record bytes and elapsed seconds for one receiver, then for the agreed set of receivers. Report individual results alongside their simultaneous total. State whether the file was cached.
Check actual playback
On a representative device, play a stream at the intended bitrate. Record startup time, stalls and the viewer network. A fast bulk download by itself does not establish smooth playback.
This is a test method, not a BryskBits benchmark. Keep the target, configuration and sampling window with each result.
Questions & answers
Reading network tests
Browse all topicsCan I run a live BryskBits looking-glass test here?
No public test endpoint is available here. Before ordering, ask for a test IP or file for the network you plan to use. Do not benchmark an unrelated host as a substitute.
Does one slow traceroute hop mean my stream will buffer?
Not by itself. Routers may handle diagnostic replies differently from forwarded traffic. Compare the destination results with application latency and repeat measurements during the affected period.
Which direction does a route test measure?
It shows the path from the vantage point toward the destination. The return route may differ; test both directions when diagnosing asymmetric performance.
What information makes two test results comparable?
Keep source, destination, protocol, test duration and time zone with each result. Avoid comparing a quiet local run with a distant peak-hour run.
Can a test download predict disk performance?
Not by itself. The object may already be cached in memory. Use representative file sizes and distinguish cached reads from cold reads when evaluating an origin.