Analyze network telemetry, compare service performance and investigate changes across sites and links. l brings q/k array expressions to the numerical work behind network analysis, capacity studies and recorded service data.
01 / The workload
Traffic, latency, packet loss and configuration changes describe different parts of the same network. Engineers need the detail around an unusual interval and a consistent way to compare it across sites and services. Repeatedly exporting smaller files can separate a measurement from the context that explains it.
Select sites, links and time windows from recorded network measurements. Calculate traffic, latency and packet-loss statistics while retaining the samples needed to explain a change.
Group prepared service measurements by region, access technology or service class. Compare distributions and failure counts to understand where the observed experience changes.
Analyze recorded traffic and utilization by site, link and operating period. Compare busy intervals and supplied capacity scenarios to inform your planning process.
Apply the same expressions to recorded intervals before and after an incident or configuration change. Compare related network measurements and retain the observations behind an unexpected result.
02 / From data to analysis
Keep the time window, site and service identity together as you narrow recorded network data into a focused comparison.
Bring normalized, timestamped measurements with explicit site, link and service identifiers and the relevant configuration metadata.
Filter a defined window, calculate network statistics and compare related measurements across sites or service classes.
Return a focused sample extract and a comparison across sites or periods to the notebook or tool where the network investigation takes place.
/ Inspect network latency l>time:til 8 l>latency:20 21 21 24 29 31 27 23 l>where time within 3 5 3 4 5 l>avg latency where time within 3 5 28f l>max latency 31 l>count where latency>28 2 l>
03 / Why l
You write q/k. The runtime considers the physical representation of the data, choosing encoding-aware operations where supported and materializing values when needed.
04 / Your first evaluation
Choose a recorded network interval with a known service issue, a few related sites or links and a baseline period using the same measurement and timestamp conventions.
Available as a research preview for local evaluation.
Why moving fewer bytes matters for analytical workloads.
Read the paper Integration guideCall l from Python, pass typed values and bring results back to your application.
Read the guide Integration guideBring flat columnar datasets into l through the native Parquet reader and writer.
Read the guideYour data. Your next question.