YASS results
/ uc2 / uc2-tus-n01
uc2-tus-n01
tus
Success
Parameters
| parameter | value | unit | description |
|---|---|---|---|
| engine | tus | — | Storage/transport engine under test (edfs or tus) |
| priority | - | — | File priority class (low/default/high; EDFS PAR) |
| sat_count | 1 | sats | Number of satellites in the constellation |
KPIs
| KPI | value | unit | description |
|---|---|---|---|
| state | Success | — | Terminal experiment state (Success/TimedOut/Failure) |
| exec_date | 2026-06-10 00:45 UTC | UTC | Wall-clock date/time the experiment was launched (CR creationTimestamp) |
| duration | 12m 26s | — | Experiment duration on the simulation clock (simulationStartTime → experimentTime) |
| sim_start | 2026-05-16 23:59 UTC | UTC | Simulation clock at experiment start (simulationStartTime) |
| sim_end | 2026-05-17 00:11 UTC | UTC | Simulation clock when the experiment finished (experimentTime) |
| produced | 1 | files | Files produced by the producer agent(s) |
| files_to_GS | 1 | count | Distinct produced files that reached at least one ground station |
| delivery_success_rate | 100 | % | Distinct files reaching ≥1 GS as a percentage of files produced (UC2/UC5 primary KPI) |
| all_delivered | yes | — | Whether every produced file reached a ground station |
| first_GS_delivery_s | 742 | s | Time from sim start until the first produced file reached any ground station |
| last_GS_delivery_s | 742 | s | Time from sim start until the last produced file reached a ground station (all-delivered time) |
| GS_reached | 1 | count | Ground stations that received the file (from received_total — independent of the delivery-time metric) |
| distinct_receivers | 1 | count | Distinct nodes that received a file (GS + relays) |
| TX_MiB | 34 | MiB | Total network bytes transmitted, all nodes |
| peak_cpu_m | 30 | millicores | Peak CPU across all containers |
| peak_mem_MiB | 13 | MiB | Peak memory across all containers |
| notes | — | — | Data-quality flags for this run (missing metrics, undelivered files) |
Charts (interactive)
Node colours are consistent across every chart and the graph: ■ ground station · ■ relay satellite · ■ producer.
Per-target delivery time
Seconds from file creation until each target node held the complete file. Bars coloured by node kind; the delivery KPI is the earliest ground-station bar.
Network TX over time (MiB)
Cumulative TX summed across all nodes, vs seconds from experiment start.
CPU and memory over time
Aggregate CPU (millicores, left axis) and memory (MiB, right axis) across all containers, vs seconds from experiment start.
Network TX by node (top 20)
Total bytes transmitted per node over the run, coloured by node kind.
Peak CPU per node (top 20)
Peak CPU (millicores) reached by each fsNode during the run, coloured by node kind.
Peak memory per node (top 20)
Peak memory (MiB) reached by each fsNode during the run, coloured by node kind.
File propagation graph
■ ground station · ● relay satellite · ● producer (same colours as the charts). Spread tree: one edge
from → to per node = where it first acquired the file by the selected time. Grouped per file; 0-byte control messages omitted; times bucketed to ~2 s for cross-node clock skew. Timeline spans the whole experiment (0 = simulation start). Delivery completeness is the File deliveries table below. File transfer graph (rate)
Same nodes; edge
from → to = bytes transferred within the trailing window ending at the selected time (instantaneous flow). Edge colour = bandwidth (bits/s, blue = low → red = high); label shows the rate. Play to watch where traffic moves moment-to-moment. Timeline is the simulation clock (0 = simulation start), shared with the charts above.File deliveries
| file | received by | delivery (s from creation) | time from start | #fsNodes with file |
|---|---|---|---|---|
| oneweb-0027_0 | oneweb-0027 producer | t0 | 8s from start ← t0 | 1 |
| oneweb-0027_0 | estrack-new-norcia GS | t0 + 742.1s | 12m 30s from start | 2 |
Charts
charts/network_tx.png
gnuplot/delivery.png
Files
- Kubernetes resources — manifests to reproduce this run (hardwaredefinitions, layout, experimentdefinition, experiment)
- metrics.json — KPIs as JSON
- raw-parquet.tar — raw metrics & events as typed, zstd-compressed Parquet (one file per series; read with pandas/polars/DuckDB)