14 - Lifecycle Timing Benchmark¶

This example measures the embedded ArcadeDB lifecycle from Python with a mixed workload.

It reports timings for:

JVM startup
database create + schema setup
database open
transaction load (table + graph + vectors)
query phase
close
reopen
reopen query phase
reopen close
server mode: close database and open database on a running bundled server, timed over HTTP (three rounds, averaged)

Run¶

From bindings/python/examples:

python3 14_lifecycle_timing.py

With custom workload:

python3 14_lifecycle_timing.py \
  --runs 5 \
  --table-records 50000 \
  --graph-vertices 10000 \
  --vector-records 10000 \
  --vector-dimensions 64 \
  --query-runs 100 \
  --jvm-heap 2g

Notes¶

The benchmark uses a random database path under the system temp directory by default; --db-path sets it instead.
The path is removed at the end (cleanup is always on). The server step uses the same path with a _server suffix and removes it too.
The script is intended for benchmarking in examples, not deterministic CI assertions.

Expected Output (Desktop Baseline)¶

After the per-run lines, the server step prints its two averages (values vary):

Server mode, the same database closed and reopened on a running server:
  close database: <seconds>s avg over 3
  open database:  <seconds>s avg over 3

On a normal desktop CPU, this is a representative summary shape you can expect:

Averages
  jvm start:   0.242371s
  create:      0.065687s
  schema:      0.036061s
  open:        0.003828s
  transaction: 1.217344s
  load:        1.217346s
  query:       5.025231s
  close:       0.002140s
  reopen:      0.011762s
  reopen query:2.343519s
  reopen close:0.001561s

Actual timings vary with CPU, storage, memory, Python/JVM versions, and current system load.