Skip to content

Hosted Demo and Interactive Playground

SketchLog ships with a zero-dependency product evaluation playground that lets visitors explore the main workflows in any browser with no installation required.

Open the playground: https://sbalavignesh123.github.io/sketchlog/demo/

You can also open demo/index.html directly in a browser. The playground is a single-page static app and does not require a server.

What the playground demonstrates

Guided product tour

The tour walks through the full launch story:

  1. bounded-memory telemetry;
  2. DDSketch percentiles;
  3. HyperLogLog-style cardinality;
  4. Count-Min frequency tracking;
  5. stream writes and reads;
  6. streaming SQL examples;
  7. anomaly comparison;
  8. SLO and canary analysis;
  9. multi-tenant namespace isolation;
  10. Sketch Mesh aggregation;
  11. exporter payloads;
  12. storage durability proof links;
  13. local Docker proof instructions;
  14. GitHub and local-trial call to action.

Dashboard mode

The dashboard uses deterministic synthetic telemetry to show:

  • p50, p95, and p99 latency cards;
  • unique-user and session cardinality concepts;
  • top event/error frequency concepts;
  • anomaly, SLO burn, and canary risk summaries;
  • namespace selection;
  • Sketch Mesh node status cards;
  • latency distribution charts.

DDSketch quantile estimation

  • Add arbitrary positive values such as latencies, request sizes, and durations.
  • See p50, p95, and p99 estimates update in real time.
  • Adjust the relative accuracy parameter and see the accuracy guarantee change.
  • Inspect the visual bucket distribution chart.

Stream operations

  • Simulate writing JSON records to any stream path.
  • Read the records back and inspect the log output.
  • Demonstrate the core SketchLog stream abstraction without running a backend.

Streaming SQL playground

The SQL panel provides copyable examples for:

  • service percentile summaries;
  • tenant isolation;
  • top error/event frequency;
  • stable versus canary comparison.

The SQL results are browser-side deterministic fixtures. They are useful for understanding the product workflow, not as a replacement for the local proof commands.

Export payload preview

  • Loki: preview the JSON body sent to /loki/api/v1/push.
  • Datadog: preview the Metrics API v2 series payload.
  • New Relic: preview the Insights Events API payload for US and EU regions.

No exporter request is sent from the browser.

Real proof section

The playground links browser simulation to reproducible local evidence:

  • Docker demo smoke verification;
  • PostgreSQL durability proof stack;
  • unified storage proof for memory, PostgreSQL, and OmniKV;
  • realistic telemetry load proof.

Browser demo mode versus local proof mode

The hosted playground is static because it runs on GitHub Pages.

Mode Where it runs Purpose
Browser demo mode In the page with local JavaScript Safe product tour, dashboard, sketches, streams, SQL examples, and exporter payload previews
Local proof mode On your machine with Docker/Python Real service verification, PostgreSQL durability proof, and storage proof commands
Future live backend mode Optional future hosted API Could provide a real shared backend later with authentication, rate limiting, and tenant isolation

The page should not be interpreted as a hosted production SketchLog service.

Running locally

The playground is pure HTML, CSS, and JavaScript.

python -m http.server 8080 --directory demo

Then open http://localhost:8080.

You can also open demo/index.html directly from the filesystem.

Reproducible proof commands

Run the full Docker demo:

docker compose -f demo/compose.yml up --build --wait
python demo/smoke.py
docker compose -f demo/compose.yml down -v

Run the PostgreSQL durability proof:

python scripts/postgres_durability_proof.py --start --stop

Run unified storage proofs:

python scripts/storage_proof.py --backend memory
python scripts/storage_proof.py --backend postgres --postgres-start --postgres-stop
python scripts/storage_proof.py --backend omnikv

Run the realistic telemetry load proof:

python scripts/telemetry_load_proof.py --backend omnikv

Deploying to GitHub Pages

The repository Pages workflow copies the static website, documentation, and playground into the public GitHub Pages artifact.

The playground must keep working under:

https://sbalavignesh123.github.io/sketchlog/demo/

Architecture

File Role
demo/index.html Single-page app with hero, guided tour, dashboard, sketches, streams, SQL, exporters, proof section, snippets, and CTA
demo/assets/demo.css Dark responsive design system for screenshot-ready launch visuals
demo/assets/demo.js DDSketch implementation, deterministic telemetry fixtures, stream simulation, SQL examples, exporter preview, snippets, and UI behavior

The playground has zero external dependencies, zero JavaScript frameworks, zero build step, and works offline.

DDSketch in the browser

The playground includes a browser implementation of DDSketch:

const sketch = new DDSketch(0.01); // 1% relative accuracy
sketch.add(42.5);
sketch.add(150.0);
const p95 = sketch.quantile(0.95); // guaranteed +/-1% relative error

This demonstrates the same accuracy model used by the Python sketchlog package.

Security

  • No external CDN dependencies.
  • No frontend credentials.
  • No cookies or tracking.
  • No eval() or dynamic code execution.
  • No outbound telemetry from the playground page.