# timescaledb

<https://github.com/timescale/timescaledb>

## Test

### Récup datas

```bash
ssh grab-4
docker exec client-demo-postgres-1 pg_dump -Fc -U postgres postgres > /tmp/backup_postgres.dump
exit

cd /tmp
scp grab-4:/tmp/backup_postgres.dump .
pg_restore -c -x -t mesure -d postgres -h localhost -U postgres -W  backup_postgres.dump
pg_restore -c -x -I mesure_idx1 -d postgres -h localhost -U postgres -W  backup_postgres.dump
pg_restore -c -x -I mesure_idx2 -d postgres -h localhost -U postgres -W  backup_postgres.dump
```

### Création d'une hypertable

```sql
CREATE TABLE mesure_ng (
  "date" TIMESTAMPTZ,
  id_captation INT NOT NULL,
  valeur float4 NOT NULL
) WITH (
  tsdb.hypertable,
  timescaledb.segmentby = 'id_captation',
  timescaledb.orderby='date DESC'
)

INSERT INTO mesure_ng ("date", id_captation, valeur)
  SELECT "date", id_captation, valeur
    FROM mesure
```

### Compression des données (stockage en colonnes)

`SELECT add_columnstore_policy('mesure_ng', INTERVAL '15 days')`


### Ex. de requête

```sql
SELECT
  time_bucket('1 hour', "date") AS heure,
  round(
    average(
      time_weight('linear', "date", valeur)
    )
  ) AS moyenne_conso,
  COUNT(*) AS nombre_de_mesures
FROM
  mesure_ng
WHERE
  "date" BETWEEN '2026-03-01 00:00:00' AND '2026-03-31 23:59:59'
  and id_captation = 59 -- puissance PAC
GROUP BY
  heure
ORDER BY
  heure;
```

### Exploit

```sql
SELECT *
FROM timescaledb_information.chunks
WHERE hypertable_name = 'mesure_ng'
```

`SELECT * FROM timescaledb_information.jobs`

### Continuous aggregates (WIP)

<https://www.tigerdata.com/docs/use-timescale/latest/continuous-aggregates/create-a-continuous-aggregate>

* CREATE MATERIALIZED VIEW
* CALL add_continuous_aggregate_policy
* supprimer les "vieilles" données de mesure_ng (add_retention_policy ?)

#### Exemples

##### Gauges

CREATE MATERIALIZED VIEW sensor_hourly
WITH (timescaledb.continuous)
AS
SELECT
    time_bucket('1 hour', time) AS bucket,
    sensor_id,
    average(
      time_weight('linear', "date", valeur)
    ) AS moyenne_temporelle,
    percentile_agg(value) AS pct
FROM sensor_data
GROUP BY bucket, sensor_id;

Puis

SELECT
    bucket,
    sensor_id,
    approx_percentile(0.95, pct) AS p95,
    approx_percentile(0.99, pct) AS p99
FROM sensor_hourly;

##### Index

delta(counter_agg(time, value))