Stream Analytics — Job asa-shopnow

Configuration

Paramètre	Valeur
Nom	asa-shopnow
Streaming Units	1
Compatibilité	1.2
Mode démarrage	JobStartTime (automatique via Terraform)
Statut actuel	Running

Inputs

Input	Event Hub	Format	Groupe de consommateurs
InputOrders	orders	JSON	$Default
InputClickstream	clickstream	JSON	$Default
InputProducts	products	JSON	$Default

Outputs

Output	Table SQL	Opération
OutputOrders	dbo.fact_order	INSERT
OutputClickstream	dbo.fact_clickstream	INSERT
OutputProducts	dbo.dim_product	INSERT (upsert via clé)
OutputCustomers	dbo.dim_customer	INSERT (upsert via clé)

Requêtes ASA (logique de transformation)

-- fact_order
SELECT
    order_id,
    product_id,
    customer_id,
    quantity,
    unit_price,
    status,
    order_timestamp
INTO OutputOrders
FROM InputOrders;

-- fact_clickstream
SELECT
    event_id,
    session_id,
    user_id,
    url,
    event_type,
    event_timestamp
INTO OutputClickstream
FROM InputClickstream;

-- dim_product (upsert)
SELECT
    product_id,
    name,
    category
INTO OutputProducts
FROM InputProducts;

Monitoring du job

-- Vérifier la fraîcheur des données (dernière insertion)
SELECT
    'fact_order'       AS table_name, MAX(order_timestamp)  AS last_event FROM dbo.fact_order
UNION ALL
SELECT
    'fact_clickstream' AS table_name, MAX(event_timestamp)  AS last_event FROM dbo.fact_clickstream
UNION ALL
SELECT
    'dim_product'      AS table_name, MAX(CAST(product_id AS NVARCHAR(50))) AS last_event FROM dbo.dim_product;

Vérification statut Azure CLI :

az stream-analytics job list \
  --resource-group rg-e6-sbuasa \
  --query "[].{name:name, state:jobState}" -o table

Point d'architecture — pourquoi Stream Analytics et pas ADF

Critère	Stream Analytics	ADF
Latence clickstream (2s)	✓ Natif	✗ Batch minimum 15 min
Coût MVP	✓ 1 SU inclus	✗ Coût par activité
Complexité déploiement	✓ Terraform simple	✗ ARM templates lourds
SCD2 vendeurs	✗ Non adapté (pas de MERGE)	✓ Possible mais surdimensionné

→ Stream Analytics pour le streaming, procédure stockée pour le batch SCD2.