2.2 — Design Data Storage Solutions for Semi-Structured and Unstructured Data

Hands-On: Create Cosmos DB Container with Optimal Partition Key

Step 1: Create Cosmos DB Account

1. Navigate to Azure Cosmos DB > Create
2. Select API: Core (SQL) for JSON with SQL queries
3. Configure:
   • Account name: Globally unique
   • Consistency level: Session (default)
   • Geo-redundancy: Enable for HA
   • Multi-region writes: Enable for active-active
4. Review and create

Step 2: Create Database and Container

1. Open Cosmos DB account > Data Explorer > New Container
2. Database ID: Create or select
3. Container ID: Enter meaningful name (e.g., user-profiles)
4. Partition key: Enter /userId (high-cardinality property)
5. Throughput: Autoscale with max 4,000 RU/s (recommended)
6. Enable TTL if items should auto-expire (e.g., session data)

Step 3: Configure Blob Storage Lifecycle Management

1. Open Storage Account > Lifecycle management
2. Click Add a rule > Code view and enter:

   {
     "rules": [{
       "name": "tier-and-delete",
       "type": "Lifecycle",
       "definition": {
         "actions": {
           "baseBlob": {
             "tierToCool": {"daysAfterModificationGreaterThan": 30},
             "tierToArchive": {"daysAfterModificationGreaterThan": 90},
             "delete": {"daysAfterModificationGreaterThan": 2555}
           }
         },
         "filters": {"blobTypes": ["blockBlob"]}
       }
     }]
   }
   

3. Save and apply — transitions occur automatically

Step 4: Query Cosmos DB with SQL API

1. Open Data Explorer > select container > New SQL Query
2. Example queries:

   -- Efficient: filters on partition key
   SELECT * FROM c WHERE c.userId = 'user-123'
   
   -- Cross-partition: avoid in hot paths
   SELECT * FROM c WHERE c.region = 'US'
   

3. Click Execute Query — note RU consumption for each query
4. Cross-partition queries consume significantly more RUs