PartiTables 1.0.3

PartiTables

Azure Table Storage made simple for .NET

🧠 Why PartiTables?

Azure Table Storage is 95% cheaper than other NoSQL solutions and blazing fast — but painful to use. PartiTables fixes that by providing clean, Entity Framework-style data access patterns.

Save Money - $10/month instead of $240/month
Type-Safe - IntelliSense, compile-time checking
Auto-Retry - Built-in resilience with Polly
Batch Operations - Save multiple entities atomically
Less Code - One class replaces hundreds of lines

📦 Installation

dotnet add package PartiTables

🚀 Quick Start

Before vs After

Without PartiTables:

// The old way: manual TableEntity manipulation
var entity = new TableEntity("partition", "row-key");
entity["FirstName"] = "John";
entity["LastName"] = "Doe";
await tableClient.UpsertEntityAsync(entity);
// ... tedious manual parsing when reading
// ... complex batch operations
// ... manual retry logic

With PartiTables:

// Use them like Entity Framework
var patient = new Patient { PatientId = "patient-123" };
patient.Meta.Add(new PatientMeta { FirstName = "John", Email = "john@example.com" });
patient.Consents.Add(new Consent { Type = "DataSharing" });

await repo.SaveAsync(patient);
//   ✅ All related records saved in ONE batch operation
//   ✅ Automatic retry with Polly resilience
//   ✅ RowKeys auto-generated from patterns
//   ✅ Strong typing, IntelliSense, compile-time safety

💡 Key Features

Define Your Model with Declarative Patterns

[TablePartition("Customers", "{CustomerId}")]
public class Customer
{
    public string CustomerId { get; set; }
    
    [RowKeyPrefix("")]
    public List<Order> Orders { get; set; } = new();
    
    [RowKeyPrefix("")]
    public List<Address> Addresses { get; set; } = new();
}

[RowKeyPattern("{CustomerId}-order-{OrderId}")]
public class Order : RowEntity
{
    public string OrderId { get; set; } = Guid.NewGuid().ToString("N")[..8];
    public decimal Amount { get; set; }
    public string Status { get; set; } = "Pending";
}

[RowKeyPattern("{CustomerId}-address-{AddressId}")]
public class Address : RowEntity
{
    public string AddressId { get; set; } = Guid.NewGuid().ToString("N")[..8];
    public string City { get; set; } = default!;
}

Benefits:

• ✅ Self-documenting - pattern visible at a glance
• ✅ 60% less code than manual implementation
• ✅ Automatic key generation from properties
• ✅ Type-safe and compile-time validated

CRUD Operations

// Create
var customer = new Customer { CustomerId = "cust-123" };
customer.Orders.Add(new Order { Amount = 99.99m, Status = "Pending" });
await repo.SaveAsync(customer);

// Read
var customer = await repo.FindAsync("cust-123");
var orders = await repo.QueryCollectionAsync("cust-123", c => c.Orders);

// Update
loaded.Orders[0].Status = "Shipped";
await repo.SaveAsync(loaded);

// Delete
await repo.DeleteAsync("cust-123");

Automatic Batch Transactions

var customer = new Customer { CustomerId = "cust-123" };
customer.Orders.Add(new Order { Amount = 99.99m });
customer.Orders.Add(new Order { Amount = 45.50m });
customer.Addresses.Add(new Address { City = "Seattle" });

await repo.SaveAsync(customer);
//   ✅ All entities saved in ONE atomic batch operation
//   ✅ Either all succeed or all fail (within partition)
//   ✅ Up to 100 operations per batch
//   ✅ Automatic grouping by partition key

Built-in Resilience with Polly

// Automatic retry on transient failures
await repo.SaveAsync(customer);
//   ✅ Retries on network errors
//   ✅ Exponential backoff
//   ✅ Circuit breaker protection

You don't need to:

• ❌ Write retry logic
• ❌ Handle transient failures
• ❌ Implement exponential backoff
• ❌ Track failed operations

Handles Big Data with Automatic Rollback

PartiTables automatically handles datasets larger than Azure's 100-item batch limit with automatic rollback if any batch fails:

// Save 10,000 records across 100 batches
var salesData = GenerateSalesData("store-001", 10_000);

await repo.SaveAsync(salesData);
//   ✅ Automatically split into 100-item batches
//   ✅ If ANY batch fails, ALL previous batches are rolled back
//   ✅ Your data stays consistent - it's all-or-nothing!

What happens:

1. Automatic batching - Splits into 100-item batches
2. Sequential submission - Submits batches one at a time
3. Rollback on failure - If any batch fails, previous batches are automatically deleted
4. Exception preserved - Original error is re-thrown after cleanup

Benefits:

• ✅ No partial data - Either all items save or none do
• ✅ Unlimited scale - Handle 10,000+ items automatically
• ✅ Zero configuration - Works out of the box

🔍 Powerful Querying

Partition-Scoped Queries (Fast)

// Get all data for a partition
var allCustomerData = await repo.FindAsync("customer-123");

// Get specific collection (more efficient)
var orders = await repo.QueryCollectionAsync("customer-123", c => c.Orders);

// Prefix-based queries
var orders2024 = await client.QueryByPrefixAsync("customer-123", "order-2024-");

Collection Filtering

// Load and filter in memory (for small datasets)
var customer = await repo.FindAsync("customer-123");
var pendingOrders = customer.Orders.Where(o => o.Status == "Pending").ToList();

// For larger datasets, query specific collection first
var allOrders = await repo.QueryCollectionAsync("customer-123", c => c.Orders);
var shipped = allOrders.Where(o => o.Status == "Shipped").ToList();

Query Best Practices

• ✅ Query within a single partition for best performance
• ✅ Use prefix-based queries for time-series data
• ✅ Load only the collections you need
• ✅ Avoid cross-partition queries when possible
• ✅ Filter in memory for small result sets

🏗️ How It Works

PartiTables transforms your object graph into optimized Table Storage entities:

// Your code
var patient = new Patient { PatientId = "patient-123" };
patient.Meta.Add(new PatientMeta { FirstName = "John", LastName = "Doe" });
patient.Consents.Add(new Consent { Type = "DataSharing" });
patient.Devices.Add(new DeviceLink { DeviceId = "device-001" });

await repo.SaveAsync(patient);

What happens behind the scenes:

✅ Batch Transaction to Table Storage
PartitionKey: clinic-001
   patient-123-meta          (Auto-generated from pattern)
   patient-123-consent-a7b8  (Auto-generated from pattern)
   patient-123-device-001    (Auto-generated from pattern)

✅ All saved atomically in one batch
✅ Automatic retry on failure
✅ Optimistic concurrency handled

Single Partition (Fast Queries):

PartitionKey: customer-123
RowKeys:
   customer-123-order-001
   customer-123-order-002
   customer-123-address-001

Multi-Tenant Isolation:

PartitionKey: tenant-789
RowKeys:
   tenant-789-user-001
   tenant-789-user-002
   tenant-789-setting-001

🎯 Perfect For

📱 Multi-tenant SaaS | 🛒 Customer orders & history | 🏥 Healthcare records
📋 Audit logs | 👤 User profiles | 🌐 IoT device data
📊 Time-series metrics | 🔔 Notification queues | 📦 Inventory tracking
🎫 Event ticketing | 💬 Chat history | 🔐 Session management

🧪 Try It Now

cd PartiSample
dotnet run

Choose from 6 interactive demos showing real-world scenarios.

📚 Documentation

• Sample Demos - Real-world examples
• Quick Start Guide - Get started
• Big Data Tests - 10,000+ item examples

⚙️ Requirements

• .NET 8.0+
• Azure Storage or Azurite (local development)

📜 License

MIT License — © 2025 PartiTech

🤝 Contributing

Contributions are welcome! Please feel free to submit a Pull Request.