Data Stores lets you use simple cloud storage to create serverless, distributed, highly available, low-cost databases and caches with sophisticated keying, richly structured values, and compression at-rest.

Data Stores integrates with a wide variety of storage providers and formats, all using a single, simple abstraction.

• Inexpensive Serverless Databases: Use simple cloud storage to create serverless, distributed, highly available, low-cost databases.

• Easy Distributed Caches: Create persistent caches to increase performance of backend API controllers, or reduce the number of calls to expensive third-party web APIs.

• Many Supported Storage Providers: Including Azure Blobs, Azure Tables, Amazon S3, FTP, SFTP, and Local File Storage.

• Many Supported Formats: Including human-readable structured data like JSON and YAML, efficient structured binary data with Bond or Protobuf, delimited tabular data like CSV and TSV, and binary data such as images and raw text.

• Optional Compression: Flip a switch and compress data payloads up to 10x with GZip or LZMA, improving response times and reducing storage and bandwidth costs.

• One Easy Abstraction: Trade in your error prone bare metal integrations and high ceremony repository patterns for a single, elegant IDataStore interface that is uniform across storage platforms and formats.

1. Install NuGet Packages: Install the NuGet packages for your desired storage providers and formats:

   Install-Package Halforbit.DataStores

   See the NuGet Packages section below for a list of available NuGet packages and what storage providers and formats they support.

2. Define Your Stores: Use the DataStore type to create ad-hoc stores, or define them as properties on a data context.

3. Use Your Stores: Persist and retrieve data with your stores, or inspect write times, sizes, metadata and more with the stores' Context property.

There are many options to choose from when creating a store. A store can:

• be defined as a property of a data context, or created ad-hoc and stored in a local variable.
• be keyed with simple types like Guid or string, with the properties of a class, record, or struct, or be a keyless singleton storing a single value.
• store values as simple types like string or byte[], or structured class, record, or struct types serialized as e.g. JSON or TSV.

Let's make a simple ad-hoc store to persist a record value using Azure Blobs, formatted as JSON, and key it with the properties of a record key.

First, we define a type to represent our value:

public record Person(
    string Department,
    Guid PersonId,
    string FirstName,
    string LastName);

Next we define a type to represent our key:

public record PersonKey(
    string Department,
    Guid PersonId);

Next we create a store ad-hoc with a key type of PersonKey, and a value type of Person:

var store = DataStore
    .Describe()
    .BlobStorage()
    .ConnectionString("<connection-string-here>")
    .Container("<container-name-here>")
    .ContentType("application/json")
    .DefaultContentEncoding()
    .JsonSerialization()
    .NoCompression()
    .FileExtension(".json")
    .Map<PersonKey, Person>(k => $"people/{k.Department}/{k.PersonId}")
    .Build();

The Map<,> method lets us define a bi-directional mapping between a PersonKey and a string.

Note that people here is acting similarly to a table or collection name, and the Department property is acting as a kind of partition key.

Now we can put a Person in the store with the Upsert method:

var key = new PersonKey(
    Department: "development",
    PersonId: Guid.NewGuid());

var value = new Person(
    Department: key.Department,
    PersonId: key.PersonId,
    FirstName: "Steve",
    LastName: "Smith");

await store.Upsert(key, value);

We can get a Person by their PersonKey:

var value = await store.Get(key);

We can delete a Person by their PersonKey:

await store.Delete(key);

We can list the PersonKeys in the store, optionally filtered by PersonKey.Department:

var keys = await store.ListKeys(k => k.Department == "development");

We can get all of the Persons in the store, optionally filtered by PersonKey.Department:

var values = await store.ListValues(k => k.Department == "development");

The following NuGet packages are provided, parted out by their dependencies. Install the ones that contain the storage providers and formats you wish to use.

Data Stores is licensed under the MIT License. See the LICENSE file for details.