Skip to content

Enable query support for your stateful services in Service Fabric via the OData protocol.

License

Notifications You must be signed in to change notification settings

jessebenson/service-fabric-queryable

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

ServiceFabric.Extensions.Services.Queryable

Getting Started

  1. Create a stateful ASP.NET Core service.

  2. Add the ServiceFabric.Extensions.Services.Queryable nuget package.

  3. Add the ODataQueryable middleware to your Startup.cs. This will intercept calls to the /$query endpoint to expose OData query capabilities over the reliable collections in your service.

using ServiceFabric.Extensions.Services.Queryable;

public class Startup
{
	public void Configure(IApplicationBuilder app, IHostingEnvironment env, ILoggerFactory loggerFactory)
	{
		...
		app.UseODataQueryable();
		...
	}
}

Add the ODataQueryable middleware to your stateful services (using ASP.NET Core stateful services), ensure Reverse Proxy is enabled, and start querying your reliable collections.

Use rcctl to query your middleware

There is a command line tool you can use to query your middleware endpoints called rcctl.

You can install it and start using it with

pip install rcctl

Use REST calls to query your middleware

You can interact directly with the middleware by constructing REST calls. If your service is named 'fabric:/MyApp/MyService' and your reliable dictionary is named 'my-dictionary', try queries like:

Get OData metadata about a single partition stateful service:

  • GET http://localhost:19081/MyApp/MyService/$query/$metadata

Get OData metadata about a partitioned stateful service:

  • GET http://localhost:19081/MyApp/MyService/$query/$metadata?PartitionKind=Int64Range&PartitionKey=0

Get 10 items from the reliable dictionary.

  • GET http://localhost:19081/MyApp/MyService/$query/my-dictionary?$top=10

Get 10 items with Quantity between 2 and 4, inclusively.

  • GET http://localhost:19081/MyApp/MyService/$query/my-dictionary?$top=10&$filter=Quantity ge 2 and Quantity le 4

Get 10 items, returning only the Price and Quantity properties, sorted by Price in descending order.

  • GET http://localhost:19081/MyApp/MyService/$query/my-dictionary?$top=10&$select=Price,Quantity&$orderby=Price desc

Use ReliableIndexedDictionaries to speed up your queries by taking advantage of secondary indices

Some queries may take a long time or not complete. If you expect that you will be querying against a field often, you should consider adding a secondary index on that field, which will dramatically speed up your queries.

To do so, when you are making your dictionary, instead of using

StateManager.GetOrAddAsync<IReliableDictionary>("name")

you should use

StateManager.GetOrAddIndexedAsync("name",
	FilterableIndex<KeyType, ValueType, Property1Type>.CreateQueryableInstance("Property1Name"),
	FilterableIndex<KeyType, ValueType, Property2Type>.CreateQueryableInstance("Property2Name"),
	etc.)

This will create a dictionary with secondary indices on ValueType.Property1Name and ValueType.Property2Name. To find out more about ReliableIndexedDictionary go to the indexing repository

Note: you have to create your indexes the first time you define your dictionary. If you make them later, they will not be truly consistent with the dictionary and this can lead to failed requests or data corruption.

Now, if we made a secondary index on a property called Age on our ValueType, these queries would be faster because of indexing:

  • $filter=Value.Age eq 20
  • $filter=Value.Age gt 25
  • $filter=Value.Age le 40 and Value.Name eq "John"

However the following would not be faster than without using an indexed dictionary:

  • $filter=Value.Name eq "John"
  • $filter=Value.Age le 40 or Value.Name eq "John"

If we added a secondary index on both Age and Name, then all the above queries would be faster.

Check out UserSvc in the BasicApp sample to see both an unindexed and an indexed dictionary being constructed.

Using LINQ to query ReliableIndexedDictionaries

In addition to external requests through the OData middleware, ReliableIndexedDictionaries can be queried from your application code using LINQ. However, similarly to external queries, not all queries will work effectively with indexing. If your query is not supported, you should use GetAllAsync() to get the entire result set and apply your LINQ query against that.

To create a queryable instance of your IReliableIndexedDictionary, call:

var qd = new QueryableReliableIndexedDictionary<TKey, TValue, TValue>(indexedDictionary, stateManager);

Then you can carry out queries such as:

var query = qd.Where(x => x.Age == 25);
var query = qd.Where(x => x.Age >= 25).Select(x => x.Email);
var query = from Person person in qd
            where person.Age >= 25 && person.Email == "[email protected]"
            select person.Email;
var query = qd.Where(x => x.Name.CompareTo("Name1") >= 0);

Some import notes for querying:

  1. Put all your WHERE logic in a single Where statement and join using && and || operators, as the query may not be efficient if it is spread across multiple WHERE clauses.
  2. If you want to compare an IComparable type that does not have ">", "<=" operators, you must give it in the form: property.CompareTo(constant) '>,<,<=,=>' 0

Samples