Creating and deploying indexes
Indexes are used by server to satisfy queries, whenever a user issues a query RavenDB will use an existing index if it matches query or create a new one if no match is found.
Remember
Indexes created by issuing a query are called dynamic
or Auto
indexes and can be easily identified, due to their name that starts with Auto/
prefix.
Indexes created explicitly by user are called static
.
Static indexes
There are a couple of ways to create static index
and send it to server, we can use low-level commands or create a custom class. There is also a possibility to scan an assembly and deploy all found indexes or even deploy indexes side by side.
using AbstractIndexCreationTask
If you are interested in having a strongly-typed syntax during index creation, have an ability to deploy indexes using assembly scanner, avoid hard-coding index names in every query and do not worry about setting sorting for numerical fields, then AbstractIndexCreationTask
should be your choice.
Note
We recommend creating and using indexes in this form due to its simplicity, many benefits and few disadvantages.
Safe By Default
When index is deployed using AbstractIndexCreationTask
then all numerical fields in generated index definition will contain appropriate sorting options set.
Naming conventions
Actually there is only one naming conventions: each _
in class name will be translated to /
in index name.
e.g.
In Northwind
samples there is a index called Orders/Totals
. To get such a index name, we need to create class called Orders_Totals
.
public class Orders_Totals : AbstractIndexCreationTask<Order>
{
// ...
}
Sending to server
There is not much use from an index if it is not deployed to server. In order to do so, we need to create instance of our class that inherits from AbstractIndexCreationTask
and use one of the deployment methods: Execute
or ExecuteAsync
for asynchronous call.
// deploy index to `DefaultDatabase` for given `DocumentStore`
// using store `Conventions`
new Orders_Totals().Execute(store);
// deploy asynchronously index to `DefaultDatabase` for given `DocumentStore`
// using store `Conventions`
await new Orders_Totals().ExecuteAsync(store.AsyncDatabaseCommands, store.Conventions);
// deploy index to `Northwind` database
// using store `Conventions`
new Orders_Totals().Execute(store.DatabaseCommands.ForDatabase("Northwind"), store.Conventions);
Safe By Default
If index exists on server and stored definition is the same as the one that was send, then it will not be overwritten, which implies that indexed data will not be deleted and indexation will not start from scratch.
Using assembly scanner
All classes that inherit from AbstractIndexCreationTask
can be deployed at once using one of IndexCreation.CreateIndexes
method overloads.
// deploy all indexes (and transformers)
// from assembly where `Orders_Totals` is found
// to `DefaultDatabase` for given `DocumentStore`
IndexCreation.CreateIndexes(typeof(Orders_Totals).Assembly, store);
Underneath, the IndexCreation
will attempt to create all indexes in a single request. If it fails, then it will repeat the execution by calling
Execute
method one-by-one for each of found indexes in separate requests.
Warning
IndexCreation.CreateIndexes
will also deploy all classes that inherit from AbstractTransformerCreationTask
(more about it here).
Example
public class Orders_Totals : AbstractIndexCreationTask<Order>
{
public class Result
{
public string Employee { get; set; }
public string Company { get; set; }
public decimal Total { get; set; }
}
public Orders_Totals()
{
Map = orders => from order in orders
select new
{
order.Employee,
order.Company,
Total = order.Lines.Sum(l => (l.Quantity * l.PricePerUnit) * (1 - l.Discount))
};
}
}
public static void Main(string[] args)
{
using (DocumentStore store = new DocumentStore
{
Url = "http://localhost:8080",
DefaultDatabase = "Northwind"
})
{
store.Initialize();
new Orders_Totals().Execute(store);
using (IDocumentSession session = store.OpenSession())
{
IList<Order> orders = session
.Query<Orders_Totals.Result, Orders_Totals>()
.Where(x => x.Total > 100)
.OfType<Order>()
.ToList();
}
}
}
using Commands
Low-level PutIndex
command from DatabaseCommands
(which API reference can be found here) can be used also to send index to server.
The benefit of this approach is that you can choose the name as you feel fit and change various settings available in IndexDefinition
, but loose the ability to deploy using assembly scanner. Also you will have to use string-based names of indexes when querying.
store
.DatabaseCommands
.PutIndex("Orders/Totals", new IndexDefinition
{
Map = @"from order in docs.Orders
select new { order.Employee, order.Company, Total = order.Lines.Sum(l => (l.Quantity * l.PricePerUnit) * (1 - l.Discount)) }"
});
IndexDefinitionBuilder
IndexDefinitionBuilder
is a very useful class that enables you to create IndexDefinitions
using strongly-typed syntax with an access to low-level settings not available when AbstractIndexCreationTask
approach is used.
IndexDefinitionBuilder<Order> builder = new IndexDefinitionBuilder<Order>();
builder.Map = orders => from order in orders
select new
{
order.Employee,
order.Company,
Total = order.Lines.Sum(l => (l.Quantity * l.PricePerUnit) * (1 - l.Discount))
};
store
.DatabaseCommands
.PutIndex("Orders/Totals", builder.ToIndexDefinition(store.Conventions));
Remarks
Information
Commands or IndexDefinitionBuilder
approaches are not recommended and should be used only if you can't do that by inheriting from AbstractIndexCreationTask
.
Auto indexes
Auto-indexes are created when queries that do not specify an index name are executed and (after in-depth query analysis) no matching index is found on server-side.
Naming convention
As mentioned earlier auto-indexes can be recognized by Auto/
prefix in name. Their name also contains name of a collection that was queried and list of fields that were required to find valid query results.
For instance, issuing query like this
List<Employee> employees = session
.Query<Employee>()
.Where(x => x.FirstName == "Robert" && x.LastName == "King")
.ToList();
will result in a creation of a index named Auto/Employees/ByFirstNameAndLastName
.
Disabling creation of Auto indexes
By default, Raven/CreateAutoIndexesForAdHocQueriesIfNeeded
configuration option is set to true
, which allows auto-index creation if needed. To disable it server or database-wide please refer to this article.
Auto indexes and indexing prioritization
To reduce the server load, if auto-indexes are not queried for a certain amount of time defined in Raven/TimeToWaitBeforeMarkingAutoIndexAsIdle
setting (1 hour by default), then they will be marked as Idle
. You can read more about implications of marking index as Idle
here.
If this is disabled, you'll have to manually ensure that all queries have covering indexes. This is not a recommended configuration.
Remarks
In-memory indexes
By default, new indexes are created in-memory (can be disabled by changing Raven/DisableInMemoryIndexing
to false
) and persisted to disk when one of the following conditions are met:
- index is not stale after write,
- when in-memory size of an index exceeds 64MB (can be altered by changing
Raven/NewIndexInMemoryMaxMB
), - when difference between current write time and index creation time exceeds 15 minutes (can be altered by changing
Raven/NewIndexInMemoryMaxTime
), - when backup is created,
- when database is closed,
- on user request. More info here.