Using a DBMS for the ImageMosaic index

The ImageMosaic plugin for GeoServer allows users to place the index for the granules into a spatial DBMS as explained below. This is especially important when one or more of the following conditions apply:

  • The granules depicts a phenomenon that depends on multiple dimensions (e.g. Time, Elevation and so on)
  • The granules will be added/remove/update frequently at runtime, hence a full-blown transactional DMBS would be a bettere choice for the index.

Datastore.properties Keys Description

The datastore.properties is a properties file which is used by the ImageMosaic plugin in GeoServer in order to place the granule index in a spatial DBMS. Currently the supported ones are as follows:

  • Postgis
  • Oracle
  • H2
  • SQLServer

Internally the ImageMosaic uses the key-value pairs contained in this property file to create a GeoTools DataStore (using the provided parameters) therefore you might notice a parallel between the names of the parameters we will discuss below and those you can configure when creating vector DataStore through the GeoServer user interface, since the underlying machinery is the same.

All the keys that can be configured are explained here below. Generally speaking they depend on the DBMS you are connecting to hence we suggest to check the GeoServer documentation for the relative DataStore for more information:

  • dbtype : database type.
  • host : host name.
  • port : port number.
  • database : database name.
  • schema : used Schema.
  • user : user name to login.
  • passwd : password used to login.
  • namespace : namespace prefix.
  • Data\ Source : data source used.
  • max\ connections : maximum number of open connections. (Default 10)
  • min\ connections : minimum number of pooled connection. (Default 1)
  • validate\ connections : boolean used for checking if the connection is alive before using it. (Default Boolean.FALSE)
  • fetch\ size : number of records read with each iteraction with the dbms.
  • Connection\ timeout : number of seconds the connection pool will wait before timing out attempting to get a new connection. (Default 20 seconds)
  • Primarykey\ metadata\ table : optional table containing primary key structure and sequence associations. Can be expressed as ‘schema.name’ or just ‘name’.
  • Max\ open\ prepared\ statements : maximum number of prepared statements kept open and cached for each connection in the pool. Set to 0 to have unbounded caching, to -1 to disable caching. (Default 50)
  • Expose\ primary\ keys : Expose primary key columns as attributes of the feature type. (Default Boolean.FALSE)
  • create\ database\ params=WITH\ TEMPLATE\ : optional parameter used for setting the proper PostGis template for PostGres version lower than 9.2.3.
  • SPI : DataStoreFactory used. Note the ‘SPI’ key can substitute the ‘dbtype’.
  • Loose\ bbox : boolean indicating if the datastore performs only primary filter on bbox. (Default Boolean.TRUE)
  • Estimated\ extends : use the spatial index information to quickly get an estimate of the data bounds. (Default Boolean.TRUE)
  • preparedStatements : use prepared statements. (Default Boolean.FALSE)
  • jndiReferenceName : used for referencing a preconfigured JNDI connection pool, without having to set the database connection parameters.

datastore.properties Examples

Below a few examples of datastore.properties file are shown for different database types. All the 4 examples create a database with name dbname, username usr and password psw.

Note

For each type of database you want to use, you shall add to the WEB-INF/lib directory of your geoserver installation the related jars, which can be found here. This is especially true for Oracle support as ths DBMS is not supported by default in GeoServer.

Plain PostGIS Example

SPI=org.geotools.data.postgis.PostgisNGDataStoreFactory
host=localhost
port=5434
database=dbname
schema=public
user=usr
passwd=psw
Loose\ bbox=true
Estimated\ extends=false
validate\ connections=true
Connection\ timeout=10
preparedStatements=true

This snippet contains the definition of the host and the port of a PostGIS instance on where the database will be stored. Loose\ bbox=true, Estimated\ extends=false and preparedStatements=true are configured for achieving better performances.

Note

For a PostGIS datastore, it is not needed that the database is already present in the PostGIS instance; it will be created automatically by the imagemosaic reader.

Plain Oracle Example

SPI=org.geotools.data.oracle.OracleNGDataStoreFactory
port=1521
host=localhost
database=dbname
Loose\ bbox=true
Estimated\ extends=false
user=usr
passwd=psw
validate \connections=true
Connection\ timeout=10

This snippet contains the definition of the host and the port of an Oracle instance on where the database will be stored.

Warning

When using an Oracle database to host the index it is necessary to create the database before serving the mosaic because it is not automatically created by the underlying code. This is because database creation is not supported in a standard way through JDBC drivers, and we have a custom solution working for PostGIS only at the time being.

Plain H2 Example

SPI=org.geotools.data.h2.H2DataStoreFactory
type=javax.sql.DataSource
driver=org.h2.Driver
database=dbname
user=usr
password=psw

In this snippet host and port parameters are not defined because we are going to create and embedded database. This database can be accessed directly with an URL like this:

jdbc:h2://path/to/the/db/directory

Plain SQLServer Example

SPI=org.geotools.data.sqlserver.SQLServerDataStoreFactory
port=1433
host=localhost
database=dbname
schema=dbo
Loose\ bbox=true
Estimated\ extends=false
user=usr
passwd=psw
validate \connections=true
Connection\ timeout=10
Geometry\ metadata\ table= GEOMETRY_COLUMNS

This snippet contains the definition of the host and the port of a SQL Server instance on where the database will be stored. A *Geometry metadata table property* has also been defined, to allow the SQL Server datastore to determine the geometry type and the native SRID of the geometry column. For more information see the GeoServer documentation

Warning

As per the Oracle case When using a SQLServer database to host the index it is necessary to create the database before serving the mosaic because it is not automatically created by the underlying code.

Using a JNDI Connection Pool

Java Naming and Directory Interface (JNDI) is a Java API used for accessing resources via a single name mapped as a key.

A common way of using JNDI is the configuration of a JDBC connection pool, which is a global container of database connections. This container is typically used for reducing the database resources usage by sharing the same connection pool between multiple objects accessing the same database. In fact, by maximizing the sharing database connections we minimize the need of creating new ones, which are resource-expensive objects, each time they are requested.

It is worth to point out that GeoServer when creating vector DataStores without JNDI implicitly creates an embeeded pool reducing the possibility of sharing connections between multiple stores.

For configuring a JDBC connection pool with Tomcat, you have to edit the context.xml file inside the $TOMCAT_HOME/conf directory ($TOMCAT_HOME is the directory where Tomcat is installed) defining the parameters associated to the new database connections.

These parameters are:

  • name : The name of the JNDI object.
  • driverClassName : the name of the JDBC driver used.
  • url : URL for connecting to the JDBC database.
  • username : username for accessing the database.
  • password : password for accessing the database.
  • maxActive : The number of maximum active connections to use.
  • maxIdle : The number of maximum unused connections.
  • maxWait : The maximum number of milliseconds that the pool will wait.

Optionally you can set also other parameters like:

  • poolPreparedStatements : Enable the prepared statement pooling (very important for good performance).
  • maxOpenPreparedStatements : The maximum number of prepared statements in pool.
  • validationQuery : (default null) A validation query that double checks the connection is still alive before actually using it.
  • timeBetweenEvictionRunsMillis : (default -1) The number of milliseconds to sleep between runs of the idle object evictor thread. When non-positive, no idle object evictor thread will be run.
  • numTestsPerEvictionRun : (default 3) The number of objects to examine during each run of the idle object evictor thread (if any).
  • minEvictableIdleTimeMillis : : (default 1000 * 60 * 30) The minimum amount of time in milliseconds an object may sit idle in the pool before it is eligible for eviction by the idle object evictor (if any).
  • removeAbandoned : (default false) Flag to remove abandoned connections if they exceed the removeAbandonedTimout. If set to true a connection is considered abandoned and eligible for removal if it has been idle longer than the removeAbandonedTimeout. Setting this to true can recover db connections from poorly written applications which fail to close a connection.
  • removeAbandonedTimeout : (default 300) Timeout in seconds before an abandoned connection can be removed.
  • logAbandoned : (default false) Flag to log stack traces for application code which abandoned a Statement or Connection.
  • testWhileIdle : (default false) Flag used to test connections when idle.

Warning

The previous settings should be modified only by experienced users. Using wrong low values for removedAbandonedTimeout and minEvictableIdleTimeMillis may result in connection failures; if so try it is important to set-up logAbandoned to true and check your catalina.out log file.

These 3 examples below configure a connection pool to a local database called dbname, with username usr and password psw. Also these 3 configurations have the same connection parameters:

  • maxActive=”20”
  • maxIdle=”10”
  • maxWait=”10000”
  • minEvictableIdleTimeMillis=”300000”
  • timeBetweenEvictionRunsMillis=”300000”
  • validationQuery

Note

Validation query may change between the DataBases.

The datastore.properties will then reference the connection pool by name using the jndiReferenceName parameter, and also configure all the store parameters that do not deal with the pool creation itself, including:

  • dbtype : database type.
  • SPI : DataStoreFactory used. Note the ‘SPI’ key can substitute the ‘dbtype’.
  • schema : used Schema.
  • namespace : namespace prefix.
  • Expose\ primary\ keys : Expose primary key columns as attributes of the feature type. (Default Boolean.FALSE)
  • Loose\ bbox : boolean indicating if the datastore performs only primary filter on bbox. (Default Boolean.TRUE)
  • Estimated\ extends : use the spatial index information to quickly get an estimate of the data bounds. (Default Boolean.TRUE)
  • preparedStatements : use prepared statements. (Default Boolean.FALSE)

The list might contain other parameters depending on the specific store being configured, it is advisable to check the documentation about it.

JNDI PostGIS Example

For setting up a PostgreSQL JNDI pool you have to remove the Postgres JDBC driver (it should be named postgresql-X.X-XXX.jdbc3.jar) from the GeoServer WEB-INF/lib folder and put it into the $TOMCAT_HOME/lib folder.

A context.xml example file for PostGIS could be:

<Context>
 <Resource
  name="jdbc/postgres"
  auth="Container"
  type="javax.sql.DataSource"
  driverClassName="org.postgresql.Driver"
  url="jdbc:postgresql://localhost/dbname"
  username="usr"
  password="psw"
  maxActive="20"
  maxIdle="10"
  maxWait="10000"
  minEvictableIdleTimeMillis="300000"
  timeBetweenEvictionRunsMillis="300000"
  validationQuery="SELECT 1"/>
</Context>

and here is the relative datastore.properties file.

# JNDI specific #
#dbtype=
SPI=org.geotools.data.postgis.PostgisNGJNDIDataStoreFactory
#String
# JNDI data source
# Default "java:comp/env/"+"jdbc/mydatabase"
jndiReferenceName=postgres

#Boolean
# perform only primary filter on bbox
# Default Boolean.TRUE
Loose\ bbox=true

#Boolean
# use prepared statements
#Default Boolean.FALSE
preparedStatements=false

JNDI Oracle Example

For setting up an Oracle JNDI pool you have to remove the Oracle JDBC jar ojdbc14.jar file from the WEB-INF/lib folder and put it inside the $TOMCAT_HOME/lib folder.

A context.xml example file could be:

<Context>
  <Resource
   name="jdbc/oralocal"
   auth="Container"
   type="javax.sql.DataSource"
   url="jdbc:oracle:thin:@localhost:1521:dbname"
   driverClassName="oracle.jdbc.driver.OracleDriver"
   username="usr"
   password="psw"
   maxActive="20"
   maxIdle="10"
   maxWait="10000"
   minEvictableIdleTimeMillis="300000"
   timeBetweenEvictionRunsMillis="300000"
   poolPreparedStatements="true"
   maxOpenPreparedStatements="100"
   validationQuery="SELECT SYSDATE FROM DUAL" />
</Context>

JNDI H2 Example

For setting up an H2 JNDI pool you have to remove the H2 JDBC jar file h2-XXXX.jar from the WEB-INF/lib folder and put it inside the $TOMCAT_HOME/lib folder.

context.xml:

<Context>
 <Resource
    name="jdbc/h2"
    auth="Container"
    type="javax.sql.DataSource"
    driverClassName="org.h2.Driver"
    url="jdbc:h2:mem:dbname"
    username="usr"
    password="psw"
    maxActive="20"
    maxIdle="10"
    maxWait="10000"
    minEvictableIdleTimeMillis="300000"
    timeBetweenEvictionRunsMillis="300000"
    validationQuery="SELECT 1"/>
</Context>

JNDI SQLServer Example

For setting up a SQLServer JNDI pool you have to remove the SqlServer JDBC jar file sqljdbcX.jar from the WEB-INF/lib folder and put it inside the $TOMCAT_HOME/lib folder.

context.xml:

<Context>
 <Resource name="jdbc/sqlserver"
 auth="Container"
 type="javax.sql.DataSource"
 driverClassName="com.microsoft.sqlserver.jdbc.SQLServerDriver"
 url="jdbc:sqlserver://localhost:1433;databaseName=test;user=admin;password=admin;"
 username="admin" password="admin"
 maxActive="20"
 initialSize="0"
 minIdle="0"
 maxIdle="8"
 maxWait="10000"
 timeBetweenEvictionRunsMillis="30000"
 minEvictableIdleTimeMillis="60000"
 testWhileIdle="true"
 poolPreparedStatements="true"
 maxOpenPreparedStatements="100"
 validationQuery="SELECT 1"
 maxAge="600000"
 rollbackOnReturn="true"
 />
</Context>