Questo contenuto non è disponibile nella lingua selezionata.
Chapter 143. HBase Component
Available as of Camel version 2.10
This component provides an idemptotent repository, producers and consumers for Apache HBase.
Maven users will need to add the following dependency to their pom.xml
for this component:
<dependency> <groupId>org.apache.camel</groupId> <artifactId>camel-hbase</artifactId> <version>x.x.x</version> <!-- use the same version as your Camel core version --> </dependency>
143.1. Apache HBase Overview
HBase is an open-source, distributed, versioned, column-oriented store modeled after Google’s Bigtable: A Distributed Storage System for Structured Data. You can use HBase when you need random, realtime read/write access to your Big Data. More information at Apache HBase.
143.2. Camel and HBase
When using a datasotre inside a camel route, there is always the chalenge of specifying how the camel message will stored to the datastore. In document based stores things are more easy as the message body can be directly mapped to a document. In relational databases an ORM solution can be used to map properties to columns etc. In column based stores things are more challenging as there is no standard way to perform that kind of mapping.
HBase adds two additional challenges:
- HBase groups columns into families, so just mapping a property to a column using a name convention is just not enough.
- HBase doesn’t have the notion of type, which means that it stores everything as byte[] and doesn’t know if the byte[] represents a String, a Number, a serialized Java object or just binary data.
To overcome these challenges, camel-hbase makes use of the message headers to specify the mapping of the message to HBase columns. It also provides the ability to use some camel-hbase provided classes that model HBase data and can be easily convert to and from xml/json etc.
Finally it provides the ability to the user to implement and use his own mapping strategy.
Regardless of the mapping strategy camel-hbase will convert a message into an org.apache.camel.component.hbase.model.HBaseData object and use that object for its internal operations.
143.3. Configuring the component
The HBase component can be provided a custom HBaseConfiguration object as a property or it can create an HBase configuration object on its own based on the HBase related resources that are found on classpath.
<bean id="hbase" class="org.apache.camel.component.hbase.HBaseComponent"> <property name="configuration" ref="config"/> </bean>
If no configuration object is provided to the component, the component will create one. The created configuration will search the class path for an hbase-site.xml file, from which it will draw the configuration. You can find more information about how to configure HBase clients at: HBase client configuration and dependencies
143.4. HBase Producer
As mentioned above camel provides produers endpoints for HBase. This allows you to store, delete, retrieve or query data from HBase using your camel routes.
hbase://table[?options]
where table is the table name.
The supported operations are:
- Put
- Get
- Delete
- Scan
143.4.1. Supported URI options
The HBase component supports 3 options, which are listed below.
Name | Description | Default | Type |
---|---|---|---|
configuration (advanced) | To use the shared configuration | Configuration | |
poolMaxSize (common) | Maximum number of references to keep for each table in the HTable pool. The default value is 10. | 10 | int |
resolveProperty Placeholders (advanced) | Whether the component should resolve property placeholders on itself when starting. Only properties which are of String type can use property placeholders. | true | boolean |
The HBase endpoint is configured using URI syntax:
hbase:tableName
with the following path and query parameters:
143.4.2. Path Parameters (1 parameters):
Name | Description | Default | Type |
---|---|---|---|
tableName | Required The name of the table | String |
143.4.3. Query Parameters (16 parameters):
Name | Description | Default | Type |
---|---|---|---|
cellMappingStrategyFactory (common) | To use a custom CellMappingStrategyFactory that is responsible for mapping cells. | CellMappingStrategy Factory | |
filters (common) | A list of filters to use. | List | |
mappingStrategyClassName (common) | The class name of a custom mapping strategy implementation. | String | |
mappingStrategyName (common) | The strategy to use for mapping Camel messages to HBase columns. Supported values: header, or body. | String | |
rowMapping (common) | To map the key/values from the Map to a HBaseRow. The following keys is supported: rowId - The id of the row. This has limited use as the row usually changes per Exchange. rowType - The type to covert row id to. Supported operations: CamelHBaseScan. family - The column family. Supports a number suffix for referring to more than one columns. qualifier - The column qualifier. Supports a number suffix for referring to more than one columns. value - The value. Supports a number suffix for referring to more than one columns valueType - The value type. Supports a number suffix for referring to more than one columns. Supported operations: CamelHBaseGet, and CamelHBaseScan. | Map | |
rowModel (common) | An instance of org.apache.camel.component.hbase.model.HBaseRow which describes how each row should be modeled | HBaseRow | |
userGroupInformation (common) | Defines privileges to communicate with HBase such as using kerberos. | UserGroupInformation | |
bridgeErrorHandler (consumer) | Allows for bridging the consumer to the Camel routing Error Handler, which mean any exceptions occurred while the consumer is trying to pickup incoming messages, or the likes, will now be processed as a message and handled by the routing Error Handler. By default the consumer will use the org.apache.camel.spi.ExceptionHandler to deal with exceptions, that will be logged at WARN or ERROR level and ignored. | false | boolean |
maxMessagesPerPoll (consumer) | Gets the maximum number of messages as a limit to poll at each polling. Is default unlimited, but use 0 or negative number to disable it as unlimited. | int | |
operation (consumer) | The HBase operation to perform | String | |
remove (consumer) | If the option is true, Camel HBase Consumer will remove the rows which it processes. | true | boolean |
removeHandler (consumer) | To use a custom HBaseRemoveHandler that is executed when a row is to be removed. | HBaseRemoveHandler | |
exceptionHandler (consumer) | To let the consumer use a custom ExceptionHandler. Notice if the option bridgeErrorHandler is enabled then this option is not in use. By default the consumer will deal with exceptions, that will be logged at WARN or ERROR level and ignored. | ExceptionHandler | |
exchangePattern (consumer) | Sets the exchange pattern when the consumer creates an exchange. | ExchangePattern | |
maxResults (producer) | The maximum number of rows to scan. | 100 | int |
synchronous (advanced) | Sets whether synchronous processing should be strictly used, or Camel is allowed to use asynchronous processing (if supported). | false | boolean |
143.5. Spring Boot Auto-Configuration
The component supports 4 options, which are listed below.
Name | Description | Default | Type |
---|---|---|---|
camel.component.hbase.configuration | To use the shared configuration. The option is a org.apache.hadoop.conf.Configuration type. | String | |
camel.component.hbase.enabled | Enable hbase component | true | Boolean |
camel.component.hbase.pool-max-size | Maximum number of references to keep for each table in the HTable pool. The default value is 10. | 10 | Integer |
camel.component.hbase.resolve-property-placeholders | Whether the component should resolve property placeholders on itself when starting. Only properties which are of String type can use property placeholders. | true | Boolean |
143.5.1. Put Operations.
HBase is a column based store, which allows you to store data into a specific column of a specific row. Columns are grouped into families, so in order to specify a column you need to specify the column family and the qualifier of that column. To store data into a specific column you need to specify both the column and the row.
The simplest scenario for storing data into HBase from a camel route, would be to store part of the message body to specified HBase column.
<route> <from uri="direct:in"/> <!-- Set the HBase Row --> <setHeader headerName="CamelHBaseRowId"> <el>${in.body.id}</el> </setHeader> <!-- Set the HBase Value --> <setHeader headerName="CamelHBaseValue"> <el>${in.body.value}</el> </setHeader> <to uri="hbase:mytable?operation=CamelHBasePut&family=myfamily&qualifier=myqualifier"/> </route>
The route above assumes that the message body contains an object that has an id and value property and will store the content of value in the HBase column myfamily:myqualifier in the row specified by id. If we needed to specify more than one column/value pairs we could just specify additional column mappings. Notice that you must use numbers from the 2nd header onwards, eg RowId2, RowId3, RowId4, etc. Only the 1st header does not have the number 1.
<route> <from uri="direct:in"/> <!-- Set the HBase Row 1st column --> <setHeader headerName="CamelHBaseRowId"> <el>${in.body.id}</el> </setHeader> <!-- Set the HBase Row 2nd column --> <setHeader headerName="CamelHBaseRowId2"> <el>${in.body.id}</el> </setHeader> <!-- Set the HBase Value for 1st column --> <setHeader headerName="CamelHBaseValue"> <el>${in.body.value}</el> </setHeader> <!-- Set the HBase Value for 2nd column --> <setHeader headerName="CamelHBaseValue2"> <el>${in.body.othervalue}</el> </setHeader> <to uri="hbase:mytable?operation=CamelHBasePut&family=myfamily&qualifier=myqualifier&family2=myfamily&qualifier2=myqualifier2"/> </route>
It is important to remember that you can use uri options, message headers or a combination of both. It is recommended to specify constants as part of the uri and dynamic values as headers. If something is defined both as header and as part of the uri, the header will be used.
143.5.2. Get Operations.
A Get Operation is an operation that is used to retrieve one or more values from a specified HBase row. To specify what are the values that you want to retrieve you can just specify them as part of the uri or as message headers.
<route> <from uri="direct:in"/> <!-- Set the HBase Row of the Get --> <setHeader headerName="CamelHBaseRowId"> <el>${in.body.id}</el> </setHeader> <to uri="hbase:mytable?operation=CamelHBaseGet&family=myfamily&qualifier=myqualifier&valueType=java.lang.Long"/> <to uri="log:out"/> </route>
In the example above the result of the get operation will be stored as a header with name CamelHBaseValue.
143.5.3. Delete Operations.
You can also you camel-hbase to perform HBase delete operation. The delete operation will remove an entire row. All that needs to be specified is one or more rows as part of the message headers.
<route> <from uri="direct:in"/> <!-- Set the HBase Row of the Get --> <setHeader headerName="CamelHBaseRowId"> <el>${in.body.id}</el> </setHeader> <to uri="hbase:mytable?operation=CamelHBaseDelete"/> </route>
143.5.4. Scan Operations.
A scan operation is the equivalent of a query in HBase. You can use the scan operation to retrieve multiple rows. To specify what columns should be part of the result and also specify how the values will be converted to objects you can use either uri options or headers.
<route> <from uri="direct:in"/> <to uri="hbase:mytable?operation=CamelHBaseScan&family=myfamily&qualifier=myqualifier&valueType=java.lang.Long&rowType=java.lang.String"/> <to uri="log:out"/> </route>
In this case its probable that you also also need to specify a list of filters for limiting the results. You can specify a list of filters as part of the uri and camel will return only the rows that satisfy ALL the filters.
To have a filter that will be aware of the information that is part of the message, camel defines the ModelAwareFilter. This will allow your filter to take into consideration the model that is defined by the message and the mapping strategy.
When using a ModelAwareFilter camel-hbase will apply the selected mapping strategy to the in message, will create an object that models the mapping and will pass that object to the Filter.
For example to perform scan using as criteria the message headers, you can make use of the ModelAwareColumnMatchingFilter as shown below.
<route> <from uri="direct:scan"/> <!-- Set the Criteria --> <setHeader headerName="CamelHBaseFamily"> <constant>name</constant> </setHeader> <setHeader headerName="CamelHBaseQualifier"> <constant>first</constant> </setHeader> <setHeader headerName="CamelHBaseValue"> <el>in.body.firstName</el> </setHeader> <setHeader headerName="CamelHBaseFamily2"> <constant>name</constant> </setHeader> <setHeader headerName="CamelHBaseQualifier2"> <constant>last</constant> </setHeader> <setHeader headerName="CamelHBaseValue2"> <el>in.body.lastName</el> </setHeader> <!-- Set additional fields that you want to be return by skipping value --> <setHeader headerName="CamelHBaseFamily3"> <constant>address</constant> </setHeader> <setHeader headerName="CamelHBaseQualifier3"> <constant>country</constant> </setHeader> <to uri="hbase:mytable?operation=CamelHBaseScan&filters=#myFilterList"/> </route> <bean id="myFilters" class="java.util.ArrayList"> <constructor-arg> <list> <bean class="org.apache.camel.component.hbase.filters.ModelAwareColumnMatchingFilter"/> </list> </constructor-arg> </bean>
The route above assumes that a pojo is with properties firstName and lastName is passed as the message body, it takes those properties and adds them as part of the message headers. The default mapping strategy will create a model object that will map the headers to HBase columns and will pass that model the ModelAwareColumnMatchingFilter. The filter will filter out any rows, that do not contain columns that match the model. It is like query by example.
143.6. HBase Consumer
The Camel HBase Consumer, will perform repeated scan on the specified HBase table and will return the scan results as part of the message. You can either specify header mapping (default) or body mapping. The later will just add the org.apache.camel.component.hbase.model.HBaseData as part of the message body.
hbase://table[?options]
You can specify the columns that you want to be return and their types as part of the uri options:
hbase:mutable?family=name&qualifer=first&valueType=java.lang.String&family=address&qualifer=number&valueType2=java.lang.Integer&rowType=java.lang.Long
The example above will create a model object that is consisted of the specified fields and the scan results will populate the model object with values. Finally the mapping strategy will be used to map this model to the camel message.
143.7. HBase Idempotent repository
The camel-hbase component also provides an idempotent repository which can be used when you want to make sure that each message is processed only once. The HBase idempotent repository is configured with a table, a column family and a column qualifier and will create to that table a row per message.
HBaseConfiguration configuration = HBaseConfiguration.create(); HBaseIdempotentRepository repository = new HBaseIdempotentRepository(configuration, tableName, family, qualifier); from("direct:in") .idempotentConsumer(header("messageId"), repository) .to("log:out);
143.8. HBase Mapping
It was mentioned above that you the default mapping strategies are header and body mapping.
Below you can find some detailed examples of how each mapping strategy works.
143.8.1. HBase Header mapping Examples
The header mapping is the default mapping. To put the value "myvalue" into HBase row "myrow" and column "myfamily:mycolum" the message should contain the following headers:
Header | Value |
---|---|
CamelHBaseRowId | myrow |
CamelHBaseFamily | myfamily |
CamelHBaseQualifier | myqualifier |
CamelHBaseValue | myvalue |
To put more values for different columns and / or different rows you can specify additional headers suffixed with the index of the headers, e.g:
Header | Value |
---|---|
CamelHBaseRowId | myrow |
CamelHBaseFamily | myfamily |
CamelHBaseQualifier | myqualifier |
CamelHBaseValue | myvalue |
CamelHBaseRowId2 | myrow2 |
CamelHBaseFamily2 | myfamily |
CamelHBaseQualifier2 | myqualifier |
CamelHBaseValue2 | myvalue2 |
In the case of retrieval operations such as get or scan you can also specify for each column the type that you want the data to be converted to. For exampe:
Header | Value |
---|---|
CamelHBaseFamily | myfamily |
CamelHBaseQualifier | myqualifier |
CamelHBaseValueType | Long |
Please note that in order to avoid boilerplate headers that are considered constant for all messages, you can also specify them as part of the endpoint uri, as you will see below.
143.8.2. Body mapping Examples
In order to use the body mapping strategy you will have to specify the option mappingStrategy as part of the uri, for example:
hbase:mytable?mappingStrategyName=body
To use the body mapping strategy the body needs to contain an instance of org.apache.camel.component.hbase.model.HBaseData. You can construct t
HBaseData data = new HBaseData(); HBaseRow row = new HBaseRow(); row.setId("myRowId"); HBaseCell cell = new HBaseCell(); cell.setFamily("myfamily"); cell.setQualifier("myqualifier"); cell.setValue("myValue"); row.getCells().add(cell); data.addRows().add(row);
The object above can be used for example in a put operation and will result in creating or updating the row with id myRowId and add the value myvalue to the column myfamily:myqualifier.
The body mapping strategy might not seem very appealing at first. The advantage it has over the header mapping strategy is that the HBaseData object can be easily converted to or from xml/json.
143.9. See also
- Polling Consumer
- Apache HBase