http://git-wip-us.apache.org/repos/asf/incubator-impala/blob/3c2c8f12/docs/topics/impala_hbase.xml ---------------------------------------------------------------------- diff --git a/docs/topics/impala_hbase.xml b/docs/topics/impala_hbase.xml index d8880d3..5ce47e4 100644 --- a/docs/topics/impala_hbase.xml +++ b/docs/topics/impala_hbase.xml @@ -4,7 +4,16 @@ <title id="hbase">Using Impala to Query HBase Tables</title> <titlealts audience="PDF"><navtitle>HBase Tables</navtitle></titlealts> - + <prolog> + <metadata> + <data name="Category" value="Impala"/> + <data name="Category" value="HBase"/> + <data name="Category" value="Querying"/> + <data name="Category" value="Data Analysts"/> + <data name="Category" value="Developers"/> + <data name="Category" value="Tables"/> + </metadata> + </prolog> <conbody> @@ -17,7 +26,881 @@ ranges of values. </p> - + <p> + From the perspective of an Impala user, coming from an RDBMS background, HBase is a kind of key-value store + where the value consists of multiple fields. The key is mapped to one column in the Impala table, and the + various fields of the value are mapped to the other columns in the Impala table. + </p> + + <p> + For background information on HBase, see the snapshot of the Apache HBase site (including documentation) for + the level of HBase that comes with + <xref href="https://archive.cloudera.com/cdh4/cdh/4/hbase/"; scope="external" format="html">CDH 4</xref> or + <xref href="https://archive.cloudera.com/cdh5/cdh/5/hbase/"; scope="external" format="html">CDH 5</xref>. To + install HBase on a CDH cluster, see the installation instructions for + <xref href="http://www.cloudera.com/content/cloudera-content/cloudera-docs/CDH4/latest/CDH4-Installation-Guide/cdh4ig_topic_20.html"; scope="external" format="html">CDH + 4</xref> or +<!-- Original URL: http://www.cloudera.com/content/cloudera-content/cloudera-docs/CDH5/latest/CDH5-Installation-Guide/cdh_ig_hbase_installation.html --> + <xref href="http://www.cloudera.com/documentation/enterprise/latest/topics/cdh_ig_hbase_installation.html"; scope="external" format="html">CDH + 5</xref>. + </p> + + <p outputclass="toc inpage"/> + </conbody> + + <concept id="hbase_using"> + + <title>Overview of Using HBase with Impala</title> + <prolog> + <metadata> + <data name="Category" value="Concepts"/> + </metadata> + </prolog> + + <conbody> + + <p> + When you use Impala with HBase: + </p> + + <ul> + <li> + You create the tables on the Impala side using the Hive shell, because the Impala <codeph>CREATE + TABLE</codeph> statement currently does not support custom SerDes and some other syntax needed for these + tables: + <ul> + <li> + You designate it as an HBase table using the <codeph>STORED BY + 'org.apache.hadoop.hive.hbase.HBaseStorageHandler'</codeph> clause on the Hive <codeph>CREATE + TABLE</codeph> statement. + </li> + + <li> + You map these specially created tables to corresponding tables that exist in HBase, with the clause + <codeph>TBLPROPERTIES("hbase.table.name" = "<varname>table_name_in_hbase</varname>")</codeph> on the + Hive <codeph>CREATE TABLE</codeph> statement. + </li> + + <li> + See <xref href="#hbase_queries"/> for a full example. + </li> + </ul> + </li> + + <li> + You define the column corresponding to the HBase row key as a string with the <codeph>#string</codeph> + keyword, or map it to a <codeph>STRING</codeph> column. + </li> + + <li> + Because Impala and Hive share the same metastore database, once you create the table in Hive, you can + query or insert into it through Impala. (After creating a new table through Hive, issue the + <codeph>INVALIDATE METADATA</codeph> statement in <cmdname>impala-shell</cmdname> to make Impala aware of + the new table.) + </li> + + <li> + You issue queries against the Impala tables. For efficient queries, use <codeph>WHERE</codeph> clauses to + find a single key value or a range of key values wherever practical, by testing the Impala column + corresponding to the HBase row key. Avoid queries that do full-table scans, which are efficient for + regular Impala tables but inefficient in HBase. + </li> + </ul> + + <p> + To work with an HBase table from Impala, ensure that the <codeph>impala</codeph> user has read/write + privileges for the HBase table, using the <codeph>GRANT</codeph> command in the HBase shell. For details + about HBase security, see the + <xref href="http://hbase.apache.org/book/ch08s04.html"; format="html" scope="external">Security chapter in + the HBase Reference Guide</xref>. + </p> + </conbody> + </concept> + + <concept id="hbase_config"> + + <title>Configuring HBase for Use with Impala</title> + <prolog> + <metadata> + <data name="Category" value="Configuring"/> + </metadata> + </prolog> + + <conbody> + + <p> + HBase works out of the box with Impala. There is no mandatory configuration needed to use these two + components together. + </p> + + <p> + To avoid delays if HBase is unavailable during Impala startup or after an <codeph>INVALIDATE + METADATA</codeph> statement, Cloudera recommends setting timeout values as follows in + <filepath>/etc/impala/conf/hbase-site.xml</filepath> (for environments not managed by Cloudera Manager): + </p> + +<codeblock><property> + <name>hbase.client.retries.number</name> + <value>3</value> +</property> +<property> + <name>hbase.rpc.timeout</name> + <value>3000</value> +</property> +</codeblock> + + <p> + Currently, Cloudera Manager does not have an Impala-only override for HBase settings, so any HBase + configuration change you make through Cloudera Manager would take affect for all HBase applications. + Therefore, this change is not recommended on systems managed by Cloudera Manager. + </p> + </conbody> + </concept> + + <concept id="hbase_types"> + + <title>Supported Data Types for HBase Columns</title> + + <conbody> + + <p> + To understand how Impala column data types are mapped to fields in HBase, you should have some background + knowledge about HBase first. You set up the mapping by running the <codeph>CREATE TABLE</codeph> statement + in the Hive shell. See + <xref href="https://cwiki.apache.org/confluence/display/Hive/HBaseIntegration"; scope="external" format="html">the + Hive wiki</xref> for a starting point, and <xref href="#hbase_queries"/> for examples. + </p> + + <p> + HBase works as a kind of <q>bit bucket</q>, in the sense that HBase does not enforce any typing for the + key or value fields. All the type enforcement is done on the Impala side. + </p> + + <p> + For best performance of Impala queries against HBase tables, most queries will perform comparisons in the + <codeph>WHERE</codeph> against the column that corresponds to the HBase row key. When creating the table + through the Hive shell, use the <codeph>STRING</codeph> data type for the column that corresponds to the + HBase row key. Impala can translate conditional tests (through operators such as <codeph>=</codeph>, + <codeph><</codeph>, <codeph>BETWEEN</codeph>, and <codeph>IN</codeph>) against this column into fast + lookups in HBase, but this optimization (<q>predicate pushdown</q>) only works when that column is + defined as <codeph>STRING</codeph>. + </p> + + <p> + Starting in Impala 1.1, Impala also supports reading and writing to columns that are defined in the Hive + <codeph>CREATE TABLE</codeph> statement using binary data types, represented in the Hive table definition + using the <codeph>#binary</codeph> keyword, often abbreviated as <codeph>#b</codeph>. Defining numeric + columns as binary can reduce the overall data volume in the HBase tables. You should still define the + column that corresponds to the HBase row key as a <codeph>STRING</codeph>, to allow fast lookups using + those columns. + </p> + </conbody> + </concept> + + <concept id="hbase_performance"> + + <title>Performance Considerations for the Impala-HBase Integration</title> + <prolog> + <metadata> + <data name="Category" value="Performance"/> + </metadata> + </prolog> + + <conbody> + + <p> + To understand the performance characteristics of SQL queries against data stored in HBase, you should have + some background knowledge about how HBase interacts with SQL-oriented systems first. See + <xref href="https://cwiki.apache.org/confluence/display/Hive/HBaseIntegration"; scope="external" format="html">the + Hive wiki</xref> for a starting point; because Impala shares the same metastore database as Hive, the + information about mapping columns from Hive tables to HBase tables is generally applicable to Impala too. + </p> + + <p> + Impala uses the HBase client API via Java Native Interface (JNI) to query data stored in HBase. This + querying does not read HFiles directly. The extra communication overhead makes it important to choose what + data to store in HBase or in HDFS, and construct efficient queries that can retrieve the HBase data + efficiently: + </p> + + <ul> + <li> + Use HBase table for queries that return a single row or a range of rows, not queries that scan the entire + table. (If a query has no <codeph>WHERE</codeph> clause, that is a strong indicator that it is an + inefficient query for an HBase table.) + </li> + + <li> + If you have join queries that do aggregation operations on large fact tables and join the results against + small dimension tables, consider using Impala for the fact tables and HBase for the dimension tables. + (Because Impala does a full scan on the HBase table in this case, rather than doing single-row HBase + lookups based on the join column, only use this technique where the HBase table is small enough that + doing a full table scan does not cause a performance bottleneck for the query.) + </li> + </ul> + + <p> + Query predicates are applied to row keys as start and stop keys, thereby limiting the scope of a particular + lookup. If row keys are not mapped to string columns, then ordering is typically incorrect and comparison + operations do not work. For example, if row keys are not mapped to string columns, evaluating for greater + than (>) or less than (<) cannot be completed. + </p> + + <p> + Predicates on non-key columns can be sent to HBase to scan as <codeph>SingleColumnValueFilters</codeph>, + providing some performance gains. In such a case, HBase returns fewer rows than if those same predicates + were applied using Impala. While there is some improvement, it is not as great when start and stop rows are + used. This is because the number of rows that HBase must examine is not limited as it is when start and + stop rows are used. As long as the row key predicate only applies to a single row, HBase will locate and + return that row. Conversely, if a non-key predicate is used, even if it only applies to a single row, HBase + must still scan the entire table to find the correct result. + </p> + + <example> + + <title>Interpreting EXPLAIN Output for HBase Queries</title> + + <p> + For example, here are some queries against the following Impala table, which is mapped to an HBase table. + The examples show excerpts from the output of the <codeph>EXPLAIN</codeph> statement, demonstrating what + things to look for to indicate an efficient or inefficient query against an HBase table. + </p> + + <p> + The first column (<codeph>cust_id</codeph>) was specified as the key column in the <codeph>CREATE + EXTERNAL TABLE</codeph> statement; for performance, it is important to declare this column as + <codeph>STRING</codeph>. Other columns, such as <codeph>BIRTH_YEAR</codeph> and + <codeph>NEVER_LOGGED_ON</codeph>, are also declared as <codeph>STRING</codeph>, rather than their + <q>natural</q> types of <codeph>INT</codeph> or <codeph>BOOLEAN</codeph>, because Impala can optimize + those types more effectively in HBase tables. For comparison, we leave one column, + <codeph>YEAR_REGISTERED</codeph>, as <codeph>INT</codeph> to show that filtering on this column is + inefficient. + </p> + +<codeblock>describe hbase_table; +Query: describe hbase_table ++-----------------------+--------+---------+ +| name | type | comment | ++-----------------------+--------+---------+ +| cust_id | <b>string</b> | | +| birth_year | <b>string</b> | | +| never_logged_on | <b>string</b> | | +| private_email_address | string | | +| year_registered | <b>int</b> | | ++-----------------------+--------+---------+ +</codeblock> + + <p> + The best case for performance involves a single row lookup using an equality comparison on the column + defined as the row key: + </p> + +<codeblock>explain select count(*) from hbase_table where cust_id = '[email protected]'; ++------------------------------------------------------------------------------------+ +| Explain String | ++------------------------------------------------------------------------------------+ +| Estimated Per-Host Requirements: Memory=1.01GB VCores=1 | +| WARNING: The following tables are missing relevant table and/or column statistics. | +| hbase.hbase_table | +| | +| 03:AGGREGATE [MERGE FINALIZE] | +| | output: sum(count(*)) | +| | | +| 02:EXCHANGE [PARTITION=UNPARTITIONED] | +| | | +| 01:AGGREGATE | +| | output: count(*) | +| | | +<b>| 00:SCAN HBASE [hbase.hbase_table] |</b> +<b>| start key: [email protected] |</b> +<b>| stop key: [email protected]\0 |</b> ++------------------------------------------------------------------------------------+ +</codeblock> + + <p> + Another type of efficient query involves a range lookup on the row key column, using SQL operators such + as greater than (or equal), less than (or equal), or <codeph>BETWEEN</codeph>. This example also includes + an equality test on a non-key column; because that column is a <codeph>STRING</codeph>, Impala can let + HBase perform that test, indicated by the <codeph>hbase filters:</codeph> line in the + <codeph>EXPLAIN</codeph> output. Doing the filtering within HBase is more efficient than transmitting all + the data to Impala and doing the filtering on the Impala side. + </p> + +<codeblock>explain select count(*) from hbase_table where cust_id between 'a' and 'b' + and never_logged_on = 'true'; ++------------------------------------------------------------------------------------+ +| Explain String | ++------------------------------------------------------------------------------------+ +... +<!--| Estimated Per-Host Requirements: Memory=1.01GB VCores=1 | +| WARNING: The following tables are missing relevant table and/or column statistics. | +| hbase.hbase_table | +| | +| 03:AGGREGATE [MERGE FINALIZE] | +| | output: sum(count(*)) | +| | | +| 02:EXCHANGE [PARTITION=UNPARTITIONED] | +| | |--> +| 01:AGGREGATE | +| | output: count(*) | +| | | +<b>| 00:SCAN HBASE [hbase.hbase_table] |</b> +<b>| start key: a |</b> +<b>| stop key: b\0 |</b> +<b>| hbase filters: cols:never_logged_on EQUAL 'true' |</b> ++------------------------------------------------------------------------------------+ +</codeblock> + + <p> + The query is less efficient if Impala has to evaluate any of the predicates, because Impala must scan the + entire HBase table. Impala can only push down predicates to HBase for columns declared as + <codeph>STRING</codeph>. This example tests a column declared as <codeph>INT</codeph>, and the + <codeph>predicates:</codeph> line in the <codeph>EXPLAIN</codeph> output indicates that the test is + performed after the data is transmitted to Impala. + </p> + +<codeblock>explain select count(*) from hbase_table where year_registered = 2010; ++------------------------------------------------------------------------------------+ +| Explain String | ++------------------------------------------------------------------------------------+ +... +<!--| Estimated Per-Host Requirements: Memory=1.01GB VCores=1 | +| WARNING: The following tables are missing relevant table and/or column statistics. | +| hbase.hbase_table | +| | +| 03:AGGREGATE [MERGE FINALIZE] | +| | output: sum(count(*)) | +| | | +| 02:EXCHANGE [PARTITION=UNPARTITIONED] | +| | |--> +| 01:AGGREGATE | +| | output: count(*) | +| | | +<b>| 00:SCAN HBASE [hbase.hbase_table] |</b> +<b>| predicates: year_registered = 2010 |</b> ++------------------------------------------------------------------------------------+ +</codeblock> + + <p> + The same inefficiency applies if the key column is compared to any non-constant value. Here, even though + the key column is a <codeph>STRING</codeph>, and is tested using an equality operator, Impala must scan + the entire HBase table because the key column is compared to another column value rather than a constant. + </p> + +<codeblock>explain select count(*) from hbase_table where cust_id = private_email_address; ++------------------------------------------------------------------------------------+ +| Explain String | ++------------------------------------------------------------------------------------+ +... +<!--| Estimated Per-Host Requirements: Memory=1.01GB VCores=1 | +| WARNING: The following tables are missing relevant table and/or column statistics. | +| hbase.hbase_table | +| | +| 03:AGGREGATE [MERGE FINALIZE] | +| | output: sum(count(*)) | +| | | +| 02:EXCHANGE [PARTITION=UNPARTITIONED] | +| | |--> +| 01:AGGREGATE | +| | output: count(*) | +| | | +<b>| 00:SCAN HBASE [hbase.hbase_table] |</b> +<b>| predicates: cust_id = private_email_address |</b> ++------------------------------------------------------------------------------------+ +</codeblock> + + <p> + Currently, tests on the row key using <codeph>OR</codeph> or <codeph>IN</codeph> clauses are not + optimized into direct lookups either. Such limitations might be lifted in the future, so always check the + <codeph>EXPLAIN</codeph> output to be sure whether a particular SQL construct results in an efficient + query or not for HBase tables. + </p> + +<codeblock>explain select count(*) from hbase_table where + cust_id = '[email protected]' or cust_id = '[email protected]'; ++----------------------------------------------------------------------------------------+ +| Explain String | ++----------------------------------------------------------------------------------------+ +... +<!--| Estimated Per-Host Requirements: Memory=1.01GB VCores=1 | +| WARNING: The following tables are missing relevant table and/or column statistics. | +| hbase.hbase_table | +| | +| 03:AGGREGATE [MERGE FINALIZE] | +| | output: sum(count(*)) | +| | | +| 02:EXCHANGE [PARTITION=UNPARTITIONED] | +| | |--> +| 01:AGGREGATE | +| | output: count(*) | +| | | +<b>| 00:SCAN HBASE [hbase.hbase_table] |</b> +<b>| predicates: cust_id = '[email protected]' OR cust_id = '[email protected]' |</b> ++----------------------------------------------------------------------------------------+ + +explain select count(*) from hbase_table where + cust_id in ('[email protected]', '[email protected]'); ++------------------------------------------------------------------------------------+ +| Explain String | ++------------------------------------------------------------------------------------+ +... +<!--| Estimated Per-Host Requirements: Memory=1.01GB VCores=1 | +| WARNING: The following tables are missing relevant table and/or column statistics. | +| hbase.hbase_table | +| | +| 03:AGGREGATE [MERGE FINALIZE] | +| | output: sum(count(*)) | +| | | +| 02:EXCHANGE [PARTITION=UNPARTITIONED] | +| | |--> +| 01:AGGREGATE | +| | output: count(*) | +| | | +<b>| 00:SCAN HBASE [hbase.hbase_table] |</b> +<b>| predicates: cust_id IN ('[email protected]', '[email protected]') |</b> ++------------------------------------------------------------------------------------+ +</codeblock> + + <p> + Either rewrite into separate queries for each value and combine the results in the application, or + combine the single-row queries using UNION ALL: + </p> + +<codeblock>select count(*) from hbase_table where cust_id = '[email protected]'; +select count(*) from hbase_table where cust_id = '[email protected]'; + +explain + select count(*) from hbase_table where cust_id = '[email protected]' + union all + select count(*) from hbase_table where cust_id = '[email protected]'; ++------------------------------------------------------------------------------------+ +| Explain String | ++------------------------------------------------------------------------------------+ +... +<!--| Estimated Per-Host Requirements: Memory=1.01GB VCores=1 | +| WARNING: The following tables are missing relevant table and/or column statistics. | +| hbase.hbase_table | +| | +| 09:EXCHANGE [PARTITION=UNPARTITIONED] | +| | | +| |−−11:MERGE | +| | | | +| | 08:AGGREGATE [MERGE FINALIZE] | +| | | output: sum(count(*)) | +| | | | +| | 07:EXCHANGE [PARTITION=UNPARTITIONED] | +| | | |--> +| | 04:AGGREGATE | +| | | output: count(*) | +| | | | +<b>| | 03:SCAN HBASE [hbase.hbase_table] |</b> +<b>| | start key: [email protected] |</b> +<b>| | stop key: [email protected]\0 |</b> +| | | +| 10:MERGE | +... +<!--| | | +| 06:AGGREGATE [MERGE FINALIZE] | +| | output: sum(count(*)) | +| | | +| 05:EXCHANGE [PARTITION=UNPARTITIONED] | +| | |--> +| 02:AGGREGATE | +| | output: count(*) | +| | | +<b>| 01:SCAN HBASE [hbase.hbase_table] |</b> +<b>| start key: [email protected] |</b> +<b>| stop key: [email protected]\0 |</b> ++------------------------------------------------------------------------------------+ +</codeblock> + + </example> + + <example> + + <title>Configuration Options for Java HBase Applications</title> + + <p> If you have an HBase Java application that calls the + <codeph>setCacheBlocks</codeph> or <codeph>setCaching</codeph> + methods of the class <xref + href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Scan.html"; + scope="external" format="html" + >org.apache.hadoop.hbase.client.Scan</xref>, you can set these same + caching behaviors through Impala query options, to control the memory + pressure on the HBase RegionServer. For example, when doing queries in + HBase that result in full-table scans (which by default are + inefficient for HBase), you can reduce memory usage and speed up the + queries by turning off the <codeph>HBASE_CACHE_BLOCKS</codeph> setting + and specifying a large number for the <codeph>HBASE_CACHING</codeph> + setting. + <!-- That recommendation w.r.t. full-table scans comes from the Cloudera HBase forum: http://community.cloudera.com/t5/Realtime-Random-Access-Apache/How-to-optimise-Full-Table-Scan-FTS-in-HBase/td-p/97 --> + </p> + + <p> + To set these options, issue commands like the following in <cmdname>impala-shell</cmdname>: + </p> + +<codeblock>-- Same as calling setCacheBlocks(true) or setCacheBlocks(false). +set hbase_cache_blocks=true; +set hbase_cache_blocks=false; + +-- Same as calling setCaching(rows). +set hbase_caching=1000; +</codeblock> + + <p> + Or update the <cmdname>impalad</cmdname> defaults file <filepath>/etc/default/impala</filepath> and + include settings for <codeph>HBASE_CACHE_BLOCKS</codeph> and/or <codeph>HBASE_CACHING</codeph> in the + <codeph>-default_query_options</codeph> setting for <codeph>IMPALA_SERVER_ARGS</codeph>. See + <xref href="impala_config_options.xml#config_options"/> for details. + </p> + + <note> + In Impala 2.0 and later, these options are settable through the JDBC or ODBC interfaces using the + <codeph>SET</codeph> statement. + </note> + + </example> + </conbody> + </concept> + + <concept id="hbase_scenarios"> + + <title>Use Cases for Querying HBase through Impala</title> + <prolog> + <metadata> + <data name="Category" value="Use Cases"/> + </metadata> + </prolog> + + <conbody> + + <p> + The following are popular use cases for using Impala to query HBase tables: + </p> + + <ul> + <li> + Keeping large fact tables in Impala, and smaller dimension tables in HBase. The fact tables use Parquet + or other binary file format optimized for scan operations. Join queries scan through the large Impala + fact tables, and cross-reference the dimension tables using efficient single-row lookups in HBase. + </li> + + <li> + Using HBase to store rapidly incrementing counters, such as how many times a web page has been viewed, or + on a social network, how many connections a user has or how many votes a post received. HBase is + efficient for capturing such changeable data: the append-only storage mechanism is efficient for writing + each change to disk, and a query always returns the latest value. An application could query specific + totals like these from HBase, and combine the results with a broader set of data queried from Impala. + </li> + + <li> + <p> + Storing very wide tables in HBase. Wide tables have many columns, possibly thousands, typically + recording many attributes for an important subject such as a user of an online service. These tables + are also often sparse, that is, most of the columns values are <codeph>NULL</codeph>, 0, + <codeph>false</codeph>, empty string, or other blank or placeholder value. (For example, any particular + web site user might have never used some site feature, filled in a certain field in their profile, + visited a particular part of the site, and so on.) A typical query against this kind of table is to + look up a single row to retrieve all the information about a specific subject, rather than summing, + averaging, or filtering millions of rows as in typical Impala-managed tables. + </p> + <p> + Or the HBase table could be joined with a larger Impala-managed table. For example, analyze the large + Impala table representing web traffic for a site and pick out 50 users who view the most pages. Join + that result with the wide user table in HBase to look up attributes of those users. The HBase side of + the join would result in 50 efficient single-row lookups in HBase, rather than scanning the entire user + table. + </p> + </li> + </ul> + </conbody> + </concept> + + <concept audience="Cloudera" id="hbase_create_new"> + + <title>Creating a New HBase Table for Impala to Use</title> + + <conbody> + + <p> + You can create an HBase-backed table through a <codeph>CREATE TABLE</codeph> statement in the Hive shell, + without going into the HBase shell at all: + </p> + + <!-- To do: + Add example. (Not critical because this subtopic is currently hidden.) + --> + </conbody> + </concept> + + <concept audience="Cloudera" id="hbase_reuse_existing"> + + <title>Associate Impala with an Existing HBase Table</title> + + <conbody> + + <p> + If you already have some HBase tables created through the HBase shell, you can make them accessible to + Impala through a <codeph>CREATE TABLE</codeph> statement in the Hive shell: + </p> + + <!-- To do: + Add example. (Not critical because this subtopic is currently hidden.) + --> </conbody> </concept> + <concept audience="Cloudera" id="hbase_column_families"> + + <title>Map HBase Columns and Column Families to Impala Columns</title> + + <conbody> + + <p/> + </conbody> + </concept> + + <concept id="hbase_loading"> + + <title>Loading Data into an HBase Table</title> + <prolog> + <metadata> + <data name="Category" value="ETL"/> + <data name="Category" value="Ingest"/> + </metadata> + </prolog> + + <conbody> + + <p> + The Impala <codeph>INSERT</codeph> statement works for HBase tables. The <codeph>INSERT ... VALUES</codeph> + syntax is ideally suited to HBase tables, because inserting a single row is an efficient operation for an + HBase table. (For regular Impala tables, with data files in HDFS, the tiny data files produced by + <codeph>INSERT ... VALUES</codeph> are extremely inefficient, so you would not use that technique with + tables containing any significant data volume.) + </p> + + <!-- To do: + Add examples throughout this section. + --> + + <p> + When you use the <codeph>INSERT ... SELECT</codeph> syntax, the result in the HBase table could be fewer + rows than you expect. HBase only stores the most recent version of each unique row key, so if an + <codeph>INSERT ... SELECT</codeph> statement copies over multiple rows containing the same value for the + key column, subsequent queries will only return one row with each key column value: + </p> + + <p> + Although Impala does not have an <codeph>UPDATE</codeph> statement, you can achieve the same effect by + doing successive <codeph>INSERT</codeph> statements using the same value for the key column each time: + </p> + + </conbody> + </concept> + + <concept id="hbase_limitations"> + + <title>Limitations and Restrictions of the Impala and HBase Integration</title> + + <conbody> + + <p> + The Impala integration with HBase has the following limitations and restrictions, some inherited from the + integration between HBase and Hive, and some unique to Impala: + </p> + + <ul> + <li> + <p> + If you issue a <codeph>DROP TABLE</codeph> for an internal (Impala-managed) table that is mapped to an + HBase table, the underlying table is not removed in HBase. The Hive <codeph>DROP TABLE</codeph> + statement also removes the HBase table in this case. + </p> + </li> + + <li> + <p> + The <codeph>INSERT OVERWRITE</codeph> statement is not available for HBase tables. You can insert new + data, or modify an existing row by inserting a new row with the same key value, but not replace the + entire contents of the table. You can do an <codeph>INSERT OVERWRITE</codeph> in Hive if you need this + capability. + </p> + </li> + + <li> + <p> + If you issue a <codeph>CREATE TABLE LIKE</codeph> statement for a table mapped to an HBase table, the + new table is also an HBase table, but inherits the same underlying HBase table name as the original. + The new table is effectively an alias for the old one, not a new table with identical column structure. + Avoid using <codeph>CREATE TABLE LIKE</codeph> for HBase tables, to avoid any confusion. + </p> + </li> + + <li> + <p> + Copying data into an HBase table using the Impala <codeph>INSERT ... SELECT</codeph> syntax might + produce fewer new rows than are in the query result set. If the result set contains multiple rows with + the same value for the key column, each row supercedes any previous rows with the same key value. + Because the order of the inserted rows is unpredictable, you cannot rely on this technique to preserve + the <q>latest</q> version of a particular key value. + </p> + </li> + <li rev="2.3.0"> + <p> + Because the complex data types (<codeph>ARRAY</codeph>, <codeph>STRUCT</codeph>, and <codeph>MAP</codeph>) + available in CDH 5.5 / Impala 2.3 and higher are currently only supported in Parquet tables, you cannot + use these types in HBase tables that are queried through Impala. + </p> + </li> + </ul> + </conbody> + </concept> + + <concept id="hbase_queries"> + + <title>Examples of Querying HBase Tables from Impala</title> + + <conbody> + + <p> + The following examples create an HBase table with four column families, + create a corresponding table through Hive, + then insert and query the table through Impala. + </p> + <p> + In HBase shell, the table + name is quoted in <codeph>CREATE</codeph> and <codeph>DROP</codeph> statements. Tables created in HBase + begin in <q>enabled</q> state; before dropping them through the HBase shell, you must issue a + <codeph>disable '<varname>table_name</varname>'</codeph> statement. + </p> + +<codeblock>$ hbase shell +15/02/10 16:07:45 +HBase Shell; enter 'help<RETURN>' for list of supported commands. +Type "exit<RETURN>" to leave the HBase Shell +Version 0.94.2-cdh4.2.0, rUnknown, Fri Feb 15 11:51:18 PST 2013 + +hbase(main):001:0> create 'hbasealltypessmall', 'boolsCF', 'intsCF', 'floatsCF', 'stringsCF' +0 row(s) in 4.6520 seconds + +=> Hbase::Table - hbasealltypessmall +hbase(main):006:0> quit +</codeblock> + + <p> + Issue the following <codeph>CREATE TABLE</codeph> statement in the Hive shell. (The Impala <codeph>CREATE + TABLE</codeph> statement currently does not support the <codeph>STORED BY</codeph> clause, so you switch into Hive to + create the table, then back to Impala and the <cmdname>impala-shell</cmdname> interpreter to issue the + queries.) + </p> + + <p> + This example creates an external table mapped to the HBase table, usable by both Impala and Hive. It is + defined as an external table so that when dropped by Impala or Hive, the original HBase table is not touched at all. + </p> + + <p> + The <codeph>WITH SERDEPROPERTIES</codeph> clause + specifies that the first column (<codeph>ID</codeph>) represents the row key, and maps the remaining + columns of the SQL table to HBase column families. The mapping relies on the ordinal order of the + columns in the table, not the column names in the <codeph>CREATE TABLE</codeph> statement. + The first column is defined to be the lookup key; the + <codeph>STRING</codeph> data type produces the fastest key-based lookups for HBase tables. + </p> + + <note> + For Impala with HBase tables, the most important aspect to ensure good performance is to use a + <codeph>STRING</codeph> column as the row key, as shown in this example. + </note> + +<codeblock>$ hive +Logging initialized using configuration in file:/etc/hive/conf.dist/hive-log4j.properties +Hive history file=/tmp/cloudera/hive_job_log_cloudera_201502101610_1980712808.txt +hive> use hbase; +OK +Time taken: 4.095 seconds +hive> CREATE EXTERNAL TABLE hbasestringids ( + > id string, + > bool_col boolean, + > tinyint_col tinyint, + > smallint_col smallint, + > int_col int, + > bigint_col bigint, + > float_col float, + > double_col double, + > date_string_col string, + > string_col string, + > timestamp_col timestamp) + > STORED BY 'org.apache.hadoop.hive.hbase.HBaseStorageHandler' + > WITH SERDEPROPERTIES ( + > "hbase.columns.mapping" = + > ":key,boolsCF:bool_col,intsCF:tinyint_col,intsCF:smallint_col,intsCF:int_col,intsCF:\ + > bigint_col,floatsCF:float_col,floatsCF:double_col,stringsCF:date_string_col,\ + > stringsCF:string_col,stringsCF:timestamp_col" + > ) + > TBLPROPERTIES("hbase.table.name" = "hbasealltypessmall"); +OK +Time taken: 2.879 seconds +hive> quit; +</codeblock> + + <p> + Once you have established the mapping to an HBase table, you can issue DML statements and queries + from Impala. The following example shows a series of <codeph>INSERT</codeph> + statements followed by a query. + The ideal kind of query from a performance standpoint + retrieves a row from the table based on a row key + mapped to a string column. + An initial <codeph>INVALIDATE METADATA <varname>table_name</varname></codeph> + statement makes the table created through Hive visible to Impala. + </p> + +<codeblock>$ impala-shell -i localhost -d hbase +Starting Impala Shell without Kerberos authentication +Connected to localhost:21000 +Server version: impalad version 2.1.0-cdh4 RELEASE (build d520a9cdea2fc97e8d5da9fbb0244e60ee416bfa) +Welcome to the Impala shell. Press TAB twice to see a list of available commands. + +Copyright (c) 2012 Cloudera, Inc. All rights reserved. + +(Shell build version: Impala Shell v2.1.0-cdh4 (d520a9c) built on Mon Dec 8 21:41:17 PST 2014) +Query: use `hbase` +[localhost:21000] > invalidate metadata hbasestringids; +Fetched 0 row(s) in 0.09s +[localhost:21000] > desc hbasestringids; ++-----------------+-----------+---------+ +| name | type | comment | ++-----------------+-----------+---------+ +| id | string | | +| bool_col | boolean | | +| double_col | double | | +| float_col | float | | +| bigint_col | bigint | | +| int_col | int | | +| smallint_col | smallint | | +| tinyint_col | tinyint | | +| date_string_col | string | | +| string_col | string | | +| timestamp_col | timestamp | | ++-----------------+-----------+---------+ +Fetched 11 row(s) in 0.02s +[localhost:21000] > insert into hbasestringids values ('0001',true,3.141,9.94,1234567,32768,4000,76,'2014-12-31','Hello world',now()); +Inserted 1 row(s) in 0.26s +[localhost:21000] > insert into hbasestringids values ('0002',false,2.004,6.196,1500,8000,129,127,'2014-01-01','Foo bar',now()); +Inserted 1 row(s) in 0.12s +[localhost:21000] > select * from hbasestringids where id = '0001'; ++------+----------+------------+-------------------+------------+---------+--------------+-------------+-----------------+-------------+-------------------------------+ +| id | bool_col | double_col | float_col | bigint_col | int_col | smallint_col | tinyint_col | date_string_col | string_col | timestamp_col | ++------+----------+------------+-------------------+------------+---------+--------------+-------------+-----------------+-------------+-------------------------------+ +| 0001 | true | 3.141 | 9.939999580383301 | 1234567 | 32768 | 4000 | 76 | 2014-12-31 | Hello world | 2015-02-10 16:36:59.764838000 | ++------+----------+------------+-------------------+------------+---------+--------------+-------------+-----------------+-------------+-------------------------------+ +Fetched 1 row(s) in 0.54s +</codeblock> + + <note conref="../shared/impala_common.xml#common/invalidate_metadata_hbase"/> +<!-- </section> --> + </conbody> + </concept> +</concept>
http://git-wip-us.apache.org/repos/asf/incubator-impala/blob/3c2c8f12/docs/topics/impala_hbase_cache_blocks.xml ---------------------------------------------------------------------- diff --git a/docs/topics/impala_hbase_cache_blocks.xml b/docs/topics/impala_hbase_cache_blocks.xml index d42cbf6..480b31c 100644 --- a/docs/topics/impala_hbase_cache_blocks.xml +++ b/docs/topics/impala_hbase_cache_blocks.xml @@ -3,11 +3,14 @@ <concept id="hbase_cache_blocks"> <title>HBASE_CACHE_BLOCKS Query Option</title> + <titlealts audience="PDF"><navtitle>HBASE_CACHE_BLOCKS</navtitle></titlealts> <prolog> <metadata> <data name="Category" value="Impala"/> <data name="Category" value="HBase"/> <data name="Category" value="Impala Query Options"/> + <data name="Category" value="Developers"/> + <data name="Category" value="Data Analysts"/> </metadata> </prolog> @@ -15,11 +18,14 @@ <p> <indexterm audience="Cloudera">HBASE_CACHE_BLOCKS query option</indexterm> - Setting this option is equivalent to calling the <codeph>setCacheBlocks</codeph> method of the class - <xref href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Scan.html"; scope="external" format="html">org.apache.hadoop.hbase.client.Scan</xref>, - in an HBase Java application. Helps to control the memory pressure on the HBase region server, in conjunction - with the <codeph>HBASE_CACHING</codeph> query option. - </p> + Setting this option is equivalent to calling the + <codeph>setCacheBlocks</codeph> method of the class <xref + href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Scan.html"; + scope="external" format="html" + >org.apache.hadoop.hbase.client.Scan</xref>, in an HBase Java + application. Helps to control the memory pressure on the HBase + RegionServer, in conjunction with the <codeph>HBASE_CACHING</codeph> query + option. </p> <p conref="../shared/impala_common.xml#common/type_boolean"/> <p conref="../shared/impala_common.xml#common/default_false_0"/> http://git-wip-us.apache.org/repos/asf/incubator-impala/blob/3c2c8f12/docs/topics/impala_hbase_caching.xml ---------------------------------------------------------------------- diff --git a/docs/topics/impala_hbase_caching.xml b/docs/topics/impala_hbase_caching.xml index e543792..03755fd 100644 --- a/docs/topics/impala_hbase_caching.xml +++ b/docs/topics/impala_hbase_caching.xml @@ -3,11 +3,14 @@ <concept id="hbase_caching"> <title>HBASE_CACHING Query Option</title> + <titlealts audience="PDF"><navtitle>HBASE_CACHING</navtitle></titlealts> <prolog> <metadata> <data name="Category" value="Impala"/> <data name="Category" value="HBase"/> <data name="Category" value="Impala Query Options"/> + <data name="Category" value="Developers"/> + <data name="Category" value="Data Analysts"/> </metadata> </prolog> @@ -15,11 +18,14 @@ <p> <indexterm audience="Cloudera">HBASE_CACHING query option</indexterm> - Setting this option is equivalent to calling the <codeph>setCaching</codeph> method of the class - <xref href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Scan.html"; scope="external" format="html">org.apache.hadoop.hbase.client.Scan</xref>, - in an HBase Java application. Helps to control the memory pressure on the HBase region server, in conjunction - with the <codeph>HBASE_CACHE_BLOCKS</codeph> query option. - </p> + Setting this option is equivalent to calling the + <codeph>setCaching</codeph> method of the class <xref + href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Scan.html"; + scope="external" format="html" + >org.apache.hadoop.hbase.client.Scan</xref>, in an HBase Java + application. Helps to control the memory pressure on the HBase + RegionServer, in conjunction with the <codeph>HBASE_CACHE_BLOCKS</codeph> + query option. </p> <p> <b>Type:</b> <codeph>BOOLEAN</codeph> http://git-wip-us.apache.org/repos/asf/incubator-impala/blob/3c2c8f12/docs/topics/impala_hints.xml ---------------------------------------------------------------------- diff --git a/docs/topics/impala_hints.xml b/docs/topics/impala_hints.xml index 429fb19..3eef3d3 100644 --- a/docs/topics/impala_hints.xml +++ b/docs/topics/impala_hints.xml @@ -3,7 +3,7 @@ <concept id="hints"> <title>Query Hints in Impala SELECT Statements</title> - <titlealts><navtitle>Hints</navtitle></titlealts> + <titlealts audience="PDF"><navtitle>Hints</navtitle></titlealts> <prolog> <metadata> <data name="Category" value="Impala"/> @@ -11,6 +11,8 @@ <data name="Category" value="Querying"/> <data name="Category" value="Performance"/> <data name="Category" value="Troubleshooting"/> + <data name="Category" value="Developers"/> + <data name="Category" value="Data Analysts"/> </metadata> </prolog> @@ -159,7 +161,7 @@ INSERT <varname>insert_clauses</varname> <xref href="impala_perf_joins.xml#straight_join"/> for details. When you use this technique, Impala does not reorder the joined tables at all, so you must be careful to arrange the join order to put the largest table (or subquery result set) first, then the smallest, second smallest, third smallest, and so on. This ordering lets Impala do the - most I/O-intensive parts of the query using local reads on the data nodes, and then reduce the size of the + most I/O-intensive parts of the query using local reads on the DataNodes, and then reduce the size of the intermediate result set as much as possible as each subsequent table or subquery result set is joined. </p> @@ -235,7 +237,7 @@ INSERT <varname>insert_clauses</varname> from t1 join <b>[shuffle]</b> t2 join <b>[broadcast]</b> t3 on t1.id = t2.id and t2.id = t3.id;</codeblock> - <draft-comment translate="no"> This is a good place to add more sample output showing before and after EXPLAIN plans. </draft-comment> + <!-- To do: This is a good place to add more sample output showing before and after EXPLAIN plans. --> <p conref="../shared/impala_common.xml#common/related_info"/> http://git-wip-us.apache.org/repos/asf/incubator-impala/blob/3c2c8f12/docs/topics/impala_identifiers.xml ---------------------------------------------------------------------- diff --git a/docs/topics/impala_identifiers.xml b/docs/topics/impala_identifiers.xml index 55477ed..7a947b3 100644 --- a/docs/topics/impala_identifiers.xml +++ b/docs/topics/impala_identifiers.xml @@ -3,7 +3,7 @@ <concept id="identifiers"> <title>Overview of Impala Identifiers</title> - <titlealts><navtitle>Identifiers</navtitle></titlealts> + <titlealts audience="PDF"><navtitle>Identifiers</navtitle></titlealts> <prolog> <metadata> <data name="Category" value="Impala"/> http://git-wip-us.apache.org/repos/asf/incubator-impala/blob/3c2c8f12/docs/topics/impala_impala_shell.xml ---------------------------------------------------------------------- diff --git a/docs/topics/impala_impala_shell.xml b/docs/topics/impala_impala_shell.xml index 3010fff..58a96bc 100644 --- a/docs/topics/impala_impala_shell.xml +++ b/docs/topics/impala_impala_shell.xml @@ -4,17 +4,98 @@ <title>Using the Impala Shell (impala-shell Command)</title> <titlealts audience="PDF"><navtitle>The Impala Shell</navtitle></titlealts> - + <prolog> + <metadata> + <data name="Category" value="Impala"/> + <data name="Category" value="impala-shell"/> + <data name="Category" value="SQL"/> + <data name="Category" value="Querying"/> + <data name="Category" value="Data Analysts"/> + <data name="Category" value="Developers"/> + <data name="Category" value="Stub Pages"/> + </metadata> + </prolog> + <conbody> <p> - + <indexterm audience="Cloudera">impala-shell</indexterm> You can use the Impala shell tool (<codeph>impala-shell</codeph>) to set up databases and tables, insert data, and issue queries. For ad hoc queries and exploration, you can submit SQL statements in an interactive session. To automate your work, you can specify command-line options to process a single statement or a - script file. + script file. The <cmdname>impala-shell</cmdname> interpreter accepts all the same SQL statements listed in + <xref href="impala_langref_sql.xml#langref_sql"/>, plus some shell-only commands that you can use for tuning + performance and diagnosing problems. + </p> + + <p> + The <codeph>impala-shell</codeph> command fits into the familiar Unix toolchain: + </p> + + <ul> + <li> + The <codeph>-q</codeph> option lets you issue a single query from the command line, without starting the + interactive interpreter. You could use this option to run <codeph>impala-shell</codeph> from inside a shell + script or with the command invocation syntax from a Python, Perl, or other kind of script. + </li> + + <li> + The <codeph>-f</codeph> option lets you process a file containing multiple SQL statements, + such as a set of reports or DDL statements to create a group of tables and views. + </li> + + <li rev="2.5.0 IMPALA-2179"> + The <codeph>--var</codeph> option lets you pass substitution variables to the statements that + are executed by that <cmdname>impala-shell</cmdname> session, for example the statements + in a script file processed by the <codeph>-f</codeph> option. You encode the substitution variable + on the command line using the notation + <codeph>--var=<varname>variable_name</varname>=<varname>value</varname></codeph>. + Within a SQL statement, you substitute the value by using the notation <codeph>${var:<varname>variable_name</varname>}</codeph>. + This feature is available in CDH 5.7 / Impala 2.5 and higher. + </li> + + <li> + The <codeph>-o</codeph> option lets you save query output to a file. + </li> + + <li> + The <codeph>-B</codeph> option turns off pretty-printing, so that you can produce comma-separated, + tab-separated, or other delimited text files as output. (Use the <codeph>--output_delimiter</codeph> option + to choose the delimiter character; the default is the tab character.) + </li> + + <li> + In non-interactive mode, query output is printed to <codeph>stdout</codeph> or to the file specified by the + <codeph>-o</codeph> option, while incidental output is printed to <codeph>stderr</codeph>, so that you can + process just the query output as part of a Unix pipeline. + </li> + + <li> + In interactive mode, <codeph>impala-shell</codeph> uses the <codeph>readline</codeph> facility to recall + and edit previous commands. + </li> + </ul> + + <p> + For information on installing the Impala shell, see <xref href="impala_install.xml#install"/>. In Cloudera + Manager 4.1 and higher, Cloudera Manager installs <cmdname>impala-shell</cmdname> automatically. You might + install <cmdname>impala-shell</cmdname> manually on other systems not managed by Cloudera Manager, so that + you can issue queries from client systems that are not also running the Impala daemon or other Apache Hadoop + components. + </p> + + <p> + For information about establishing a connection to a DataNode running the <codeph>impalad</codeph> daemon + through the <codeph>impala-shell</codeph> command, see <xref href="impala_connecting.xml#connecting"/>. + </p> + + <p> + For a list of the <codeph>impala-shell</codeph> command-line options, see + <xref href="impala_shell_options.xml#shell_options"/>. For reference information about the + <codeph>impala-shell</codeph> interactive commands, see + <xref href="impala_shell_commands.xml#shell_commands"/>. </p> - + <p outputclass="toc"/> </conbody> </concept>
