The reason for writing this blog is that I tried to use Vectorized Reader in Hive, but faced some problems with its documentation. That's why I've decided to write this blog.

Vectorized query execution is a Hive feature that greatly reduces the CPU usage for typical query operations like scans, filters, aggregates, and joins. A standard query execution system processes one row at a time. This involves long code paths and significant metadata interpretation in the inner loop of execution. Vectorized query execution streamlines operations by processing a block of 1,024 rows at a time. Within the block, each column is stored as a vector (an array of a primitive data type). Simple operations like arithmetic and comparisons are done by quickly iterating through the vectors in a tight loop, with no or very few function calls or conditional branches inside the loop.

To use vectorized query execution, you must store your data in ORC format like so:

set hive.vectorized.execution.enabled = true ;

How to Query

To use vectorized query execution, you must store your data in ORC format. Just follow the below steps:

Start Hive CLI and create an ORC table with some data:

hive> create table vectortable(id int) stored as orc;
                OK
Time taken: 0.487 seconds
hive>set hive.vectorized.execution.enabled = true;

hive> insert into vectortable values(1);

Query ID = hduser_20170713203731_09db3954-246b-4b23-8d34-1d9d7b62965c

Total jobs = 1 Launching Job 1 out of 1 Number of reduce tasks is set to 0 since there's no reduce operator Job running in-process (local Hadoop)

2017-07-13 20:37:33,237 Stage-1 map = 100%, reduce = 0% Ended Job = job_local722393542_0002

Stage-4 is selected by condition resolver.

Stage-3 is filtered out by condition resolver.

Stage-5 is filtered out by condition resolver. Moving data to: hdfs://localhost:54310/user/hive/warehouse/vectortable/.hive-staging_hive_2017-07-13_20-37-31_172_3262390557269287245-1/-ext-10000

Loading data to table default.vectortable Table default.vectortable stats: [numFiles=1, numRows=1, totalSize=199, rawDataSize=4]

MapReduce Jobs Launched: Stage-Stage-1: HDFS Read: 321 HDFS Write: 545 SUCCESS Total MapReduce CPU Time Spent: 0 msec OK Time taken: 2.672 seconds

Now, query the table with the explain command to see whether Have is using vectorized execution or not.

Note: When Fetch is used in the plan instead of Map, it does not vectorize. So, first set hive.fetch.task.conversion=none:

hive> explain select id from vectortable where id>=1;
OK
STAGE DEPENDENCIES:
Stage-1 is a root stage
Stage-0 depends on stages: Stage-1

STAGE PLANS:
Stage: Stage-1
Map Reduce
Map Operator Tree:
TableScan
alias: vectortable
Statistics: Num rows: 1 Data size: 4 Basic stats: COMPLETE Column stats: NONE
Filter Operator
predicate: (id >= 1) (type: boolean)
Statistics: Num rows: 1 Data size: 4 Basic stats: COMPLETE Column stats: NONE
Select Operator
expressions: id (type: int)
outputColumnNames: _col0
Statistics: Num rows: 1 Data size: 4 Basic stats: COMPLETE Column stats: NONE
File Output Operator
compressed: false
Statistics: Num rows: 1 Data size: 4 Basic stats: COMPLETE Column stats: NONE
table:
input format: org.apache.hadoop.mapred.TextInputFormat
output format: org.apache.hadoop.hive.ql.io.HiveIgnoreKeyTextOutputFormat
serde: org.apache.hadoop.hive.serde2.lazy.LazySimpleSerDe
Execution mode: vectorized

Stage: Stage-0
Fetch Operator
limit: -1
Processor Tree:
ListSink

Time taken: 0.081 seconds, Fetched: 33 row(s)

As you can see in the explain command, Execution mode: vectorized is printed and is enabled for the query.