[feature](step:one)Add arrow format #161
Conversation
| private RecordBatch(Iterator<InternalRow> iterator, String format, String sep, byte[] delim, | ||
| private VectorSchemaRoot arrowRoot = null; | ||
|
|
||
| private int arrowBatchSize = 1000; |
There was a problem hiding this comment.
Can this variable be configured? Or is this the best performance value obtained through testing?
There was a problem hiding this comment.
This is the number obtained after comparing the performance of arrows when testing the performance.
However, there may be other scenarios, and perhaps a configuration item can be added
spark-doris-connector/src/main/scala/org/apache/spark/sql/doris/spark/ArrowSchemaUtils.scala
Show resolved
Hide resolved
spark-doris-connector/src/main/scala/org/apache/spark/sql/doris/spark/ArrowSchemaUtils.scala
Show resolved
Hide resolved
spark-doris-connector/src/main/java/org/apache/doris/spark/load/DorisStreamLoad.java
Show resolved
Hide resolved
|
There are some code conflicts, please fix them, thank you |
wuwenchi
force-pushed
the
add_arrow_format
branch
from
a9d37eb
to
29912e8
Compare
wuwenchi
force-pushed
the
add_arrow_format
branch
from
29912e8
to
94d7bb6
Compare
| @@ -72,7 +72,7 @@ | |||
| <spark.major.version>3.1</spark.major.version> | |||
| <scala.version>2.12</scala.version> | |||
| <libthrift.version>0.16.0</libthrift.version> | |||
| <arrow.version>5.0.0</arrow.version> | |||
| <arrow.version>13.0.0</arrow.version> | |||
| byte[] bytes = new byte[1]; | ||
| int read = read(bytes, 0, 1); | ||
| if (read < 0) { | ||
| return -1; | ||
| } else { | ||
| return bytes[0]; | ||
| } |
There was a problem hiding this comment.
use same logic from read(x,x,x)
| @@ -31,10 +31,8 @@ | |||
| import java.nio.charset.StandardCharsets; | |||
| import java.util.HashMap; | |||
| import java.util.Map; | |||
| import java.util.Objects; | |||
There was a problem hiding this comment.
seem useless, delete...
| import java.util.stream.Collectors; | ||
| import java.util.stream.IntStream; | ||
| import java.util.stream.Stream; |
There was a problem hiding this comment.
seem useless, delete...
|
|
||
| object ArrowSchemaUtils { | ||
| def toArrowSchema(schema: StructType, timeZoneId: String): Schema = { | ||
| ArrowUtils.toArrowSchema(schema, timeZoneId) |
There was a problem hiding this comment.
This is a direct call to the Spark function, because ArrowUtils is private in org.apache.spark.sql
Proposed changes
Issue Number: close #xxx
Problem Summary:
Describe the overview of changes.
Checklist(Required)
Further comments
If this is a relatively large or complex change, kick off the discussion at dev@doris.apache.org by explaining why you chose the solution you did and what alternatives you considered, etc...