- Pig
- Procedural data-flow language
- Pig is used by Programmers and Researchers.
- Pig is on the client side.
- For managing and querying unstructured data.
- Hive
- Declarative SQLish Language
- Hive is used by analysts generating data reports.
- Hive is on cluster side.
- For managing and querying structured data.
Features
|
Hive
|
Pig
|
Language
|
SQL-like
|
PigLatin
|
Schemas/Type
|
Yes (explicit)
We have to create "tables" beforehand and stores the schema in a either shared or local database for metadata.
|
Yes (implicit)
No need to create table.
|
Partitions
|
Yes
|
No
|
Server
|
Optional (Thrift)
|
No
|
UDF
|
Yes (Java)
|
Yes (Java)
|
Custom Serialize/Deserializer
|
Yes
|
Yes
|
DFS Direct Access
|
Yes (implicit)
We never point to the actual HDFS folder.
|
Yes (explict)
We explicitly point to HDFS folder.
|
Join/Order/Sort
|
Yes
|
Yes
|
Shell
|
Yes
|
Yes
|
Streaming
|
Yes
|
Yes
|
WebInterface
|
Yes
|
No
|
JDBC/ODBC
|
Yes (limited)
|
No
|
No comments:
Post a Comment