ORC Files
Hive’s RCFile has been the standard format for storing Hive data for the last 3 years. However, RCFile has limitations because it treats each column as a binary blob without semantics. The upcoming Hive 0.11 will add a new file format named Optimized Row Columnar (ORC) file that uses and retains the type information from the table definition. ORC uses type specific readers and writers that provide light weight compression techniques such as dictionary encoding, bit packing, delta encoding, and run length encoding -- resulting in dramatically smaller files. Additionally, ORC can apply generic compression using zlib, LZO, or Snappy on top of the lightweight compression for even smaller files. However, storage savings are only part of the gain. ORC supports projection, which selects subsets of the columns for reading, so that queries reading only one column read only the required bytes. Furthermore, ORC files include light weight indexes that include the minimum and maximum values for each column in each set of 10,000 rows and the entire file. Using pushdown filters from Hive, the file reader can skip entire sets of rows that aren’t important for this query. Finally, ORC works together with the upcoming query vectorization work providing a high bandwidth reader/writer interface.
Recommended
More Related Content
What's hot
Similar to ORC Files
More from Owen O'Malley
ORC Files
- 1. © Hortonworks Inc. 2012 ORC Files June 2013 Page 1 Owen O’Malley owen@hortonworks.com @owen_omalley owen@hortonworks.com
- 2. © Hortonworks Inc. 2012 Who Am I? Page 2
- 3. © Hortonworks Inc. 2012 History Page 3
- 4. © Hortonworks Inc. 2012 Remaining Challenges Page 4
- 5. © Hortonworks Inc. 2012 Requirements Page 5
- 6. © Hortonworks Inc. 2012 File Structure Page 6
- 7. © Hortonworks Inc. 2012 Stripe Structure Page 7
- 8. © Hortonworks Inc. 2012 File Layout Page 8 File Footer Postscript Index Data Row Data Stripe Footer 256MBStripe Index Data Row Data Stripe Footer 256MBStripe Index Data Row Data Stripe Footer 256MBStripe Column 1 Column 2 Column 7 Column 8 Column 3 Column 6 Column 4 Column 5 Column 1 Column 2 Column 7 Column 8 Column 3 Column 6 Column 4 Column 5 Stream 2.1 Stream 2.2 Stream 2.3 Stream 2.4
- 9. © Hortonworks Inc. 2012 Compression Page 9
- 10. © Hortonworks Inc. 2012 Integer Column Serialization Page 10
- 11. © Hortonworks Inc. 2012 String Column Serialization Page 11
- 12. © Hortonworks Inc. 2012 Hive Compound Types Page 12 0 Struct 4 Struct 3 String 1 Int 2 Map 7 Time 5 String 6 Double
- 13. © Hortonworks Inc. 2012 Compound Type Serialization Page 13
- 14. © Hortonworks Inc. 2012 Generic Compression Page 14
- 15. © Hortonworks Inc. 2012 Column Projection Page 15
- 16. © Hortonworks Inc. 2012 How Do You Use ORC Page 16
- 17. © Hortonworks Inc. 2012 Managing Memory Page 17
- 18. © Hortonworks Inc. 2012 Pavan’s Trick Page 18
- 19. © Hortonworks Inc. 2012 Looking at ORC File Structures Page 19
- 20. © Hortonworks Inc. 2012 Looking at ORC File Structures Page 20
- 21. © Hortonworks Inc. 2012 TPC-DS File Sizes Page 21
- 22. © Hortonworks Inc. 2012 TPC-DS Query Performance Page 22
- 23. © Hortonworks Inc. 2012 Additional Details Page 23
- 24. © Hortonworks Inc. 2012 Current work Page 24
- 25. © Hortonworks Inc. 2012 Vectorization Page 25
- 26. © Hortonworks Inc. 2012 Vectorization Preliminary Results Page 26
- 27. © Hortonworks Inc. 2012 Future Work Page 27
- 28. © Hortonworks Inc. 2012 Thanks! Page 28
- 29. © Hortonworks Inc. 2012 Comparison Page 29 RC File Trevni Parquet ORC File Hive Type Model N N N Y Separate complex columns N Y Y Y Splits found quickly N Y Y Y Default column group size 4MB 64MB* 64MB* 256MB Files per a bucket 1 > 1 1* 1 Store min, max, sum, count N N N Y Versioned metadata N Y Y Y Run length data encoding N N Y Y Store strings in dictionary N N N Y Store row count N Y N Y Skip compressed blocks N N N Y Store internal indexes N N N Y