Download as PDF, PPTX
![© 2016 Dremio Corporation @DremioHQ
Columnar
data
persons = [{
name: ’Joe',
age: 18,
phones: [
‘555-‐111-‐1111’,
‘555-‐222-‐2222’
]
}, {
name: ’Jack',
age: 37,
phones: [ ‘555-‐333-‐3333’ ]
}]
J
o
e
J
a
c
k
0
3
7
Offset Values
Name
18
37
Age
5
5
5
-
1
1
1
0
12
24
-
Str.
Offset
Values
Phones
1
1
1
1
5
5
5
-
2
2
2
-
…
36
0
2
3
4
List
Offset
Values](https://image.slidesharecdn.com/stratalondonparquetarrow-160602155004/75/Strata-London-2016-The-future-of-column-oriented-data-processing-with-Arrow-and-Parquet-16-2048.jpg){kind=icon}
![© 2016 Dremio Corporation @DremioHQ
Record
Batch
Construction
Schema
Negotiation
Dictionary
Batch
Record
Batch
Record
Batch
Record
Batch
name
(offset)
name
(data)
age
(data)
phones
(list
offset)
phones
(data)
data
header
(describes
offsets
into
data)
name
(bitmap)
age
(bitmap)
phones
(bitmap)
phones
(offset)
{
name: ’Joe',
age: 18,
phones: [
‘555-‐111-‐1111’,
‘555-‐222-‐2222’
]
}
Each
box
(vector)
is
contiguous
memory
The
entire
record
batch
is
contiguous
on
wire](https://image.slidesharecdn.com/stratalondonparquetarrow-160602155004/75/Strata-London-2016-The-future-of-column-oriented-data-processing-with-Arrow-and-Parquet-28-2048.jpg){kind=icon}
![© 2016 Dremio Corporation @DremioHQ
Columnar
data
persons = [{
name: ’Joe',
age: 18,
phones: [
‘555-‐111-‐1111’,
‘555-‐222-‐2222’
]
}, {
name: ’Jack',
age: 37,
phones: [ ‘555-‐333-‐3333’ ]
}]
J
o
e
J
a
c
k
0
3
7
Offset Values
Name
18
37
Age
5
5
5
-
1
1
1
0
12
24
-
Str.
Offset
Values
Phones
1
1
1
1
5
5
5
-
2
2
2
-
…
36
0
2
3
4
List
Offset
Values](https://crownmelresort.com/image.slidesharecdn.com/stratalondonparquetarrow-160602155004/75/Strata-London-2016-The-future-of-column-oriented-data-processing-with-Arrow-and-Parquet-16-2048.jpg){kind=icon}
![© 2016 Dremio Corporation @DremioHQ
Record
Batch
Construction
Schema
Negotiation
Dictionary
Batch
Record
Batch
Record
Batch
Record
Batch
name
(offset)
name
(data)
age
(data)
phones
(list
offset)
phones
(data)
data
header
(describes
offsets
into
data)
name
(bitmap)
age
(bitmap)
phones
(bitmap)
phones
(offset)
{
name: ’Joe',
age: 18,
phones: [
‘555-‐111-‐1111’,
‘555-‐222-‐2222’
]
}
Each
box
(vector)
is
contiguous
memory
The
entire
record
batch
is
contiguous
on
wire](https://crownmelresort.com/image.slidesharecdn.com/stratalondonparquetarrow-160602155004/75/Strata-London-2016-The-future-of-column-oriented-data-processing-with-Arrow-and-Parquet-28-2048.jpg){kind=icon}
Uploaded byJulien Le Dem
Strata London 2016: The future of column oriented data processing with Arrow and Parquet
The document discusses the future of column-oriented data processing using Apache Arrow and Parquet, highlighting their benefits for on-disk and in-memory formats. It emphasizes the importance of a community-driven standard for interoperability and the advantages of columnar formats for improving CPU efficiency and reducing serialization overhead. Additionally, it outlines the goals of the Apache Arrow project and its ecosystem, focusing on enhancing performance and facilitating data exchange between different systems.
More Related Content
Similar to Strata London 2016: The future of column oriented data processing with Arrow and Parquet
Strata London 2016: The future of column oriented data processing with Arrow and Parquet
- 1. © 2016 Dremio Corporation @DremioHQ The future of column oriented data processing with Arrow and Parquet Julien Le Dem Principal Architect, Dremio VP Apache Parquet, Apache Arrow PMC
- 2. © 2016 Dremio Corporation @DremioHQ • Architect at @DremioHQ • Formerly Tech Lead at Twitter on Data Platforms. • Creator of Parquet • Apache member • Apache PMCs: Arrow, Incubator, Pig, Parquet Julien Le Dem @J_ Julien
- 3. © 2016 Dremio Corporation @DremioHQ Agenda • Benefits of Columnar formats – On disk (Apache Parquet) – In memory (Apache Arrow) • Community Driven Standard • Interoperability and Ecosystem
- 4. © 2016 Dremio Corporation @DremioHQ Benefits of Columnar formats @EmrgencyKittens
- 5. © 2016 Dremio Corporation @DremioHQ Columnar layout Logical table representation Row layout Column layout
- 6. © 2016 Dremio Corporation @DremioHQ On Disk and in Memory • Different trade offs – On disk: Storage. • Accessed by multiple queries. • Priority to I/O reduction (but still needs good CPU throughput). • Mostly Streaming access. – In memory: Transient. • Specific to one query execution. • Priority to CPU throughput (but still needs good I/O). • Streaming and Random access.
- 7. © 2016 Dremio Corporation @DremioHQ Parquet on disk columnar format
- 8. © 2016 Dremio Corporation @DremioHQ Parquet on disk columnar format • Nested data structures • Compact format: – type aware encodings – better compression • Optimized I/O: – Projection push down (column pruning) – Predicate push down (filters based on stats)
- 9. © 2016 Dremio Corporation @DremioHQ Access only the data you need a b c a1 b1 c1 a2 b2 c2 a3 b3 c3 a4 b4 c4 a5 b5 c5 a b c a1 b1 c1 a2 b2 c2 a3 b3 c3 a4 b4 c4 a5 b5 c5 a b c a1 b1 c1 a2 b2 c2 a3 b3 c3 a4 b4 c4 a5 b5 c5 + = Columnar Statistics Read only the data you need!
- 10. © 2016 Dremio Corporation @DremioHQ Parquet nested representation Document DocId Links Name Backward Forward Language Url Code Country Columns: docid links.backward links.forward name.language.code name.language.country name.url Borrowed from the Google Dremel paper https://blog.twitter.com/2013/dremel-made-simple-with-parquet
- 11. © 2016 Dremio Corporation @DremioHQ Arrow in memory columnar format
- 12. © 2016 Dremio Corporation @DremioHQ Arrow in memory columnar format • Nested Data Structures • Maximize CPU throughput – Pipelining – SIMD – cache locality • Scatter/gather I/O
- 13. © 2016 Dremio Corporation @DremioHQ CPU pipeline
- 14. © 2016 Dremio Corporation @DremioHQ Minimize CPU cache misses a cache miss costs 10 to 100s cycles depending on the level
- 15. © 2016 Dremio Corporation @DremioHQ Focus on CPU Efficiency 1331246660 1331246351 1331244570 1331261196 3/8/2012 2:44PM 3/8/2012 2:38PM 3/8/2012 2:09PM 3/8/2012 6:46PM 99.155.155.225 65.87.165.114 71.10.106.181 76.102.156.138 Row 1 Row 2 Row 3 Row 4 1331246660 1331246351 1331244570 1331261196 3/8/2012 2:44PM 3/8/2012 2:38PM 3/8/2012 2:09PM 3/8/2012 6:46PM 99.155.155.225 65.87.165.114 71.10.106.181 76.102.156.138 session_id timestamp source_ip Traditional Memory Buffer Arrow Memory Buffer • Cache Locality • Super-‐scalar & vectorized operation • Minimal Structure Overhead • Constant value access – With minimal structure overhead • Operate directly on columnar compressed data
- 16. © 2016 Dremio Corporation @DremioHQ Columnar data persons = [{ name: ’Joe', age: 18, phones: [ ‘555-‐111-‐1111’, ‘555-‐222-‐2222’ ] }, { name: ’Jack', age: 37, phones: [ ‘555-‐333-‐3333’ ] }] J o e J a c k 0 3 7 Offset Values Name 18 37 Age 5 5 5 - 1 1 1 0 12 24 - Str. Offset Values Phones 1 1 1 1 5 5 5 - 2 2 2 - … 36 0 2 3 4 List Offset Values
- 17. © 2016 Dremio Corporation @DremioHQ Java: Memory Management • Chunk-‐based managed allocator – Built on top of Netty’s JEMalloc implementation • Create a tree of allocators – Limit and transfer semantics across allocators – Leak detection and location accounting • Wrap native memory from other applications
- 18. © 2016 Dremio Corporation @DremioHQ Community Driven Standard
- 19. © 2016 Dremio Corporation @DremioHQ An open source standard • Arrow: Common need for in memory columnar. • Benefits: – Share the effort – Create an ecosystem • Building on the success of Parquet. • Standard from the start
- 20. © 2016 Dremio Corporation @DremioHQ Shared Need > Open Source Opportunity “We are also considering switching to a columnar canonical in-‐memory format for data that needs to be materialized during query processing, in order to take advantage of SIMD instructions” -‐Impala Team “A large fraction of the CPU time is spent waiting for data to be fetched from main memory…we are designing cache-‐friendly algorithms and data structures so Spark applications will spend less time waiting to fetch data from memory and more time doing useful work” – Spark Team “Drill provides a flexible hierarchical columnar data model that can represent complex, highly dynamic and evolving data models and allows efficient processing of it without need to flatten or materialize.” -‐Drill Team
- 21. © 2016 Dremio Corporation @DremioHQ Arrow goals • Well-‐documented and cross language compatible • Designed to take advantage of modern CPU characteristics • Embeddable in execution engines, storage layers, etc. • Interoperable
- 22. © 2016 Dremio Corporation @DremioHQ The Apache Arrow Project • New Top-‐level Apache Software Foundation project – Announced Feb 17, 2016 • Focused on Columnar In-‐Memory Analytics 1. 10-‐100x speedup on many workloads 2. Common data layer enables companies to choose best of breed systems 3. Designed to work with any programming language 4. Support for both relational and complex data as-‐is • Developers from 13+ major open source projects involved – A significant % of the world’s data will be processed through Arrow! Calcite Cassandra Deeplearning4j Drill Hadoop HBase Ibis Impala Kudu Pandas Parquet Phoenix Spark Storm R
- 23. © 2016 Dremio Corporation @DremioHQ Interoperability and Ecosystem
- 24. © 2016 Dremio Corporation @DremioHQ High Performance Sharing & Interchange Today With Arrow • Each system has its own internal memory format • 70-‐80% CPU wasted on serialization and deserialization • Functionality duplication and unnecessary conversions • All systems utilize the same memory format • No overhead for cross-‐system communication • Projects can share functionality (eg: Parquet-‐to-‐Arrow reader) Pandas Drill Impala HBase KuduCassandra Parquet Spark Arrow Memory Pandas Drill Impala HBase KuduCassandra Parquet Spark Copy & Convert Copy & Convert Copy & Convert Copy & Convert Copy & Convert
- 25. © 2016 Dremio Corporation @DremioHQ Language Bindings Parquet • Target Languages – Java – CPP (underway) – Python & Pandas (underway) Arrow • Target Languages – Java (beta) – CPP (underway) – Python & Pandas (underway) – R – Julia • Initial Focus – Read a structure – Write a structure – Manage Memory
- 26. © 2016 Dremio Corporation @DremioHQ RPC & IPC
- 27. © 2016 Dremio Corporation @DremioHQ Common Message Pattern • Schema Negotiation – Logical Description of structure – Identification of dictionary encoded Nodes • Dictionary Batch – Dictionary ID, Values • Record Batch – Batches of records up to 64K – Leaf nodes up to 2B values Schema Negotiation Dictionary Batch Record Batch Record Batch Record Batch 1..N Batches 0..N Batches
- 28. © 2016 Dremio Corporation @DremioHQ Record Batch Construction Schema Negotiation Dictionary Batch Record Batch Record Batch Record Batch name (offset) name (data) age (data) phones (list offset) phones (data) data header (describes offsets into data) name (bitmap) age (bitmap) phones (bitmap) phones (offset) { name: ’Joe', age: 18, phones: [ ‘555-‐111-‐1111’, ‘555-‐222-‐2222’ ] } Each box (vector) is contiguous memory The entire record batch is contiguous on wire
- 29. © 2016 Dremio Corporation @DremioHQ Moving Data Between Systems RPC • Avoid Serialization & Deserialization • Layer TBD: Focused on supporting vectored io – Scatter/gather reads/writes against socket IPC • Alpha implementation using memory mapped files – Moving data between Python and Drill • Working on shared allocation approach – Shared reference counting and well-‐defined ownership semantics
- 30. © 2016 Dremio Corporation @DremioHQ Example data exchanges:
- 31. © 2016 Dremio Corporation @DremioHQ RPC: Query execution Immutable Arrow Batch Scanner Scanner Scanner Parquet files projection push down read only a and b Partial Agg Partial Agg Partial Agg Agg Agg Agg Shuffle Arrow batches Actor Result SELECT SUM(a) FROM t GROUP BY b The memory representation is sent over the wire. No serialization overhead.
- 32. © 2016 Dremio Corporation @DremioHQ IPC: Python with Spark or Drill SQL engine Python process User defined function SQL Operator 1 SQL Operator 2 reads reads Immutable Arrow Batch
- 33. © 2016 Dremio Corporation @DremioHQ What’s Next • Parquet – Arrow conversion for Python & C++ • Arrow IPC Implementation • Apache {Spark, Drill} to Arrow Integration – Faster UDFs, Storage interfaces • Support for integration with Intel’s Persistent Memory library via Apache Mnemonic
- 34. © 2016 Dremio Corporation @DremioHQ Get Involved • Join the community – dev@arrow.apache.org, dev@parquet.apache.org – Slack: https://apachearrowslackin.herokuapp.com/ – http://arrow.apache.orghttp://parquet.apache.org – Follow @ApacheParquet, @ApacheArrow