6/13/13 HBaseCon 2013
2
SQL
3
MySQL
Storage Engine
Optimizer
Parser
Local Disk
Honeycomb
open-source MySQL storage engine
backed by HBase
4
MySQL
Honeycomb
HBase
Features
6
Honeycomb InnoDB
Cross table joins ✓ ✓
Compound indices ✓ ✓
Unique indices ✓ ✓
Auto Increment Columns ✓ ✓
Store...
7
Honeycomb HBase Schema
8
Metadata
Rows
Indices
Honeycomb HBase Schema
Row format
9
MySQL row[empty]6 table id UUID
RowKey Qualifier Value
variable length encoded
unsigned integer
serialized Avr...
Index Format
10
1 or more
variable length encoded
unsigned integer
serialized Avro
data type
single byte prefix
7 table id ...
Data Duplication
11
duplication factor = 1 + 2i
where i = number of indices
Performance
12
Range Limit
Relative Scan
Performance
10 Rows 75%
100 Rows 68%
1000 Rows 51%
13
nearinfinity.github.io/honeycomb
github.com/honeycomb
Upcoming SlideShare
Loading in...5
×

HBaseCon 2013: Honeycomb - MySQL Backed by Apache HBase

1,150

Published on

Presented by Dan Burkert, Near Infinity

Published in: Technology

HBaseCon 2013: Honeycomb - MySQL Backed by Apache HBase

  1. 1. 6/13/13 HBaseCon 2013
  2. 2. 2 SQL
  3. 3. 3 MySQL Storage Engine Optimizer Parser Local Disk
  4. 4. Honeycomb open-source MySQL storage engine backed by HBase 4
  5. 5. MySQL Honeycomb HBase
  6. 6. Features 6 Honeycomb InnoDB Cross table joins ✓ ✓ Compound indices ✓ ✓ Unique indices ✓ ✓ Auto Increment Columns ✓ ✓ Stored procedures & views ✓ ✓ Transactions ✓ Foreign key constraints ✓ Automatic sharding ✓ Replication & failover ✓ Map/reduce integration ✓ Map/reduce bulkload ✓
  7. 7. 7 Honeycomb HBase Schema
  8. 8. 8 Metadata Rows Indices Honeycomb HBase Schema
  9. 9. Row format 9 MySQL row[empty]6 table id UUID RowKey Qualifier Value variable length encoded unsigned integer serialized Avro data type single byte prefix
  10. 10. Index Format 10 1 or more variable length encoded unsigned integer serialized Avro data type single byte prefix 7 table id index id UUIDnull indicator value MySQL row[empty] Qualifier ValueRowKey
  11. 11. Data Duplication 11 duplication factor = 1 + 2i where i = number of indices
  12. 12. Performance 12 Range Limit Relative Scan Performance 10 Rows 75% 100 Rows 68% 1000 Rows 51%
  13. 13. 13 nearinfinity.github.io/honeycomb github.com/honeycomb

×