Apache Phoenix Performance Result :: Mon Mar 24 11:27:10 PDT 2014

Queries by Category


Aggregation

select count(*) from AGGREGATION_TABLE
0.95s 2.2.3
0.6s 3.0-RC1

select count(1) from AGGREGATION_TABLE group by HOST
0.97s 2.2.3
0.56s 3.0-RC1

select core, count(1) from AGGREGATION_TABLE group by CORE
1.21s 2.2.3
0.71s 3.0-RC1

select core, host, sum(ACTIVE_VISITOR) from AGGREGATION_TABLE group by core, host
2.05s 2.2.3
2.11s 3.0-RC1

select AVG(DB), MIN(CORE), MAX(ACTIVE_VISITOR) from AGGREGATION_TABLE where host='CS' and DOMAIN<>'Google.com'
2.57s 2.2.3
2.62s 3.0-RC1

select round(date, 'day',50), sum(core)+2 from AGGREGATION_TABLE where (CORE<10 or CORE>90) group by round(date, 'day',50), host
1.02s 2.2.3
0.92s 3.0-RC1

select trunc(date, 'hour', 1000), count(DB), count(ACTIVE_VISITOR) from AGGREGATION_TABLE where feature='Login' and (core<50 or db<20 or active_visitor > 200) group by trunc(date, 'hour', 1000)
1.28s 2.2.3
1.27s 3.0-RC1

select count(active_visitor) count from AGGREGATION_TABLE
1.62s 2.2.3
1.7s 3.0-RC1

select count(distinct active_visitor) count from AGGREGATION_TABLE
1.9s 2.2.3
2s 3.0-RC1

select count(core), count(db), count(active_visitor) from AGGREGATION_TABLE
2.31s 2.2.3
2.23s 3.0-RC1

select core, count(db) count from AGGREGATION_TABLE group by CORE
1.64s 2.2.3
0.77s 3.0-RC1

select count(core) from AGGREGATION_TABLE where core < 10 and db < 200
0.96s 2.2.3
0.82s 3.0-RC1

select count(core) from AGGREGATION_TABLE where core < 10 and active_visitor < 1000
0.77s 2.2.3
0.83s 3.0-RC1

select count(core) from AGGREGATION_TABLE where core > 90
0.84s 2.2.3
0.74s 3.0-RC1

select count(*) from WIDE_PK
2.15s 2.2.3
1.94s 3.0-RC1


Aggregation with filter no non-pk

select count(*) from TABLE_6CF where f < 10000
2.22s 2.2.3
2.08s 3.0-RC1

select count(*) from TABLE_1CF where f < 10000
2.17s 2.2.3
1.73s 3.0-RC1

select count(*) from WIDE_PK where column2<10
3.38s 2.2.3
2.23s 3.0-RC1

select count(*) from WIDE_PK where column2<50 and column3<50 and column4<50 and column5<50
4.79s 2.2.3
2.68s 3.0-RC1


ColumnFamilies-Single

select count(1) from TABLE_6CF where K1 in ('A', 'B', 'C', 'D')
1.44s 2.2.3
0.74s 3.0-RC1

select count(1) from TABLE_6CF where K1 in ('A', 'C')
1.35s 2.2.3
0.66s 3.0-RC1

select count(*) from TABLE_6CF where f < 1000
1.26s 2.2.3
1.2s 3.0-RC1

select a,b,c,d,e,f from TABLE_6CF where B>1000 and B<2000 and f>1000 and f<2000
1.31s 2.2.3
1.22s 3.0-RC1

select sum(a),sum(b),sum(c),sum(d) from TABLE_6CF group by K1
5.28s 2.2.3
5.01s 3.0-RC1

select count(1) from TABLE_1CF where K1 in ('A', 'B', 'C', 'D')
1.59s 2.2.3
1.56s 3.0-RC1

select count(1) from TABLE_1CF where K1 in ('A', 'C')
1.56s 2.2.3
1.46s 3.0-RC1

select count(*) from TABLE_1CF where f < 1000
1.83s 2.2.3
1.85s 3.0-RC1

select a,b,c,d,e,f from TABLE_1CF where B>1000 and B<2000 and f>1000 and f<2000
2.21s 2.2.3
2.18s 3.0-RC1

select sum(a),sum(b),sum(c),sum(d) from TABLE_1CF group by K1
3.82s 2.2.3
2.02s 3.0-RC1


Count Distinct

select count(distinct core), count(distinct db), count(distinct active_visitor) from AGGREGATION_TABLE
2.7s 2.2.3
2.72s 3.0-RC1

select core, count(distinct db) count from AGGREGATION_TABLE group by CORE
2.67s 2.2.3
1.96s 3.0-RC1


IN/LIKE Clause

select count(core) from AGGREGATION_TABLE where core IN (1,25,50,75,100)
0.76s 2.2.3
0.72s 3.0-RC1

select count(1) from IN_LIKE_TABLE where K1 in ('Z', 'D', 'F', 'X')
0.74s 2.2.3
0.61s 3.0-RC1

select * from IN_LIKE_TABLE where K1 in ('A','B') and K2 in ('xxx1000000','xxx2000000','xxx2000001')
0s 2.2.3
0.01s 3.0-RC1

select * from IN_LIKE_TABLE where K1 in ('A', 'C', 'E', 'F', 'H') and K2 like 'xxxxxx1%'
0.01s 2.2.3
0.02s 3.0-RC1

select count(*) from IN_LIKE_TABLE where K1 in ('X','A','B','C','D','E','F','G','H','I','Z') and K2 like '%1%'
1.06s 2.2.3
1.09s 3.0-RC1

select count(*) from IN_LIKE_TABLE where (K1 in ('A','E','F','G','H','I','Z') or K1 like 'E%') and (K2 like '%xx1%' or K2 like '%xxxxxx2%')
1.14s 2.2.3
1.09s 3.0-RC1


Index

select count(*) from INDEXED_TABLE
0.74s 2.2.3
0.38s 3.0-RC1

select /*+ INDEX(INDEXED_TABLE idx1 idx2 idx3 idx4) */ count(*) from INDEXED_TABLE
2.22s 2.2.3
0.85s 3.0-RC1

select /*+ INDEX(INDEXED_TABLE ids1 ids2 ids3 ids4) */ count(*) from INDEXED_TABLE
0.65s 2.2.3
0.25s 3.0-RC1

select /*+NO_INDEX*/ count(*) from INDEXED_TABLE
1.3s 2.2.3
0.69s 3.0-RC1

select round(date, 'day',50), sum(core)+2 from INDEXED_TABLE where (CORE<10 or CORE>90) group by round(date, 'day',50), host
0.24s 2.2.3
0.16s 3.0-RC1

select /*+NO_INDEX*/ round(date, 'day',50), sum(core)+2 from INDEXED_TABLE where (CORE<10 or CORE>90) group by round(date, 'day',50), host
1.45s 2.2.3
1.27s 3.0-RC1

select count(core) from INDEXED_TABLE where core IN (1,100)
0.1s 2.2.3
0.11s 3.0-RC1

select /*+ INDEX(INDEXED_TABLE idx1 idx2 idx3 idx4) */ count(core) from INDEXED_TABLE where core IN (1,100)
0.08s 2.2.3
0.04s 3.0-RC1

select /*+ INDEX(INDEXED_TABLE ids1 ids2 ids3 ids4) */ count(core) from INDEXED_TABLE where core IN (1,100)
0.06s 2.2.3
0.09s 3.0-RC1

select /*+NO_INDEX*/ count(core) from INDEXED_TABLE where core IN (1,100)
1.01s 2.2.3
0.91s 3.0-RC1

select count(core) from INDEXED_TABLE where core IN (1,25,50,75,100)
0.13s 2.2.3
0.13s 3.0-RC1

select /*+ INDEX(INDEXED_TABLE idx1 idx2 idx3 idx4) */ count(core) from INDEXED_TABLE where core IN (1,25,50,75,100)
0.16s 2.2.3
0.1s 3.0-RC1

select /*+ INDEX(INDEXED_TABLE ids1 ids2 ids3 ids4) */ count(core) from INDEXED_TABLE where core IN (1,25,50,75,100)
0.11s 2.2.3
0.1s 3.0-RC1

select /*+NO_INDEX*/ count(core) from INDEXED_TABLE where core IN (1,25,50,75,100)
1.03s 2.2.3
0.93s 3.0-RC1

select count(core) from INDEXED_TABLE where core < 10 and db < 200
0.18s 2.2.3
0.19s 3.0-RC1

select /*+ INDEX(INDEXED_TABLE idx1 idx2 idx3 idx4) */ count(core) from INDEXED_TABLE where core < 10 and db < 200
0.78s 2.2.3
0.66s 3.0-RC1

select /*+ INDEX(INDEXED_TABLE ids1 ids2 ids3 ids4) */ count(core) from INDEXED_TABLE where core < 10 and db < 200
0.16s 2.2.3
0.18s 3.0-RC1

select /*+NO_INDEX*/ count(core) from INDEXED_TABLE where core < 10 and db < 200
1.27s 2.2.3
1.15s 3.0-RC1

select count(core) from INDEXED_TABLE where core < 100 and db < 2000
0.93s 2.2.3
0.54s 3.0-RC1

select /*+NO_INDEX*/ count(core) from INDEXED_TABLE where core < 100 and db < 2000
2.76s 2.2.3
1.41s 3.0-RC1

select count(core) from INDEXED_TABLE where core > 90
0.18s 2.2.3
0.13s 3.0-RC1

select /*+NO_INDEX*/ count(core) from INDEXED_TABLE where core > 90
1.15s 2.2.3
1.04s 3.0-RC1

select count(core) from INDEXED_TABLE where core > 0
0.88s 2.2.3
0.44s 3.0-RC1

select /*+NO_INDEX*/ count(core) from INDEXED_TABLE where core > 0
1.86s 2.2.3
1.13s 3.0-RC1

select db from INDEXED_TABLE where db >5 and db <7 order by db
0.06s 2.2.3
0.07s 3.0-RC1

select /*+NO_INDEX*/ db from INDEXED_TABLE where db >5 and db <7 order by db
1.77s 2.2.3
1.96s 3.0-RC1

select db from INDEXED_TABLE order by db limit 10
0.05s 2.2.3
0.1s 3.0-RC1

select /*+NO_INDEX*/ db from INDEXED_TABLE order by db limit 10
2.28s 2.2.3
1.42s 3.0-RC1

select count(*) from INDEXED_TABLE_MUTABLE
1.22s 2.2.3
0.86s 3.0-RC1

select /*+ INDEX(INDEXED_TABLE_MUTABLE midx1 midx2 midx3 midx4) */ count(*) from INDEXED_TABLE_MUTABLE
2.07s 2.2.3
0.85s 3.0-RC1

select /*+ INDEX(INDEXED_TABLE_MUTABLE mids1 mids2 mids3 mids4) */ count(*) from INDEXED_TABLE_MUTABLE
0.85s 2.2.3
0.46s 3.0-RC1

select /*+NO_INDEX*/ count(*) from INDEXED_TABLE_MUTABLE
1.53s 2.2.3
0.79s 3.0-RC1

select round(date, 'day',50), sum(core)+2 from INDEXED_TABLE_MUTABLE where (CORE<10 or CORE>90) group by round(date, 'day',50), host
0.49s 2.2.3
0.38s 3.0-RC1

select /*+NO_INDEX*/ round(date, 'day',50), sum(core)+2 from INDEXED_TABLE_MUTABLE where (CORE<10 or CORE>90) group by round(date, 'day',50), host
1.66s 2.2.3
1.36s 3.0-RC1

select count(core) from INDEXED_TABLE_MUTABLE where core IN (1,100)
0.17s 2.2.3
0.15s 3.0-RC1

select /*+ INDEX(INDEXED_TABLE_MUTABLE midx1 midx2 midx3 midx4) */ count(core) from INDEXED_TABLE_MUTABLE where core IN (1,100)
0.08s 2.2.3
0.04s 3.0-RC1

select /*+ INDEX(INDEXED_TABLE_MUTABLE mids1 mids2 mids3 mids4) */ count(core) from INDEXED_TABLE_MUTABLE where core IN (1,100)
0.14s 2.2.3
0.12s 3.0-RC1

select /*+NO_INDEX*/ count(core) from INDEXED_TABLE_MUTABLE where core IN (1,100)
1.2s 2.2.3
0.96s 3.0-RC1

select count(core) from INDEXED_TABLE_MUTABLE where core IN (1,25,50,75,100)
0.2s 2.2.3
0.17s 3.0-RC1

select /*+ INDEX(INDEXED_TABLE_MUTABLE midx1 midx2 midx3 midx4) */ count(core) from INDEXED_TABLE_MUTABLE where core IN (1,25,50,75,100)
0.12s 2.2.3
0.08s 3.0-RC1

select /*+ INDEX(INDEXED_TABLE_MUTABLE mids1 mids2 mids3 mids4) */ count(core) from INDEXED_TABLE_MUTABLE where core IN (1,25,50,75,100)
0.16s 2.2.3
0.12s 3.0-RC1

select /*+NO_INDEX*/ count(core) from INDEXED_TABLE_MUTABLE where core IN (1,25,50,75,100)
1.22s 2.2.3
0.97s 3.0-RC1

select count(core) from INDEXED_TABLE_MUTABLE where core < 10 and db < 200
0.42s 2.2.3
0.42s 3.0-RC1

select /*+ INDEX(INDEXED_TABLE_MUTABLE midx1 midx2 midx3 midx4) */ count(core) from INDEXED_TABLE_MUTABLE where core < 10 and db < 200
0.4s 2.2.3
0.4s 3.0-RC1

select /*+ INDEX(INDEXED_TABLE_MUTABLE mids1 mids2 mids3 mids4) */ count(core) from INDEXED_TABLE_MUTABLE where core < 10 and db < 200
0.41s 2.2.3
0.4s 3.0-RC1

select /*+NO_INDEX*/ count(core) from INDEXED_TABLE_MUTABLE where core < 10 and db < 200
1.5s 2.2.3
1.19s 3.0-RC1

select count(core) from INDEXED_TABLE_MUTABLE where core < 100 and db < 2000
1.34s 2.2.3
0.97s 3.0-RC1

select /*+NO_INDEX*/ count(core) from INDEXED_TABLE_MUTABLE where core < 100 and db < 2000
3.17s 2.2.3
1.55s 3.0-RC1

select count(core) from INDEXED_TABLE_MUTABLE where core > 90
0.23s 2.2.3
0.19s 3.0-RC1

select /*+NO_INDEX*/ count(core) from INDEXED_TABLE_MUTABLE where core > 90
1.34s 2.2.3
1.07s 3.0-RC1

select count(core) from INDEXED_TABLE_MUTABLE where core > 0
0.92s 2.2.3
0.56s 3.0-RC1

select /*+NO_INDEX*/ count(core) from INDEXED_TABLE_MUTABLE where core > 0
2.16s 2.2.3
1.21s 3.0-RC1

select db from INDEXED_TABLE_MUTABLE where db >5 and db <7 order by db
0.18s 2.2.3
0.16s 3.0-RC1

select /*+NO_INDEX*/ db from INDEXED_TABLE_MUTABLE where db >5 and db <7 order by db
1.98s 2.2.3
2.08s 3.0-RC1

select db from INDEXED_TABLE_MUTABLE order by db limit 10
1.03s 2.2.3
1.22s 3.0-RC1

select /*+NO_INDEX*/ db from INDEXED_TABLE_MUTABLE order by db limit 10
2.64s 2.2.3
1.44s 3.0-RC1


Index Creation After Data

CREATE INDEX idx5 ON INDEXED_TABLE (CORE) INCLUDE (DB,ACTIVE_VISITOR)
34.97s 2.2.3
34.95s 3.0-RC1

CREATE INDEX midx5 ON INDEXED_TABLE_MUTABLE (CORE) INCLUDE (DB,ACTIVE_VISITOR)
36.43s 2.2.3
37.75s 3.0-RC1


Large Group By

select count(*) from TABLE_6CF group by A
1.58s 2.2.3
0.83s 3.0-RC1

select count(*) from TABLE_6CF group by B
2.49s 2.2.3
2.54s 3.0-RC1

select count(*) from TABLE_1CF group by A
1.82s 2.2.3
1.58s 3.0-RC1

select count(*) from TABLE_1CF group by B
2.26s 2.2.3
1.71s 3.0-RC1


Load Data

LOAD DATA AGGREGATION_TABLE [2000000 ROWS]
95.32s 2.2.3
113.67s 3.0-RC1

LOAD DATA IN_LIKE_TABLE [2000000 ROWS]
82.37s 2.2.3
90.53s 3.0-RC1

LOAD DATA INDEXED_TABLE [2000000 ROWS]
438.38s 2.2.3
449.14s 3.0-RC1

LOAD DATA INDEXED_TABLE_MUTABLE [2000000 ROWS]
639.02s 2.2.3
629.08s 3.0-RC1

LOAD DATA TABLE_6CF [2000000 ROWS]
96.88s 2.2.3
104.05s 3.0-RC1

LOAD DATA SALTED_TABLE [2000000 ROWS]
94.95s 2.2.3
107.1s 3.0-RC1

LOAD DATA TABLE_1CF [2000000 ROWS]
91.83s 2.2.3
97.33s 3.0-RC1

LOAD DATA WIDE_PK [2000000 ROWS]
215.7s 2.2.3
227.82s 3.0-RC1


Percentile

select percentile_cont(0.9) within group (order by active_visitor asc) from AGGREGATION_TABLE
1.88s 2.2.3
1.99s 3.0-RC1

select percentile_cont(0.1) within group (order by active_visitor desc) from AGGREGATION_TABLE
1.9s 2.2.3
1.94s 3.0-RC1

select percentile_cont(0.5) within group (order by core asc), percentile_cont(0.5) within group (order by db asc), percentile_cont(0.5) within group (order by active_visitor asc) from AGGREGATION_TABLE
2.75s 2.2.3
2.75s 3.0-RC1

select core, percentile_cont(0.9999) within group (order by db asc) from AGGREGATION_TABLE group by CORE
2.71s 2.2.3
1.92s 3.0-RC1


Salting

select count(1) from SALTED_TABLE where K1 in ('A', 'B', 'C', 'D')
1.6s 2.2.3
1.05s 3.0-RC1

select count(*) from SALTED_TABLE where D < 1000
0.99s 2.2.3
1.08s 3.0-RC1

select avg(c),avg(d) from SALTED_TABLE group by K1
3.54s 2.2.3
3.73s 3.0-RC1




Queries by Table



AGGREGATION_TABLE


Table used for various aggregation functions

ROWS 2000000 [2M]

DDL CREATE TABLE $TABLE (HOST CHAR(2) NOT NULL,DOMAIN VARCHAR NOT NULL,FEATURE VARCHAR NOT NULL,DATE DATE NOT NULL,USAGE.CORE BIGINT,USAGE.DB BIGINT,STATS.ACTIVE_VISITOR INTEGER CONSTRAINT PK PRIMARY KEY (HOST, DOMAIN, FEATURE, DATE)) SPLIT ON ('CSGoogle','CSSalesforce','EUApple','EUGoogle','EUSalesforce','NAApple','NAGoogle','NASalesforce')

DATA GENERATOR
STRING :: Values: NA|CS|EU
STRING :: Values: Salesforce.com|Apple.com|Google.com|Yelp.com
STRING :: Values: Login|Report|Dashboard|Sales|UI
DATE :: Values: SEQUENTIAL
INTEGER :: Max. Value: 100 Values: RANDOM
INTEGER :: Max. Value: 2000 Values: RANDOM
INTEGER :: Max. Value: 10000 Values: RANDOM

select count(*) from AGGREGATION_TABLE
0.95s 2.2.3
0.6s 3.0-RC1

select count(1) from AGGREGATION_TABLE group by HOST
0.97s 2.2.3
0.56s 3.0-RC1

select core, count(1) from AGGREGATION_TABLE group by CORE
1.21s 2.2.3
0.71s 3.0-RC1

select core, host, sum(ACTIVE_VISITOR) from AGGREGATION_TABLE group by core, host
2.05s 2.2.3
2.11s 3.0-RC1

select AVG(DB), MIN(CORE), MAX(ACTIVE_VISITOR) from AGGREGATION_TABLE where host='CS' and DOMAIN<>'Google.com'
2.57s 2.2.3
2.62s 3.0-RC1

select round(date, 'day',50), sum(core)+2 from AGGREGATION_TABLE where (CORE<10 or CORE>90) group by round(date, 'day',50), host
1.02s 2.2.3
0.92s 3.0-RC1

select trunc(date, 'hour', 1000), count(DB), count(ACTIVE_VISITOR) from AGGREGATION_TABLE where feature='Login' and (core<50 or db<20 or active_visitor > 200) group by trunc(date, 'hour', 1000)
1.28s 2.2.3
1.27s 3.0-RC1

select count(active_visitor) count from AGGREGATION_TABLE
1.62s 2.2.3
1.7s 3.0-RC1

select count(distinct active_visitor) count from AGGREGATION_TABLE
1.9s 2.2.3
2s 3.0-RC1

select percentile_cont(0.9) within group (order by active_visitor asc) from AGGREGATION_TABLE
1.88s 2.2.3
1.99s 3.0-RC1

select percentile_cont(0.1) within group (order by active_visitor desc) from AGGREGATION_TABLE
1.9s 2.2.3
1.94s 3.0-RC1

select count(core), count(db), count(active_visitor) from AGGREGATION_TABLE
2.31s 2.2.3
2.23s 3.0-RC1

select count(distinct core), count(distinct db), count(distinct active_visitor) from AGGREGATION_TABLE
2.7s 2.2.3
2.72s 3.0-RC1

select percentile_cont(0.5) within group (order by core asc), percentile_cont(0.5) within group (order by db asc), percentile_cont(0.5) within group (order by active_visitor asc) from AGGREGATION_TABLE
2.75s 2.2.3
2.75s 3.0-RC1

select core, count(db) count from AGGREGATION_TABLE group by CORE
1.64s 2.2.3
0.77s 3.0-RC1

select core, count(distinct db) count from AGGREGATION_TABLE group by CORE
2.67s 2.2.3
1.96s 3.0-RC1

select core, percentile_cont(0.9999) within group (order by db asc) from AGGREGATION_TABLE group by CORE
2.71s 2.2.3
1.92s 3.0-RC1

select count(core) from AGGREGATION_TABLE where core IN (1,25,50,75,100)
0.76s 2.2.3
0.72s 3.0-RC1

select count(core) from AGGREGATION_TABLE where core < 10 and db < 200
0.96s 2.2.3
0.82s 3.0-RC1

select count(core) from AGGREGATION_TABLE where core < 10 and active_visitor < 1000
0.77s 2.2.3
0.83s 3.0-RC1

select count(core) from AGGREGATION_TABLE where core > 90
0.84s 2.2.3
0.74s 3.0-RC1

LOAD DATA AGGREGATION_TABLE [2000000 ROWS]
95.32s 2.2.3
113.67s 3.0-RC1


IN_LIKE_TABLE


Phoenix 1.2+ uses skip scan that gives significant performance improvement for IN/LIKE queries.

ROWS 2000000 [2M]

DDL CREATE TABLE IF NOT EXISTS $TABLE (K1 CHAR(1) NOT NULL, K2 VARCHAR NOT NULL, CF.A INTEGER, CF.B VARCHAR, CF.C VARCHAR CONSTRAINT PK PRIMARY KEY (K1,K2)) SPLIT ON ('B','C','D','E','F','G','H','I')

DATA GENERATOR
STRING :: Values: A|B|C|D|E|F|G|H|I
STRING :: Length: 10 Values: SEQUENTIAL
INTEGER :: Max. Value: 1000000 Values: RANDOM
STRING :: Length: 50 Values: RANDOM
STRING :: Length: 50 Values: RANDOM

select count(1) from IN_LIKE_TABLE where K1 in ('Z', 'D', 'F', 'X')
0.74s 2.2.3
0.61s 3.0-RC1

select * from IN_LIKE_TABLE where K1 in ('A','B') and K2 in ('xxx1000000','xxx2000000','xxx2000001')
0s 2.2.3
0.01s 3.0-RC1

select * from IN_LIKE_TABLE where K1 in ('A', 'C', 'E', 'F', 'H') and K2 like 'xxxxxx1%'
0.01s 2.2.3
0.02s 3.0-RC1

select count(*) from IN_LIKE_TABLE where K1 in ('X','A','B','C','D','E','F','G','H','I','Z') and K2 like '%1%'
1.06s 2.2.3
1.09s 3.0-RC1

select count(*) from IN_LIKE_TABLE where (K1 in ('A','E','F','G','H','I','Z') or K1 like 'E%') and (K2 like '%xx1%' or K2 like '%xxxxxx2%')
1.14s 2.2.3
1.09s 3.0-RC1

LOAD DATA IN_LIKE_TABLE [2000000 ROWS]
82.37s 2.2.3
90.53s 3.0-RC1


INDEXED_TABLE


Indexing test. Phoenix 2.0+ feature. See explain plan by hovering over version label to see which index table is used.

ROWS 2000000 [2M]

DDL CREATE TABLE $TABLE (HOST CHAR(2) NOT NULL,DOMAIN VARCHAR NOT NULL,FEATURE VARCHAR NOT NULL,DATE DATE NOT NULL,USAGE.CORE BIGINT,USAGE.DB BIGINT,STATS.ACTIVE_VISITOR INTEGER CONSTRAINT PK PRIMARY KEY (HOST, DOMAIN, FEATURE, DATE)) IMMUTABLE_ROWS=true,MAX_FILESIZE=30485760;CREATE INDEX idx1 ON $TABLE (CORE);CREATE INDEX idx2 ON $TABLE (DB);CREATE INDEX idx3 ON $TABLE (DB,ACTIVE_VISITOR);CREATE INDEX idx4 ON $TABLE (CORE,DB,ACTIVE_VISITOR);CREATE INDEX ids1 ON $TABLE (CORE) SALT_BUCKETS=16;CREATE INDEX ids2 ON $TABLE (DB) SALT_BUCKETS=16;CREATE INDEX ids3 ON $TABLE (DB,ACTIVE_VISITOR) SALT_BUCKETS=16;CREATE INDEX ids4 ON $TABLE (CORE,DB,ACTIVE_VISITOR) SALT_BUCKETS=16;

DATA GENERATOR
STRING :: Values: NA|CS|EU
STRING :: Values: Salesforce.com|Apple.com|Google.com|Yelp.com
STRING :: Values: Login|Report|Dashboard|Sales|UI
DATE :: Values: SEQUENTIAL
INTEGER :: Max. Value: 100 Values: RANDOM
INTEGER :: Max. Value: 2000 Values: RANDOM
INTEGER :: Max. Value: 10000 Values: RANDOM

select count(*) from INDEXED_TABLE
0.74s 2.2.3
0.38s 3.0-RC1

select /*+ INDEX(INDEXED_TABLE idx1 idx2 idx3 idx4) */ count(*) from INDEXED_TABLE
2.22s 2.2.3
0.85s 3.0-RC1

select /*+ INDEX(INDEXED_TABLE ids1 ids2 ids3 ids4) */ count(*) from INDEXED_TABLE
0.65s 2.2.3
0.25s 3.0-RC1

select /*+NO_INDEX*/ count(*) from INDEXED_TABLE
1.3s 2.2.3
0.69s 3.0-RC1

select round(date, 'day',50), sum(core)+2 from INDEXED_TABLE where (CORE<10 or CORE>90) group by round(date, 'day',50), host
0.24s 2.2.3
0.16s 3.0-RC1

select /*+NO_INDEX*/ round(date, 'day',50), sum(core)+2 from INDEXED_TABLE where (CORE<10 or CORE>90) group by round(date, 'day',50), host
1.45s 2.2.3
1.27s 3.0-RC1

select count(core) from INDEXED_TABLE where core IN (1,100)
0.1s 2.2.3
0.11s 3.0-RC1

select /*+ INDEX(INDEXED_TABLE idx1 idx2 idx3 idx4) */ count(core) from INDEXED_TABLE where core IN (1,100)
0.08s 2.2.3
0.04s 3.0-RC1

select /*+ INDEX(INDEXED_TABLE ids1 ids2 ids3 ids4) */ count(core) from INDEXED_TABLE where core IN (1,100)
0.06s 2.2.3
0.09s 3.0-RC1

select /*+NO_INDEX*/ count(core) from INDEXED_TABLE where core IN (1,100)
1.01s 2.2.3
0.91s 3.0-RC1

select count(core) from INDEXED_TABLE where core IN (1,25,50,75,100)
0.13s 2.2.3
0.13s 3.0-RC1

select /*+ INDEX(INDEXED_TABLE idx1 idx2 idx3 idx4) */ count(core) from INDEXED_TABLE where core IN (1,25,50,75,100)
0.16s 2.2.3
0.1s 3.0-RC1

select /*+ INDEX(INDEXED_TABLE ids1 ids2 ids3 ids4) */ count(core) from INDEXED_TABLE where core IN (1,25,50,75,100)
0.11s 2.2.3
0.1s 3.0-RC1

select /*+NO_INDEX*/ count(core) from INDEXED_TABLE where core IN (1,25,50,75,100)
1.03s 2.2.3
0.93s 3.0-RC1

select count(core) from INDEXED_TABLE where core < 10 and db < 200
0.18s 2.2.3
0.19s 3.0-RC1

select /*+ INDEX(INDEXED_TABLE idx1 idx2 idx3 idx4) */ count(core) from INDEXED_TABLE where core < 10 and db < 200
0.78s 2.2.3
0.66s 3.0-RC1

select /*+ INDEX(INDEXED_TABLE ids1 ids2 ids3 ids4) */ count(core) from INDEXED_TABLE where core < 10 and db < 200
0.16s 2.2.3
0.18s 3.0-RC1

select /*+NO_INDEX*/ count(core) from INDEXED_TABLE where core < 10 and db < 200
1.27s 2.2.3
1.15s 3.0-RC1

select count(core) from INDEXED_TABLE where core < 100 and db < 2000
0.93s 2.2.3
0.54s 3.0-RC1

select /*+NO_INDEX*/ count(core) from INDEXED_TABLE where core < 100 and db < 2000
2.76s 2.2.3
1.41s 3.0-RC1

select count(core) from INDEXED_TABLE where core > 90
0.18s 2.2.3
0.13s 3.0-RC1

select /*+NO_INDEX*/ count(core) from INDEXED_TABLE where core > 90
1.15s 2.2.3
1.04s 3.0-RC1

select count(core) from INDEXED_TABLE where core > 0
0.88s 2.2.3
0.44s 3.0-RC1

select /*+NO_INDEX*/ count(core) from INDEXED_TABLE where core > 0
1.86s 2.2.3
1.13s 3.0-RC1

select db from INDEXED_TABLE where db >5 and db <7 order by db
0.06s 2.2.3
0.07s 3.0-RC1

select /*+NO_INDEX*/ db from INDEXED_TABLE where db >5 and db <7 order by db
1.77s 2.2.3
1.96s 3.0-RC1

select db from INDEXED_TABLE order by db limit 10
0.05s 2.2.3
0.1s 3.0-RC1

select /*+NO_INDEX*/ db from INDEXED_TABLE order by db limit 10
2.28s 2.2.3
1.42s 3.0-RC1

CREATE INDEX idx5 ON INDEXED_TABLE (CORE) INCLUDE (DB,ACTIVE_VISITOR)
34.97s 2.2.3
34.95s 3.0-RC1

LOAD DATA INDEXED_TABLE [2000000 ROWS]
438.38s 2.2.3
449.14s 3.0-RC1


INDEXED_TABLE_MUTABLE


Indexing test. Phoenix 2.0+ feature. See explain plan by hovering over version label to see which index table is used.

ROWS 2000000 [2M]

DDL CREATE TABLE $TABLE (HOST CHAR(2) NOT NULL,DOMAIN VARCHAR NOT NULL,FEATURE VARCHAR NOT NULL,DATE DATE NOT NULL,USAGE.CORE BIGINT,USAGE.DB BIGINT,STATS.ACTIVE_VISITOR INTEGER CONSTRAINT PK PRIMARY KEY (HOST, DOMAIN, FEATURE, DATE)) MAX_FILESIZE=30485760;CREATE INDEX midx1 ON $TABLE (CORE);CREATE INDEX midx2 ON $TABLE (DB);CREATE INDEX midx3 ON $TABLE (DB,ACTIVE_VISITOR);CREATE INDEX midx4 ON $TABLE (CORE,DB,ACTIVE_VISITOR);CREATE INDEX mids1 ON $TABLE (CORE) SALT_BUCKETS=16;CREATE INDEX mids2 ON $TABLE (DB) SALT_BUCKETS=16;CREATE INDEX mids3 ON $TABLE (DB,ACTIVE_VISITOR) SALT_BUCKETS=16;CREATE INDEX mids4 ON $TABLE (CORE,DB,ACTIVE_VISITOR) SALT_BUCKETS=16;

DATA GENERATOR
STRING :: Values: NA|CS|EU
STRING :: Values: Salesforce.com|Apple.com|Google.com|Yelp.com
STRING :: Values: Login|Report|Dashboard|Sales|UI
DATE :: Values: SEQUENTIAL
INTEGER :: Max. Value: 100 Values: RANDOM
INTEGER :: Max. Value: 2000 Values: RANDOM
INTEGER :: Max. Value: 10000 Values: RANDOM

select count(*) from INDEXED_TABLE_MUTABLE
1.22s 2.2.3
0.86s 3.0-RC1

select /*+ INDEX(INDEXED_TABLE_MUTABLE midx1 midx2 midx3 midx4) */ count(*) from INDEXED_TABLE_MUTABLE
2.07s 2.2.3
0.85s 3.0-RC1

select /*+ INDEX(INDEXED_TABLE_MUTABLE mids1 mids2 mids3 mids4) */ count(*) from INDEXED_TABLE_MUTABLE
0.85s 2.2.3
0.46s 3.0-RC1

select /*+NO_INDEX*/ count(*) from INDEXED_TABLE_MUTABLE
1.53s 2.2.3
0.79s 3.0-RC1

select round(date, 'day',50), sum(core)+2 from INDEXED_TABLE_MUTABLE where (CORE<10 or CORE>90) group by round(date, 'day',50), host
0.49s 2.2.3
0.38s 3.0-RC1

select /*+NO_INDEX*/ round(date, 'day',50), sum(core)+2 from INDEXED_TABLE_MUTABLE where (CORE<10 or CORE>90) group by round(date, 'day',50), host
1.66s 2.2.3
1.36s 3.0-RC1

select count(core) from INDEXED_TABLE_MUTABLE where core IN (1,100)
0.17s 2.2.3
0.15s 3.0-RC1

select /*+ INDEX(INDEXED_TABLE_MUTABLE midx1 midx2 midx3 midx4) */ count(core) from INDEXED_TABLE_MUTABLE where core IN (1,100)
0.08s 2.2.3
0.04s 3.0-RC1

select /*+ INDEX(INDEXED_TABLE_MUTABLE mids1 mids2 mids3 mids4) */ count(core) from INDEXED_TABLE_MUTABLE where core IN (1,100)
0.14s 2.2.3
0.12s 3.0-RC1

select /*+NO_INDEX*/ count(core) from INDEXED_TABLE_MUTABLE where core IN (1,100)
1.2s 2.2.3
0.96s 3.0-RC1

select count(core) from INDEXED_TABLE_MUTABLE where core IN (1,25,50,75,100)
0.2s 2.2.3
0.17s 3.0-RC1

select /*+ INDEX(INDEXED_TABLE_MUTABLE midx1 midx2 midx3 midx4) */ count(core) from INDEXED_TABLE_MUTABLE where core IN (1,25,50,75,100)
0.12s 2.2.3
0.08s 3.0-RC1

select /*+ INDEX(INDEXED_TABLE_MUTABLE mids1 mids2 mids3 mids4) */ count(core) from INDEXED_TABLE_MUTABLE where core IN (1,25,50,75,100)
0.16s 2.2.3
0.12s 3.0-RC1

select /*+NO_INDEX*/ count(core) from INDEXED_TABLE_MUTABLE where core IN (1,25,50,75,100)
1.22s 2.2.3
0.97s 3.0-RC1

select count(core) from INDEXED_TABLE_MUTABLE where core < 10 and db < 200
0.42s 2.2.3
0.42s 3.0-RC1

select /*+ INDEX(INDEXED_TABLE_MUTABLE midx1 midx2 midx3 midx4) */ count(core) from INDEXED_TABLE_MUTABLE where core < 10 and db < 200
0.4s 2.2.3
0.4s 3.0-RC1

select /*+ INDEX(INDEXED_TABLE_MUTABLE mids1 mids2 mids3 mids4) */ count(core) from INDEXED_TABLE_MUTABLE where core < 10 and db < 200
0.41s 2.2.3
0.4s 3.0-RC1

select /*+NO_INDEX*/ count(core) from INDEXED_TABLE_MUTABLE where core < 10 and db < 200
1.5s 2.2.3
1.19s 3.0-RC1

select count(core) from INDEXED_TABLE_MUTABLE where core < 100 and db < 2000
1.34s 2.2.3
0.97s 3.0-RC1

select /*+NO_INDEX*/ count(core) from INDEXED_TABLE_MUTABLE where core < 100 and db < 2000
3.17s 2.2.3
1.55s 3.0-RC1

select count(core) from INDEXED_TABLE_MUTABLE where core > 90
0.23s 2.2.3
0.19s 3.0-RC1

select /*+NO_INDEX*/ count(core) from INDEXED_TABLE_MUTABLE where core > 90
1.34s 2.2.3
1.07s 3.0-RC1

select count(core) from INDEXED_TABLE_MUTABLE where core > 0
0.92s 2.2.3
0.56s 3.0-RC1

select /*+NO_INDEX*/ count(core) from INDEXED_TABLE_MUTABLE where core > 0
2.16s 2.2.3
1.21s 3.0-RC1

select db from INDEXED_TABLE_MUTABLE where db >5 and db <7 order by db
0.18s 2.2.3
0.16s 3.0-RC1

select /*+NO_INDEX*/ db from INDEXED_TABLE_MUTABLE where db >5 and db <7 order by db
1.98s 2.2.3
2.08s 3.0-RC1

select db from INDEXED_TABLE_MUTABLE order by db limit 10
1.03s 2.2.3
1.22s 3.0-RC1

select /*+NO_INDEX*/ db from INDEXED_TABLE_MUTABLE order by db limit 10
2.64s 2.2.3
1.44s 3.0-RC1

CREATE INDEX midx5 ON INDEXED_TABLE_MUTABLE (CORE) INCLUDE (DB,ACTIVE_VISITOR)
36.43s 2.2.3
37.75s 3.0-RC1

LOAD DATA INDEXED_TABLE_MUTABLE [2000000 ROWS]
639.02s 2.2.3
629.08s 3.0-RC1


TABLE_6CF


Data is split in 6 column families. Phoenix 1.2+ uses Essential Column Family filter feature which fetches only required columns when filtering rows.

ROWS 2000000 [2M]

DDL CREATE TABLE IF NOT EXISTS $TABLE (K1 CHAR(1) NOT NULL, K2 VARCHAR NOT NULL, CF1.A INTEGER, CF2.B INTEGER, CF3.C INTEGER, CF4.D INTEGER, CF5.E INTEGER, CF6.F INTEGER CONSTRAINT PK PRIMARY KEY (K1,K2)) SPLIT ON ('B','C','D');CREATE TABLE IF NOT EXISTS $TABLE_UPSERTSELECT (K1 CHAR(1) NOT NULL, K2 VARCHAR NOT NULL, CF1.A INTEGER, CF2.B INTEGER, CF3.C INTEGER, CF4.D INTEGER, CF5.E INTEGER, CF6.F INTEGER CONSTRAINT PK PRIMARY KEY (K1,K2)) SPLIT ON ('B','C','D')

DATA GENERATOR
STRING :: Values: A|B|C|D
STRING :: Length: 10 Values: SEQUENTIAL
INTEGER :: Max. Value: 100 Values: RANDOM
INTEGER :: Max. Value: 1000 Values: RANDOM
INTEGER :: Max. Value: 2000 Values: RANDOM
INTEGER :: Max. Value: 5000 Values: RANDOM
INTEGER :: Max. Value: 10000 Values: RANDOM
INTEGER :: Max. Value: 15000 Values: RANDOM

select count(1) from TABLE_6CF where K1 in ('A', 'B', 'C', 'D')
1.44s 2.2.3
0.74s 3.0-RC1

select count(1) from TABLE_6CF where K1 in ('A', 'C')
1.35s 2.2.3
0.66s 3.0-RC1

select count(*) from TABLE_6CF where f < 1000
1.26s 2.2.3
1.2s 3.0-RC1

select a,b,c,d,e,f from TABLE_6CF where B>1000 and B<2000 and f>1000 and f<2000
1.31s 2.2.3
1.22s 3.0-RC1

select sum(a),sum(b),sum(c),sum(d) from TABLE_6CF group by K1
5.28s 2.2.3
5.01s 3.0-RC1

select count(*) from TABLE_6CF group by A
1.58s 2.2.3
0.83s 3.0-RC1

select count(*) from TABLE_6CF group by B
2.49s 2.2.3
2.54s 3.0-RC1

select count(*) from TABLE_6CF where f < 10000
2.22s 2.2.3
2.08s 3.0-RC1

LOAD DATA TABLE_6CF [2000000 ROWS]
96.88s 2.2.3
104.05s 3.0-RC1


SALTED_TABLE


This is Phoenix 1.2+ feature. Data is split in 4 regions equal to number of region servers in test cluster by adding salted byte to each key.

ROWS 2000000 [2M]

DDL CREATE TABLE $TABLE (K1 CHAR(1) NOT NULL, K2 VARCHAR NOT NULL, CFA.A VARCHAR, CFA.B CHAR(100), CFA.C INTEGER, CFB.D INTEGER CONSTRAINT PK PRIMARY KEY (K1,K2)) SALT_BUCKETS=4

DATA GENERATOR
STRING :: Values: A|B|C|D
STRING :: Length: 10 Values: SEQUENTIAL
STRING :: Length: 100 Values: RANDOM
STRING :: Length: 100 Values: RANDOM
INTEGER :: Max. Value: 1000000 Values: RANDOM
INTEGER :: Max. Value: 1000000 Values: RANDOM

select count(1) from SALTED_TABLE where K1 in ('A', 'B', 'C', 'D')
1.6s 2.2.3
1.05s 3.0-RC1

select count(*) from SALTED_TABLE where D < 1000
0.99s 2.2.3
1.08s 3.0-RC1

select avg(c),avg(d) from SALTED_TABLE group by K1
3.54s 2.2.3
3.73s 3.0-RC1

LOAD DATA SALTED_TABLE [2000000 ROWS]
94.95s 2.2.3
107.1s 3.0-RC1


TABLE_1CF


Same data as in 6 column family table but this one only has a single column family. Performance should be worse in most cases.

ROWS 2000000 [2M]

DDL CREATE TABLE IF NOT EXISTS $TABLE (K1 CHAR(1) NOT NULL, K2 VARCHAR NOT NULL, CF.A INTEGER, CF.B INTEGER, CF.C INTEGER, CF.D INTEGER, CF.E INTEGER, CF.F INTEGER CONSTRAINT PK PRIMARY KEY (K1,K2)) SPLIT ON ('B','C','D');CREATE TABLE IF NOT EXISTS $TABLE_UPSERTSELECT (K1 CHAR(1) NOT NULL, K2 VARCHAR NOT NULL, CF.A INTEGER, CF.B INTEGER, CF.C INTEGER, CF.D INTEGER, CF.E INTEGER, CF.F INTEGER CONSTRAINT PK PRIMARY KEY (K1,K2)) SPLIT ON ('B','C','D')

DATA GENERATOR
STRING :: Values: A|B|C|D
STRING :: Length: 10 Values: SEQUENTIAL
INTEGER :: Max. Value: 100 Values: RANDOM
INTEGER :: Max. Value: 1000 Values: RANDOM
INTEGER :: Max. Value: 2000 Values: RANDOM
INTEGER :: Max. Value: 5000 Values: RANDOM
INTEGER :: Max. Value: 10000 Values: RANDOM
INTEGER :: Max. Value: 15000 Values: RANDOM

select count(1) from TABLE_1CF where K1 in ('A', 'B', 'C', 'D')
1.59s 2.2.3
1.56s 3.0-RC1

select count(1) from TABLE_1CF where K1 in ('A', 'C')
1.56s 2.2.3
1.46s 3.0-RC1

select count(*) from TABLE_1CF where f < 1000
1.83s 2.2.3
1.85s 3.0-RC1

select a,b,c,d,e,f from TABLE_1CF where B>1000 and B<2000 and f>1000 and f<2000
2.21s 2.2.3
2.18s 3.0-RC1

select sum(a),sum(b),sum(c),sum(d) from TABLE_1CF group by K1
3.82s 2.2.3
2.02s 3.0-RC1

select count(*) from TABLE_1CF group by A
1.82s 2.2.3
1.58s 3.0-RC1

select count(*) from TABLE_1CF group by B
2.26s 2.2.3
1.71s 3.0-RC1

select count(*) from TABLE_1CF where f < 10000
2.17s 2.2.3
1.73s 3.0-RC1

LOAD DATA TABLE_1CF [2000000 ROWS]
91.83s 2.2.3
97.33s 3.0-RC1


WIDE_PK




ROWS 2000000 [2M]

DDL CREATE TABLE IF NOT EXISTS $TABLE (mypk CHAR(500) NOT NULL PRIMARY KEY,CF.column1 INTEGER,CF.column2 INTEGER,CF.column3 INTEGER,CF.column4 INTEGER,CF.column5 INTEGER) SALT_BUCKETS=4;

DATA GENERATOR
STRING :: Length: 500 Values: SEQUENTIAL
INTEGER :: Values: SEQUENTIAL
INTEGER :: Max. Value: 100 Values: RANDOM
INTEGER :: Max. Value: 100 Values: RANDOM
INTEGER :: Max. Value: 100 Values: RANDOM
INTEGER :: Max. Value: 100 Values: RANDOM

select count(*) from WIDE_PK
2.15s 2.2.3
1.94s 3.0-RC1

select count(*) from WIDE_PK where column2<10
3.38s 2.2.3
2.23s 3.0-RC1

select count(*) from WIDE_PK where column2<50 and column3<50 and column4<50 and column5<50
4.79s 2.2.3
2.68s 3.0-RC1

LOAD DATA WIDE_PK [2000000 ROWS]
215.7s 2.2.3
227.82s 3.0-RC1



Note: Hover over version label to see Phoenix Query Explain Plan.