Phoenix Performance Result :: Tue Jun 04 15:18:54 PDT 2013

AGGREGATION_TABLE
Various server side grouped/un-grouped aggregation functions. Data is pre-split in 9 regions over a 4 region server cluster.

ROWS 10000000 [10M]

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

QUERIES
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)
7.95s SNAPSHOT
7.93s PHOENIX-1.2.0
13.26s PHOENIX-1.1

select round(date, 'day',50), sum(core)+2 from AGGREGATION_TABLE where (CORE<10 or CORE>90) group by round(date, 'day',50), host
5.62s SNAPSHOT
5.78s PHOENIX-1.2.0
6.67s PHOENIX-1.1

select count(1) from AGGREGATION_TABLE group by HOST
7.2s SNAPSHOT
7.31s PHOENIX-1.2.0
8.89s PHOENIX-1.1

select count(*) from AGGREGATION_TABLE
6.89s SNAPSHOT
7.04s PHOENIX-1.2.0
8.6s PHOENIX-1.1

select core, host, sum(ACTIVE_VISITOR) from AGGREGATION_TABLE group by core, host
13.22s SNAPSHOT
13.4s PHOENIX-1.2.0
15.46s PHOENIX-1.1

select core, count(1) from AGGREGATION_TABLE group by CORE
7.9s SNAPSHOT
8.16s PHOENIX-1.2.0
9.53s PHOENIX-1.1

select MIN(CORE) from AGGREGATION_TABLE group by DOMAIN
8.1s SNAPSHOT
8.2s PHOENIX-1.2.0
9.67s PHOENIX-1.1

select AVG(DB), MIN(CORE), MAX(ACTIVE_VISITOR) from AGGREGATION_TABLE where host='CS' and DOMAIN<>'Google.com'
17.09s SNAPSHOT
17.44s PHOENIX-1.2.0
17.07s PHOENIX-1.1

LOAD DATA [10000000 ROWS]
359.53s SNAPSHOT
360.17s PHOENIX-1.2.0
306.13s PHOENIX-1.1


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

ROWS 10000000 [10M]

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

QUERIES
select count(1) from IN_LIKE_TABLE where K1 in ('Z', 'D', 'F', 'X')
3.69s SNAPSHOT
3.59s PHOENIX-1.2.0
3.88s PHOENIX-1.1

select count(*) from IN_LIKE_TABLE where K1 in ('X','A','B','C','D','E','F','G','H','I','Z') and K2 like '%1%'
5.21s SNAPSHOT
5.24s PHOENIX-1.2.0
4.88s PHOENIX-1.1

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%')
5.45s SNAPSHOT
5.18s PHOENIX-1.2.0
5.02s PHOENIX-1.1

select * from IN_LIKE_TABLE where K1 in ('A','B') and K2 in ('xxx1000000','xxx2000000','xxx2000001')
0.01s SNAPSHOT
0.01s PHOENIX-1.2.0
5.69s PHOENIX-1.1

select * from IN_LIKE_TABLE where K1 in ('A', 'C', 'E', 'F', 'H') and K2 like 'xxxxxx1%'
0.02s SNAPSHOT
0.02s PHOENIX-1.2.0
24.57s PHOENIX-1.1

LOAD DATA [10000000 ROWS]
308.87s SNAPSHOT
312.78s PHOENIX-1.2.0
242.25s PHOENIX-1.1


CF_TABLE_14CF
Data is split in 14 column families. Phoenix 1.2+ uses Essential Column Family filter feature which fetches only required columns when filtering rows.

ROWS 10000000 [10M]

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, CF7.G INTEGER, CF8.H INTEGER, CF9.I INTEGER, CF10.J INTEGER, CF11.K INTEGER, CF12.L INTEGER, CF13.M INTEGER, CF14.N 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: 1000000 Values: RANDOM
INTEGER :: Max. Value: 1000000 Values: RANDOM
INTEGER :: Max. Value: 1000000 Values: RANDOM
INTEGER :: Max. Value: 1000000 Values: RANDOM
INTEGER :: Max. Value: 1000000 Values: RANDOM
INTEGER :: Max. Value: 1000000 Values: RANDOM
INTEGER :: Max. Value: 1000000 Values: RANDOM
INTEGER :: Max. Value: 1000000 Values: RANDOM
INTEGER :: Max. Value: 1000000 Values: RANDOM
INTEGER :: Max. Value: 1000000 Values: RANDOM
INTEGER :: Max. Value: 1000000 Values: RANDOM
INTEGER :: Max. Value: 1000000 Values: RANDOM
INTEGER :: Max. Value: 1000000 Values: RANDOM
INTEGER :: Max. Value: 1000000 Values: RANDOM

QUERIES
select sum(a),sum(e),sum(j),sum(n) from CF_TABLE_14CF group by K1
34.69s SNAPSHOT
35.86s PHOENIX-1.2.0
31.29s PHOENIX-1.1

select count(1) from CF_TABLE_14CF where K1 in ('A', 'B', 'C', 'D')
11.72s SNAPSHOT
11.92s PHOENIX-1.2.0
10.19s PHOENIX-1.1

select count(*) from CF_TABLE_14CF where N < 1000
5.95s SNAPSHOT
5.76s PHOENIX-1.2.0
14.16s PHOENIX-1.1

select a,b,c,d,e,f,g,h,i,j,k,l,m,n from CF_TABLE_14CF where (B>1000 and B<2000) and (M>1000 and M<2000)
7.84s SNAPSHOT
8.04s PHOENIX-1.2.0
418.67s PHOENIX-1.1

LOAD DATA [10000000 ROWS]
620.85s SNAPSHOT
632.56s PHOENIX-1.2.0
474.07s PHOENIX-1.1


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 10000000 [10M]

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

QUERIES
select count(1) from SALTED_TABLE where K1 in ('A', 'B', 'C', 'D')
6.77s SNAPSHOT
7.48s PHOENIX-1.2.0
18.41s PHOENIX-1.1

select count(*) from SALTED_TABLE where D < 1000
2.74s SNAPSHOT
2.69s PHOENIX-1.2.0
26.82s PHOENIX-1.1

select avg(c),avg(d) from SALTED_TABLE group by K1
14.46s SNAPSHOT
14.95s PHOENIX-1.2.0
27.72s PHOENIX-1.1

LOAD DATA [10000000 ROWS]
378.42s SNAPSHOT
388.32s PHOENIX-1.2.0
419.04s PHOENIX-1.1


CF_TABLE_1CF
Same data as in 14 column family table but this one only has a single column family. Performance should be worse in most cases.

ROWS 10000000 [10M]

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, CF.G INTEGER, CF.H INTEGER, CF.I INTEGER, CF.J INTEGER, CF.K INTEGER, CF.L INTEGER, CF.M INTEGER, CF.N 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: 1000000 Values: RANDOM
INTEGER :: Max. Value: 1000000 Values: RANDOM
INTEGER :: Max. Value: 1000000 Values: RANDOM
INTEGER :: Max. Value: 1000000 Values: RANDOM
INTEGER :: Max. Value: 1000000 Values: RANDOM
INTEGER :: Max. Value: 1000000 Values: RANDOM
INTEGER :: Max. Value: 1000000 Values: RANDOM
INTEGER :: Max. Value: 1000000 Values: RANDOM
INTEGER :: Max. Value: 1000000 Values: RANDOM
INTEGER :: Max. Value: 1000000 Values: RANDOM
INTEGER :: Max. Value: 1000000 Values: RANDOM
INTEGER :: Max. Value: 1000000 Values: RANDOM
INTEGER :: Max. Value: 1000000 Values: RANDOM
INTEGER :: Max. Value: 1000000 Values: RANDOM

QUERIES
select sum(a),sum(e),sum(j),sum(n) from CF_TABLE_1CF group by K1
31.77s SNAPSHOT
31.41s PHOENIX-1.2.0
32.72s PHOENIX-1.1

select count(1) from CF_TABLE_1CF where K1 in ('A', 'B', 'C', 'D')
9.98s SNAPSHOT
10.26s PHOENIX-1.2.0
10.38s PHOENIX-1.1

select count(*) from CF_TABLE_1CF where N < 1000
14.27s SNAPSHOT
14.08s PHOENIX-1.2.0
14.06s PHOENIX-1.1

select a,b,c,d,e,f,g,h,i,j,k,l,m,n from CF_TABLE_1CF where B>1000 and B<2000 and M>1000 and M<2000
16.69s SNAPSHOT
16.54s PHOENIX-1.2.0
72.49s PHOENIX-1.1

LOAD DATA [10000000 ROWS]
546.66s SNAPSHOT
544.38s PHOENIX-1.2.0
401.25s PHOENIX-1.1