Phoenix Performance Result :: Thu May 30 16:10:17 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)
7.89s SNAPSHOT
7.78s PHOENIX-1.2.0
12.93s 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
5.78s SNAPSHOT
5.78s PHOENIX-1.2.0
6.66s PHOENIX-1.1

select count(1) from AGGREGATION_TABLE group by HOST
7.64s SNAPSHOT
7.43s PHOENIX-1.2.0
9.01s PHOENIX-1.1

select count(*) from AGGREGATION_TABLE
7.32s SNAPSHOT
7.18s PHOENIX-1.2.0
8.62s PHOENIX-1.1

select core, host, sum(ACTIVE_VISITOR) from AGGREGATION_TABLE group by core, host
14.18s SNAPSHOT
13.04s PHOENIX-1.2.0
15.22s PHOENIX-1.1

select core, count(1) from AGGREGATION_TABLE group by CORE
8.55s SNAPSHOT
7.99s PHOENIX-1.2.0
9.44s PHOENIX-1.1

select MIN(CORE) from AGGREGATION_TABLE group by DOMAIN
8.79s SNAPSHOT
8.24s 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'
16.41s SNAPSHOT
17.75s PHOENIX-1.2.0
17.2s PHOENIX-1.1

LOAD DATA [10000000 ROWS]
357.65s SNAPSHOT
359.74s PHOENIX-1.2.0
293.84s 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')
3.72s SNAPSHOT
4.09s PHOENIX-1.2.0
3.92s 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%'
5.33s SNAPSHOT
5.43s PHOENIX-1.2.0
4.91s 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%')
5.14s SNAPSHOT
5.2s PHOENIX-1.2.0
5.19s 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
5.66s 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
24.76s PHOENIX-1.1

LOAD DATA [10000000 ROWS]
319.2s SNAPSHOT
320.3s PHOENIX-1.2.0
254.54s 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
38.61s SNAPSHOT
35.58s PHOENIX-1.2.0
35.52s PHOENIX-1.1

select count(1) from CF_TABLE_14CF where K1 in ('A', 'B', 'C', 'D')
12.7s SNAPSHOT
12s PHOENIX-1.2.0
12.14s PHOENIX-1.1

select count(*) from CF_TABLE_14CF where N < 1000
6.28s SNAPSHOT
6.02s PHOENIX-1.2.0
17.8s 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)
8.13s SNAPSHOT
7.74s PHOENIX-1.2.0
436.5s PHOENIX-1.1

LOAD DATA [10000000 ROWS]
635.9s SNAPSHOT
634.31s PHOENIX-1.2.0
476.53s 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')
7.9s SNAPSHOT
7.25s PHOENIX-1.2.0
17.49s PHOENIX-1.1

select count(*) from SALTED_TABLE where D < 1000
2.93s SNAPSHOT
2.74s PHOENIX-1.2.0
22.3s PHOENIX-1.1

select avg(c),avg(d) from SALTED_TABLE group by K1
16.53s SNAPSHOT
14.92s PHOENIX-1.2.0
28.3s PHOENIX-1.1

LOAD DATA [10000000 ROWS]
389.52s SNAPSHOT
387.31s PHOENIX-1.2.0
431.38s 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
35.44s SNAPSHOT
32.53s PHOENIX-1.2.0
33.79s PHOENIX-1.1

select count(1) from CF_TABLE_1CF where K1 in ('A', 'B', 'C', 'D')
10.6s SNAPSHOT
10.41s PHOENIX-1.2.0
10.6s PHOENIX-1.1

select count(*) from CF_TABLE_1CF where N < 1000
15.18s SNAPSHOT
14.84s PHOENIX-1.2.0
14.54s 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
17.76s SNAPSHOT
16.89s PHOENIX-1.2.0
78.66s PHOENIX-1.1

LOAD DATA [10000000 ROWS]
551.79s SNAPSHOT
555.07s PHOENIX-1.2.0
417.21s PHOENIX-1.1