Phoenix Performance Result :: Mon Jul 15 18:01:44 PDT 2013

AGGREGATION_TEST_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 count(*) from AGGREGATION_TEST_TABLE
7.07s CURRENT-SNAPSHOT
6.94s LAST-SNAPSHOT
7.37s PHOENIX-1.1
7.29s PHOENIX-1.2.0

select count(1) from AGGREGATION_TEST_TABLE group by HOST
7.3s CURRENT-SNAPSHOT
7.08s LAST-SNAPSHOT
7.54s PHOENIX-1.1
7.56s PHOENIX-1.2.0

select MIN(CORE) from AGGREGATION_TEST_TABLE group by DOMAIN
8.22s CURRENT-SNAPSHOT
8.06s LAST-SNAPSHOT
8.19s PHOENIX-1.1
8.25s PHOENIX-1.2.0

select core, count(1) from AGGREGATION_TEST_TABLE group by CORE
8.06s CURRENT-SNAPSHOT
7.58s LAST-SNAPSHOT
8.08s PHOENIX-1.1
8.02s PHOENIX-1.2.0

select core, host, sum(ACTIVE_VISITOR) from AGGREGATION_TEST_TABLE group by core, host
13.3s CURRENT-SNAPSHOT
12.84s LAST-SNAPSHOT
13s PHOENIX-1.1
13.12s PHOENIX-1.2.0

select AVG(DB), MIN(CORE), MAX(ACTIVE_VISITOR) from AGGREGATION_TEST_TABLE where host='CS' and DOMAIN<>'Google.com'
18.38s CURRENT-SNAPSHOT
17.35s LAST-SNAPSHOT
16.84s PHOENIX-1.1
18.26s PHOENIX-1.2.0

select round(date, 'day',50), sum(core)+2 from AGGREGATION_TEST_TABLE where (CORE<10 or CORE>90) group by round(date, 'day',50), host
5.7s CURRENT-SNAPSHOT
5.53s LAST-SNAPSHOT
5.51s PHOENIX-1.1
5.85s PHOENIX-1.2.0

select trunc(date, 'hour', 1000), count(DB), count(ACTIVE_VISITOR) from AGGREGATION_TEST_TABLE where feature='Login' and (core<50 or db<20 or active_visitor > 200) group by trunc(date, 'hour', 1000)
7.88s CURRENT-SNAPSHOT
7.55s LAST-SNAPSHOT
10.9s PHOENIX-1.1
7.98s PHOENIX-1.2.0

select count(active_visitor) count from AGGREGATION_TEST_TABLE
11.8s CURRENT-SNAPSHOT
11.77s LAST-SNAPSHOT
11.55s PHOENIX-1.1
11.71s PHOENIX-1.2.0

select count(distinct active_visitor) count from AGGREGATION_TEST_TABLE
13.2s CURRENT-SNAPSHOT
13.01s LAST-SNAPSHOT
N/A for PHOENIX-1.1
N/A for PHOENIX-1.2.0

select percentile_cont(0.9) within group (order by active_visitor asc) from AGGREGATION_TEST_TABLE
13.12s CURRENT-SNAPSHOT
12.94s LAST-SNAPSHOT
N/A for PHOENIX-1.1
N/A for PHOENIX-1.2.0

select percentile_cont(0.1) within group (order by active_visitor desc) from AGGREGATION_TEST_TABLE
13s CURRENT-SNAPSHOT
12.91s LAST-SNAPSHOT
N/A for PHOENIX-1.1
N/A for PHOENIX-1.2.0

select count(core), count(db), count(active_visitor) from AGGREGATION_TEST_TABLE
16.37s CURRENT-SNAPSHOT
16.05s LAST-SNAPSHOT
16.05s PHOENIX-1.1
15.91s PHOENIX-1.2.0

select count(distinct core), count(distinct db), count(distinct active_visitor) from AGGREGATION_TEST_TABLE
18.7s CURRENT-SNAPSHOT
18.37s LAST-SNAPSHOT
N/A for PHOENIX-1.1
N/A for PHOENIX-1.2.0

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_TEST_TABLE
18.58s CURRENT-SNAPSHOT
18.36s LAST-SNAPSHOT
N/A for PHOENIX-1.1
N/A for PHOENIX-1.2.0

select core, count(db) count from AGGREGATION_TEST_TABLE group by CORE
12.15s CURRENT-SNAPSHOT
11.79s LAST-SNAPSHOT
11.74s PHOENIX-1.1
11.94s PHOENIX-1.2.0

select core, count(distinct db) count from AGGREGATION_TEST_TABLE group by CORE
14.93s CURRENT-SNAPSHOT
14.46s LAST-SNAPSHOT
N/A for PHOENIX-1.1
N/A for PHOENIX-1.2.0

select core, percentile_cont(0.9999) within group (order by db asc) from AGGREGATION_TEST_TABLE group by CORE
15.04s CURRENT-SNAPSHOT
14.5s LAST-SNAPSHOT
N/A for PHOENIX-1.1
N/A for PHOENIX-1.2.0

LOAD DATA [10000000 ROWS]
317.75s CURRENT-SNAPSHOT
323.38s LAST-SNAPSHOT
289.68s PHOENIX-1.1
362.54s PHOENIX-1.2.0


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')
3.34s CURRENT-SNAPSHOT
3.4s LAST-SNAPSHOT
3.8s PHOENIX-1.1
3.94s PHOENIX-1.2.0

select * from IN_LIKE_TABLE where K1 in ('A','B') and K2 in ('xxx1000000','xxx2000000','xxx2000001')
0.01s CURRENT-SNAPSHOT
0.01s LAST-SNAPSHOT
5.28s PHOENIX-1.1
0.01s PHOENIX-1.2.0

select * from IN_LIKE_TABLE where K1 in ('A', 'C', 'E', 'F', 'H') and K2 like 'xxxxxx1%'
0.01s CURRENT-SNAPSHOT
0.01s LAST-SNAPSHOT
24.3s PHOENIX-1.1
0.02s PHOENIX-1.2.0

select count(*) from IN_LIKE_TABLE where K1 in ('X','A','B','C','D','E','F','G','H','I','Z') and K2 like '%1%'
5.69s CURRENT-SNAPSHOT
5.44s LAST-SNAPSHOT
4.72s PHOENIX-1.1
5.29s PHOENIX-1.2.0

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%')
5.42s CURRENT-SNAPSHOT
5.1s LAST-SNAPSHOT
4.79s PHOENIX-1.1
5.09s PHOENIX-1.2.0

LOAD DATA [10000000 ROWS]
249.57s CURRENT-SNAPSHOT
262.44s LAST-SNAPSHOT
235.99s PHOENIX-1.1
319.46s PHOENIX-1.2.0


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 count(1) from CF_TABLE_14CF where K1 in ('A', 'B', 'C', 'D')
8.81s CURRENT-SNAPSHOT
8.37s LAST-SNAPSHOT
9.86s PHOENIX-1.1
8.41s PHOENIX-1.2.0

select count(1) from CF_TABLE_14CF where K1 in ('A', 'C')
8.23s CURRENT-SNAPSHOT
8.39s LAST-SNAPSHOT
9.62s PHOENIX-1.1
8.42s PHOENIX-1.2.0

select count(*) from CF_TABLE_14CF where N < 1000
4.16s CURRENT-SNAPSHOT
4.34s LAST-SNAPSHOT
14.42s PHOENIX-1.1
3.99s PHOENIX-1.2.0

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)
5.98s CURRENT-SNAPSHOT
5.9s LAST-SNAPSHOT
367.59s PHOENIX-1.1
5.89s PHOENIX-1.2.0

select sum(a),sum(e),sum(j),sum(n) from CF_TABLE_14CF group by K1
25.93s CURRENT-SNAPSHOT
25.37s LAST-SNAPSHOT
28.74s PHOENIX-1.1
26.27s PHOENIX-1.2.0

LOAD DATA [10000000 ROWS]
504.05s CURRENT-SNAPSHOT
532.79s LAST-SNAPSHOT
475.08s PHOENIX-1.1
633.99s PHOENIX-1.2.0


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')
6.05s CURRENT-SNAPSHOT
6.51s LAST-SNAPSHOT
12.6s PHOENIX-1.1
6.12s PHOENIX-1.2.0

select count(*) from SALTED_TABLE where D < 1000
2.22s CURRENT-SNAPSHOT
2.28s LAST-SNAPSHOT
17.04s PHOENIX-1.1
2.06s PHOENIX-1.2.0

select avg(c),avg(d) from SALTED_TABLE group by K1
12.05s CURRENT-SNAPSHOT
12.02s LAST-SNAPSHOT
24.86s PHOENIX-1.1
12.45s PHOENIX-1.2.0

LOAD DATA [10000000 ROWS]
330.8s CURRENT-SNAPSHOT
335.37s LAST-SNAPSHOT
386.35s PHOENIX-1.1
388.32s PHOENIX-1.2.0


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 count(1) from CF_TABLE_1CF where K1 in ('A', 'B', 'C', 'D')
8.62s CURRENT-SNAPSHOT
8.29s LAST-SNAPSHOT
8.65s PHOENIX-1.1
8.88s PHOENIX-1.2.0

select count(1) from CF_TABLE_1CF where K1 in ('A', 'C')
8.58s CURRENT-SNAPSHOT
8.2s LAST-SNAPSHOT
8.41s PHOENIX-1.1
8.34s PHOENIX-1.2.0

select count(*) from CF_TABLE_1CF where N < 1000
11.38s CURRENT-SNAPSHOT
11.49s LAST-SNAPSHOT
11.1s PHOENIX-1.1
11.56s PHOENIX-1.2.0

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
13.53s CURRENT-SNAPSHOT
13.77s LAST-SNAPSHOT
62.28s PHOENIX-1.1
13.69s PHOENIX-1.2.0

select sum(a),sum(e),sum(j),sum(n) from CF_TABLE_1CF group by K1
24.87s CURRENT-SNAPSHOT
24.96s LAST-SNAPSHOT
24.24s PHOENIX-1.1
24.94s PHOENIX-1.2.0

LOAD DATA [10000000 ROWS]
434.81s CURRENT-SNAPSHOT
463.71s LAST-SNAPSHOT
394.26s PHOENIX-1.1
547.5s PHOENIX-1.2.0