Phoenix Performance Result :: Tue Jul 09 14:21:38 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
7.62s PHOENIX-1.1

select count(1) from AGGREGATION_TEST_TABLE group by HOST
8.06s PHOENIX-1.1

select MIN(CORE) from AGGREGATION_TEST_TABLE group by DOMAIN
8.68s PHOENIX-1.1

select core, count(1) from AGGREGATION_TEST_TABLE group by CORE
8.75s PHOENIX-1.1

select core, host, sum(ACTIVE_VISITOR) from AGGREGATION_TEST_TABLE group by core, host
13.5s PHOENIX-1.1

select AVG(DB), MIN(CORE), MAX(ACTIVE_VISITOR) from AGGREGATION_TEST_TABLE where host='CS' and DOMAIN<>'Google.com'
16.01s PHOENIX-1.1

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
5.94s PHOENIX-1.1

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)
11.8s PHOENIX-1.1

select count(active_visitor) count from AGGREGATION_TEST_TABLE
12.44s PHOENIX-1.1

select count(distinct active_visitor) count from AGGREGATION_TEST_TABLE
N/A for PHOENIX-1.1

select percentile_cont(0.9) within group (order by active_visitor asc) from AGGREGATION_TEST_TABLE
N/A for PHOENIX-1.1

select percentile_cont(0.1) within group (order by active_visitor desc) from AGGREGATION_TEST_TABLE
N/A for PHOENIX-1.1

select count(core), count(db), count(active_visitor) from AGGREGATION_TEST_TABLE
17.21s PHOENIX-1.1

select count(distinct core), count(distinct db), count(distinct active_visitor) from AGGREGATION_TEST_TABLE
N/A for PHOENIX-1.1

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
N/A for PHOENIX-1.1

select core, count(db) count from AGGREGATION_TEST_TABLE group by CORE
12.58s PHOENIX-1.1

select core, count(distinct db) count from AGGREGATION_TEST_TABLE group by CORE
N/A for PHOENIX-1.1

select core, percentile_cont(0.9999) within group (order by db asc) from AGGREGATION_TEST_TABLE group by CORE
N/A for PHOENIX-1.1

LOAD DATA [10000000 ROWS]
290.57s 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')
3.71s PHOENIX-1.1

select * from IN_LIKE_TABLE where K1 in ('A','B') and K2 in ('xxx1000000','xxx2000000','xxx2000001')
5.29s PHOENIX-1.1

select * from IN_LIKE_TABLE where K1 in ('A', 'C', 'E', 'F', 'H') and K2 like 'xxxxxx1%'
24.36s 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%'
4.8s 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%')
4.97s PHOENIX-1.1

LOAD DATA [10000000 ROWS]
238.72s 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 count(1) from CF_TABLE_14CF where K1 in ('A', 'B', 'C', 'D')
9.54s PHOENIX-1.1

select count(1) from CF_TABLE_14CF where K1 in ('A', 'C')
9.21s PHOENIX-1.1

select count(*) from CF_TABLE_14CF where N < 1000
13.72s 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)
348.44s PHOENIX-1.1

select sum(a),sum(e),sum(j),sum(n) from CF_TABLE_14CF group by K1
26.82s PHOENIX-1.1

LOAD DATA [10000000 ROWS]
474.67s 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')
20.5s PHOENIX-1.1

select count(*) from SALTED_TABLE where D < 1000
23.76s PHOENIX-1.1

select avg(c),avg(d) from SALTED_TABLE group by K1
33.86s PHOENIX-1.1

LOAD DATA [10000000 ROWS]
441.59s 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 count(1) from CF_TABLE_1CF where K1 in ('A', 'B', 'C', 'D')
9.07s PHOENIX-1.1

select count(1) from CF_TABLE_1CF where K1 in ('A', 'C')
8.92s PHOENIX-1.1

select count(*) from CF_TABLE_1CF where N < 1000
11.56s 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
63s PHOENIX-1.1

select sum(a),sum(e),sum(j),sum(n) from CF_TABLE_1CF group by K1
25.37s PHOENIX-1.1

LOAD DATA [10000000 ROWS]
395.33s PHOENIX-1.1