Phoenix Performance Result :: Mon Jul 08 20:16:12 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 7.3s CURRENT-SNAPSHOT 6.89s LAST-SNAPSHOT 6.9s PHOENIX-1.2.0

select count(1) from AGGREGATION_TEST_TABLE group by HOST 7.49s CURRENT-SNAPSHOT 7.15s LAST-SNAPSHOT 7.17s PHOENIX-1.2.0

select MIN(CORE) from AGGREGATION_TEST_TABLE group by DOMAIN 8.13s CURRENT-SNAPSHOT 7.82s LAST-SNAPSHOT 8.24s PHOENIX-1.2.0

select core, count(1) from AGGREGATION_TEST_TABLE group by CORE 7.97s CURRENT-SNAPSHOT 7.59s LAST-SNAPSHOT 7.91s PHOENIX-1.2.0

select core, host, sum(ACTIVE_VISITOR) from AGGREGATION_TEST_TABLE group by core, host 12.91s CURRENT-SNAPSHOT 12.39s LAST-SNAPSHOT 12.8s PHOENIX-1.2.0

select AVG(DB), MIN(CORE), MAX(ACTIVE_VISITOR) from AGGREGATION_TEST_TABLE where host='CS' and DOMAIN<>'Google.com' 17.29s CURRENT-SNAPSHOT 16.81s LAST-SNAPSHOT 17.34s 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 5.64s CURRENT-SNAPSHOT 5.41s LAST-SNAPSHOT 5.67s 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) 7.66s CURRENT-SNAPSHOT 7.57s LAST-SNAPSHOT 7.65s PHOENIX-1.2.0

select count(active_visitor) count from AGGREGATION_TEST_TABLE 11.44s CURRENT-SNAPSHOT 11.41s LAST-SNAPSHOT 11.62s PHOENIX-1.2.0

select count(distinct active_visitor) count from AGGREGATION_TEST_TABLE 12.93s CURRENT-SNAPSHOT 12.54s LAST-SNAPSHOT N/A for PHOENIX-1.2.0

select percentile_cont(0.9) within group (order by active_visitor asc) from AGGREGATION_TEST_TABLE 12.84s CURRENT-SNAPSHOT 12.63s LAST-SNAPSHOT N/A for PHOENIX-1.2.0

select percentile_cont(0.1) within group (order by active_visitor desc) from AGGREGATION_TEST_TABLE 12.84s CURRENT-SNAPSHOT 12.61s LAST-SNAPSHOT N/A for PHOENIX-1.2.0

select count(core), count(db), count(active_visitor) from AGGREGATION_TEST_TABLE 15.81s CURRENT-SNAPSHOT 15.76s LAST-SNAPSHOT 16s PHOENIX-1.2.0

select count(distinct core), count(distinct db), count(distinct active_visitor) from AGGREGATION_TEST_TABLE 18.25s CURRENT-SNAPSHOT 17.8s LAST-SNAPSHOT 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 18.2s CURRENT-SNAPSHOT 17.8s LAST-SNAPSHOT N/A for PHOENIX-1.2.0

select core, count(db) count from AGGREGATION_TEST_TABLE group by CORE 11.62s CURRENT-SNAPSHOT 11.67s LAST-SNAPSHOT 12.01s PHOENIX-1.2.0

select core, count(distinct db) count from AGGREGATION_TEST_TABLE group by CORE 14.87s CURRENT-SNAPSHOT 14.82s LAST-SNAPSHOT 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 14.96s CURRENT-SNAPSHOT 14.89s LAST-SNAPSHOT N/A for PHOENIX-1.2.0

LOAD DATA [10000000 ROWS] 325.18s CURRENT-SNAPSHOT 315.93s LAST-SNAPSHOT 352.4s 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') 3.75s CURRENT-SNAPSHOT 3.68s LAST-SNAPSHOT 3.6s 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 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 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%' 5.4s CURRENT-SNAPSHOT 5.58s LAST-SNAPSHOT 5.6s 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%') 5.18s CURRENT-SNAPSHOT 4.94s LAST-SNAPSHOT 5.24s PHOENIX-1.2.0

LOAD DATA [10000000 ROWS] 268.56s CURRENT-SNAPSHOT 270.52s LAST-SNAPSHOT 323.14s 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') 8.09s CURRENT-SNAPSHOT 8.47s LAST-SNAPSHOT 8.38s PHOENIX-1.2.0

select count(1) from CF_TABLE_14CF where K1 in ('A', 'C') 7.8s CURRENT-SNAPSHOT 8.28s LAST-SNAPSHOT 8.19s PHOENIX-1.2.0

select count(*) from CF_TABLE_14CF where N < 1000 4.07s CURRENT-SNAPSHOT 4.28s LAST-SNAPSHOT 4.3s 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) 5.83s CURRENT-SNAPSHOT 6.13s LAST-SNAPSHOT 5.93s PHOENIX-1.2.0

select sum(a),sum(e),sum(j),sum(n) from CF_TABLE_14CF group by K1 25.83s CURRENT-SNAPSHOT 25.73s LAST-SNAPSHOT 25.71s PHOENIX-1.2.0

LOAD DATA [10000000 ROWS] 529.58s CURRENT-SNAPSHOT 547.01s LAST-SNAPSHOT 649.25s 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') 5.94s CURRENT-SNAPSHOT 6.01s LAST-SNAPSHOT 6.04s PHOENIX-1.2.0

select count(*) from SALTED_TABLE where D < 1000 2.19s CURRENT-SNAPSHOT 2.27s LAST-SNAPSHOT 2.25s PHOENIX-1.2.0

select avg(c),avg(d) from SALTED_TABLE group by K1 12.09s CURRENT-SNAPSHOT 12.21s LAST-SNAPSHOT 12.02s PHOENIX-1.2.0

LOAD DATA [10000000 ROWS] 336.35s CURRENT-SNAPSHOT 334.35s LAST-SNAPSHOT 388.35s 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') 8.45s CURRENT-SNAPSHOT 8.92s LAST-SNAPSHOT 8.74s PHOENIX-1.2.0

select count(1) from CF_TABLE_1CF where K1 in ('A', 'C') 8.5s CURRENT-SNAPSHOT 8.1s LAST-SNAPSHOT 8.73s PHOENIX-1.2.0

select count(*) from CF_TABLE_1CF where N < 1000 11.67s CURRENT-SNAPSHOT 11.93s LAST-SNAPSHOT 12.52s 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 13.65s CURRENT-SNAPSHOT 13.98s LAST-SNAPSHOT 14.81s PHOENIX-1.2.0

select sum(a),sum(e),sum(j),sum(n) from CF_TABLE_1CF group by K1 24.82s CURRENT-SNAPSHOT 25.6s LAST-SNAPSHOT 27.3s PHOENIX-1.2.0

LOAD DATA [10000000 ROWS] 457.96s CURRENT-SNAPSHOT 472.3s LAST-SNAPSHOT 562.15s PHOENIX-1.2.0