Why is querying millions of inserted data slower than MySQL? about greptimedb HOT 9 CLOSED

@tx991020 can you provide more detailed infomation about your benchmark testing setup?

from greptimedb.

translates to The format is as follows
"1","1208026841","477458502","FKlJh","2023-08-05 08:56:00","pSvqg","gQWxL","JikNj","972076018","348948709","1","GMZHG","XzNXZ","QHfSD","0","ejliy"

6 million records
Comparison between standalone MySQL and standalone GrepTime: In terms of a count query with a non-indexed field, GrepTime is several seconds slower than MySQL in execution.

from greptimedb.

MySQL takes 6 seconds, while GrepTime seems to take 11 seconds

from greptimedb.

Thx! Can you share the create table sql as well? If there are senstive data, it's ok to change the columns names. We'd like to do the same benchmark to investigate more

from greptimedb.

CREATE TABLE IF NOT EXISTS xxxx (
a BIGINT NULL,
b BIGINT NULL,
c TIMESTAMP(3) NOT NULL DEFAULT current_timestamp(),
d TIMESTAMP(3) NULL DEFAULT current_timestamp(),
e STRING NULL DEFAULT '',
f INT NULL,
g INT NULL,
h INT NULL,
i STRING NULL,
j STRING NULL,
k STRING NULL,
l STRING NULL,
m STRING NULL,
n STRING NULL,
o INT NULL,
p INT NULL,
q INT NULL,
r STRING NULL,
s STRING NULL,
t STRING NULL,
u STRING NULL,
v STRING NULL,
w STRING NULL,
x TINYINT NULL DEFAULT 0,
y INT NULL,
z STRING NULL,
aa STRING NULL,
ab TINYINT NULL DEFAULT 0,
ac TINYINT NULL DEFAULT 0,
TIME INDEX (c),
PRIMARY KEY (a)
)
ENGINE=mito
WITH(
regions = 1
);

from greptimedb.

@tx991020 can you share the queries for benchmarking, too?

from greptimedb.

slect count(*) from xxx; slelect e from xxx where xx

from greptimedb.

thx, we'll investigate the query performance asap

from greptimedb.

@tx991020 we are currently have some issues related to scan, dragging down our overall query performance. It's planned to be optimized soon, stay tuned!

from greptimedb.