Hello, my name is Vladimir and I have questions on Cassandra DMBS.
*1 PROBLEM. How to increase the speed?*
I installed this system from the repository:
"deb http://www.apache.org/dist/cassandra/debian 311x main"
With standard settings, the system runs slowly.
_Comparison with "MySQL"._
select count (*) from login_wifly.radacct;
+ ---------- +
| | count (*) |
+ ---------- +
| | 9810806 |
+ ---------- +
real 0m3.709s --------- speed in the MySQL (without caching)
user 0m0.000s
sys 0m0.000s
_In Cassandra:_
SELECT count(*) FROM test.radacct;
count
---------
9810806
(1 rows)
Warnings :
Aggregation query used without partition key
real 3m7.661s -------- speed in the Cassandra
user 0m0.444s
sys 0m0.056s
*2 PROBLEM. How to design a table correctly?*
My test table configure:
create table radacct2 (
radacctid bigint,
acctsessionid text,
acctuniqueid text,
username text,
groupname text,
realm text,
nasid text,
nasipaddress text,
nasportid text,
nasporttype text,
acctstarttime text,
acctstoptime text,
acctsessiontime bigint,
acctauthentic text,
connectinfo_start text,
connectinfo_stop text,
acctinputoctets bigint,
"acctoutputoctets" bigint,
"calledstationid" text,
callingstationid text,
acctterminatecause text,
servicetype text,
framedprotocol text,
framedipaddress text,
acctstartdelay bigint,
acctstopdelay bigint,
xascendsessionsvrkey text,
client bigint,
method text,
zone bigint,
localDateStart text,
localDateStop text,
localDateTimeStart text,
localDateTimeStop text,
msisdn text,
PRIMARY KEY (radacctid, username)
) WITH CLUSTERING ORDER BY (username DESC);
I need do select to the "username" field.
In MySQL it looks like this:
SELECT
a.username,
COUNT (DISTINCT a.username)
FROM
radacct as a
WHERE
(LENGTH (a.username) = 17)
GROUP BY
a.username;
When I execute the query "SELECT username, count (*) FROM radacct GROUP
BY username;
InvalidRequest: Error from server: code = 2200 [Invalid query] message =
"PRIMARY KEY, got username"
*3 PROBLEM. How to improve performance using a competent configuration?*
cat /etc/cassandra/cassandra.yaml
What parameters can be adjusted to achieve maximum effect?
