CODEWITHSUNDEEP
mysqlmongodbmysql5
XyZ
XyZ

Reputation: 25

Is this MongoDB use case?

I have a dating website in which i send daily alerts and log alerts in ALERTS_LOG.

CREATE TABLE `ALERTS_LOG` (
  `RECEIVERID` mediumint(11) unsigned NOT NULL DEFAULT '0',
  `MATCHID` mediumint(11) unsigned NOT NULL DEFAULT '0',
  `DATE` smallint(6) NOT NULL DEFAULT '0',      
  KEY `RECEIVER` (`RECEIVER`),
  KEY `USER` (`USER`)
) ENGINE=MRG_MyISAM DEFAULT CHARSET=latin1 INSERT_METHOD=LAST UNION=(`ALERTS_LOG110`,`ALERTS_LOG111`,`ALERTS_LOG112`)

Logic Of Insertion : I have create merge table and each sub tables like ALERTS_LOG110 store 0-15 days record. On every 1st and 16th i create a new table and change definition of mergeMyisam. Example : INSERT_METHOD=LAST UNION=(ALERTS_LOG111,ALERTS_LOG112,ALERTS_LOG113).

Advantage : Deletion of is super fast.

Issues with this approach: 1. When i change definition, i often got site down issue as when i change the definition, indexes need to get on cache and all select queries got stuck. 2. Locking issue because of too many inserts and select.

So, can I look MongoDB for solving this issue?

Upvotes: 1

Views: 65

Answers (1)

user672739
user672739

Reputation:

No, not really. Re-engineering your application to use two different database types because of performance on this log table seems like a poor choice.

It's not really clear why you have so many entries being logged, but on the face of it look might like to look into partitioning in MySQL and partition your table by day or week and then drop those partitions. Deletion is still super fast and there would be no downtime for it because you won't be changing object names every day.

Upvotes: 1

Related Questions