Open Source RDBMS - Seamless, Scalable, Stable and Free

English | Login |Register


1
(click on this box to dismiss)

CUBRID & highload?

Hi.

I have a banner network project. I need to INSERT up to 2k rows per second (thats a peak load) in one table and UPDATE up to 300 rows in another.

Table for INSERTs looks like this: id primary key, ip integer, date (date when record was added), user agent, referer, publisher_id (references row in another table), feed_id (references row in another table).

Table for UPDATEs has 12 fields and intentified by primary key id, only one of them (integer counter) is updated.

Also I need to do 2 SELECTs: one on INSERTs table and one on another table with ~10k records.

INSERTs table can have up to 10 million records (this is a log for views/clicks).

Currently I'm using MySQL with Infobright extension, but I already see it doesn't handle load. Scaling MySQL is a problem, so I'm looking for some alternative able to handle such load.

The question is: is CUBRID ready for this? Is it stable?

링크 댓글 쓰기 (0)
질문시간 2년 전
naquad
6
45 답변들
0

I don't make sure what I'm saying is sutalble to your case.  But according to CUBRID testing result,   http://www.cubrid.org/blog/news/new-cubrid-8-4-1--is-70-faster/ , CUBRID can meet your need in case of insert performance. In case of select, If you specify more of your environment, we can estimate it

링크 댓글 쓰기 (0)
답변시간 2년 전
choi
36
0

Dear naquad,
Sorry for late asnwering.
Before I answer to your question, I'd like to ask to read an article about CUBRID Shard, which must be inlcuded in the very next version. We have completed development and binary is ready however documentating manuals takes so long. When all set, we will release out. Here is an article which might help you: http://www.cubrid.org/blog/cubrid-life/database-sharding-with-cubrid/  and http://www.cubrid.org/blog/dev-platform/database-sharding-platform-at-nhn/

Back to your service, I guess it has heavy WRITE operations into mainly 2 tables and DB scalability matters. I think DB shard solution can help your service by dividing your single DB into 2 shards having the same table schema, middleware(we call Proxy) knows where to execute SQLs based on your PK and location map(shard rule). However if your SELECT needs aggregation of 2 shards at once, it can matters. In other hands, if your application allows SELECT data from shard1 and from shard2 separately, then DB shrad si definitely suitable.  As far as I know, MySQL also has a Sharding solution, called "Spider" and I think you should look into it. If this is not urgent that much, please wait for our new version and test CUBRID Shard for your service.  
Thank you!  

 

링크 댓글 쓰기 (0)
답변시간 2년 전
cottonspan
61
tagged




You are either using a very old browser or a browser that is not supported.
In order to browse cubrid.org you need to have one of the following browsers:



Internet Explorer: Mozilla Firefox: Google Chrome: