mySQL's Robustness

I am building a large, complex, database driven site using mySQL.

I want to know if anyone has experience in how well does mySQL perform if let's say, the site getting 100 000 hits a day.

The site is generated dynamically and there are registered users that need to be tracked, data warehousing functions, etc.

Is mySQL robust enough to handle this traffic (it will be run off a Solaris machine)?

We plan on migrating to an Oracle or Informix database in about 8 months, but for now, we need to work with mySQL because it's very cost effective.

Any comments/responses would be much appreciated.




[This message has been edited by invisible_avatar (edited July 29, 2000).]

I can't give you any statistical results, but my experience with MySQL has been:

1. I have never seen it hit a serious bottleneck on ANY of my production or development servers. (one of my development servers was a Pentium 90 with 16 MB RAM running FreeBSD, and I was able to develop a complete e-commerce application on it without any noticeable drag in searching through thousands of records.)

2. It may not have transactions (commit/rollback), but I have never seen MySQL corrupt a table, even from random reboots.

3. The benchmarks I have seen generally show one thing: MySQL is the fastest way to serve out data to a dynamic website. It may not have the best performance in INSERTS, UPDATES, or some of the more complex operations, but for simply serving out data, (especially with many hits using simple queries), nothing is faster.

4. By the time you are ready to move to Oracle or Informix, you may not want to. MySQL 3.23.xx is almost out of beta, and will support many more complex operations than previous versions, including transactions, subqueries, etc... AND it will support table sizes in the multi-terabyte range. You might even want to start your development with version 3.23.21, which is a fairly stable beta version.

5. On the other hand, you just might want to check out PostgreSQL, which is already much more sophisticated (if not quite as fast) than MySQL.