I haven’t played around with this in a little while, so there may be some new tricks available out there, but one possible way would be to use the method above to move your 2006 data into a separate table and/or database with the same structure, and move the empty 2006 to your ‘live’ partitioned table.

Then apply your reorg to the ‘offline’ table to consolidate the 2006 data. Then apply the reorg to the ‘live’ table for 2006, and this should be a fast operation since it’s just reorg’ing empty tables. Then do a swap of the real data from the offline table (again, as above) and you should be able to accomplish consolidating your 2006 data without taking the rest of the table down.

Do I recommend doing this? err… see if you can get your users to accept some downtime If not, make lots of backups and think VERY carefully about every single step!

I’m really not try to kick MySQL while it is down, but I do want to speak up as a member of the user community about what things I will look for from IBM if they end up owning MySQL. You have to admit, the publicity generated in the last few months with Monty publicly blasting the 5.1 GA decision, and then leaving Sun altogether, has not been good. Of course other companies can and will step in to support the product, but what I believe has made MySQL a great success is that there has been a strong corporate backer, in touch with the needs of users, providing direction and leadership. The events of the last 6 months have made MySQL appear divided over its direction and making users nervous about what the future has in store.

If I am running MySQL on a site right now, I’m not going to dump it just because I’m a bit nervous about direction. But if I’m planning a new project, I may actually think twice now about whether to choose MySQL, given all the signals of late. Maintaining the product front and center in developer’s minds is key.

As for the tracking of customers, I have trouble understanding what other reason Oracle or IBM would have to be involved with MySQL. As you say, a lot of the big internet companies are forthcoming about what technologies they use, but this is only a small part of the overall (lucrative) database market.

For example, at one financial company i worked at as a student during the dotcom era, I built a local datamart with MySQL, partly in response to the slow turnover time of changes to the main DB2 warehouse and datamarts. For all I konw, that datamart is still kicking, and has been draining IBM of revenue they “should” be getting for one of their entry-level DB2 products for all these years. IBM needs to know who these “lost” customers are so that they can monetize them, or at least this is the approach I expect them to take — giving away free, stripped-down versions of DB2 and Oracle doesn’t seem to be doing anything to stop the growth of OSS databases because cost is not everything.

Sun is not the only company supporting MySQL, there is a significant amount of expertise present in various other companies and that includes development. There are services for emergency needs (where sufficient internal expertise is available), but also sevices for remote DBA work with design and developer support. Sun itself does not actually provide a broad range of services in terms of the actual MySQL market.

MySQL release inflation is already a fact of life, and came into play around 5.0. I’d want it to disappear.

Finally, I still don’t believe Oracle is into InnoDB to find out about MySQL customers. People were suggesting this early on too. But most of the bigger customers are abundantly public, there’s no additional inside info that would be of such big value….

So, if you’re working for Big Corporate(tm) and are in a position of relative power, a “community” edition of MySQL may not be appropriate, not because its not good enough, but because you’re dependent on the goodwill of the community, and your own team, to help you if you run into trouble.

However, if you’re working for a non-politicized web start up where just getting stuff done cheaply is important, you’d be amazed how far you can stretch mysql.

Where you are along that spectrum between saving cold hard cash and buying merely a security policy dictates what product you go with (assuming you have unlimited cash or unlimited talent/willingness to go it alone). While I haven’t actually tried it, based on the specs alone, InfoBright does seem to be a nice middle ground between a budget-busting Oracle RAC DW and a homegrown mysql setup.

You will find some info on blogs here and there that are mysql-specific, but i think that mysql as a DW solution doesn’t seem to be extremely popular just yet, or, most smaller companies build basic MySQL solutions themselves until they have enough money and scale to go with an established vendor. Take a look at Ralph Kimball’s warehousing book, it’s like the bible of DWs.

We are just starting out on a 1T-5 T DW for a web startup. I am curious- do you think MySQL is upto it (w/o resorting to the new fangled INFOBRIGHT) etc. ?

Any pointers on published /web resources that even vaguely describe multi T MySQL DWes ?