Oracle tweaks MySQL with milestone update
NoSQL ain’t the only story in town
OpenWorld While the news about NoSQL has garnered much attention, Oracle has quietly published a development milestone release (DMR) for MySQL.
The MySQL 5.6.3 DMR includes a major revision of the software’s optimizer, which the company claims will make file-sort optimizations up to three times faster by searching more intelligently and dumping unneeded data during the process.
Index condition pushdowns were also speeded up, with Oracle reporting that operation times for the function had been cut from 15 seconds down to 90 milliseconds.
Batched Key Access has been added to augment the Multi-Range Read capabilities added in the 5.6.2 DMR, enabling indexes to be sorted by data sequence for faster searching. In tests this dropped some execution times from 2000 seconds to 10, albeit with a “sufficiently large” join buffer.
More instrumentation has been added to the performance schema in the new release and further improvements have been made in InnoDB and replication.
With over 40 sessions devoted to MySQL at the Oracle OpenWorld conference this week, it’s clear that the company is maintaining its investment in the MySQL code. That said, some developers have commented privately that they fear for the long-term development of the platform, particularly its perceived creeping commercialization.
MySQL help needed
my company is trying to persuade me that we can use MySQL the company critical work, whereas I'm trying to push them towards something grown-up. I can give them my (unimpressive) experiences of MySQL, and various bits from other places to try and discourage them including its dire track record, but can anyone point to serious flaws in the thing as it stands (I know there were disastrous bugs but they seem to have been fixed by now). I can make a pretty good case already, but I want something really damning.
Bear in mind that to them a database is just a database, whether at the Access or MSSQL, so that the one is effectively free and the other costs thousands that means that they'll go for the cheap one (although I'm getting there... I shall prevail...)
(NB Scalability is not an issue for us)