What a busy week – between O’Reilly and Collaborate we gave a number of talks, sponsored a booth, and took some long flights in between the shows. But what an interesting week. The MySQL conference is a great time to catch up with customers and, as always, we heard some excellent feedback on a variety of topics. From a storage engine provider’s perspective, we were happy to hear about the improvements to the Optimizer, and appreciated having the opportunity to attend the Storage Engine Summit on Friday.
Here are a few other areas we found of interest:
1. Big data issues were everywhere to be seen and heard
- As the SF MySQL Meetup team (@sfmysql) tweeted from Baron Schwartz’s talk: Software not keeping up w/hardware, hardware not keeping up w/data
- According to Erin O’Neill (@eonarts) : Twitter has 1 billion tweets/wk! (6,939 tweets/sec on japanese new year)
- As Selena Deckelmann (@selenamarie) noted: All the big database vendors totally whiffed on the internet.
- From Gennady Gurov (@ggurov) we heard: DBAs need to know how to work LVM and backend snapshots to do their job on large data
- And as @sfmysql perhaps best wrapped it up: the future is not Big Data. It’s Extreme Data.
2. MariaDB development continues strong
- Even as 5.3/5.5 are almost in beta, Monty is planning well in advance for 5.6
- Noted Colin Charles (@bytebot): MariaDB makes subqueries FAST
3. Replication is an area of challenge, opportunity, and plenty of discussions
- According to @sfmysql: Seems like amazing work on replication is coming from all sorts of groups. Still think replication is buzz word for the year
- And as we tweeted from Sarah Novotny’s session on replication
o Replication issues: Things can fail spectacularly, data drift, scaling writes, multiple DBAs
o Replication challenges: When things go wrong, they go very, very wrong
o Replication advantages: Never a dull moment, near infinite read scaling, identical data many places
Some of these topics were familiar. For example, we keep getting pulled into big data environments as our ability to adeptly handle big data without partitions, fragmentation or downtime for schema changes insures optimal performance. As for the second point, our latest release works well with MariaDB and we are glad to see continued growth here. Other areas provided new perspectives – we heard an overall strong need for improved replication solutions, and got specific feedback from companies we spoke to at the show who are struggling to deal with distinct replication issues such as the single-threaded ingress bottleneck on slave configurations (slave lag).
Note that this quick blog is by no means complete or definitive — there were way too many good sessions going on at the same time for us to see it all. In fact, I recommend you read many of the other blog posts and articles already posted on the show – including Mark Callaghan, Shlomi Noach (part I, II, III and IV!) and Andy Oram to name a few.
Thanks to all who came by our talks and stopped by the booth. It was also great to meet with other bloggers at the show – thanks to Shlomi for the feedback on replication, Roland for the conversation on data warehousing and NoSQL solutions, Anders for coming by the booth to chat, Sheeri for the wild ride, and Sarah for sharing her thoughts on overall MySQL trends. And we got snapped in some great photos (notably from an O’Reilly photographer and a SkySQL photographer). Here are booth photos (1, 2), and one from the fantastic SkySQL party (4).
And as much as I enjoyed racking up the miles, I would have to agree with Giuseppe – I would really like to see O’Reilly in Boston next time as well. At least I made it back in time for the big battle yesterday…
