morgo

24 ms·

211.
by morgo 12y ago
The built in query stats is similar to Performance Schema in MySQL. The statistics are quite fine grained, and with a set of views on top ( https://github.com/MarkLeith/mysql-sys ) are very useful for observability.
212.
by morgo 12y ago
The analogy I would use for thread-pool is to insert a waiter in front of the chefs in the kitchen. It doesn't make sense for all workloads, but I have found thread pool to be useful in cases where application servers can overload data
213.
by morgo 12y ago
It is worth mentioning that commercial distributions of MySQL feature pool-of-threads as well. Percona Server features what is essentially a re-implementation: http://dev.mysql.com/doc/refman/5.5/en/faqs
214.
by morgo 12y ago
Delayed replicas is a MySQL 5.6 feature (and can be emulated in previous versions). You are correct in that the main use-case is fast recovery from human error. But as a DBA, I can tell you that accidents like this cover 90% of disasters :
215.
by morgo 12y ago
FWIW, MySQL's InnoDB Storage Engine switched a few years ago from an LRU to something similar to what is described as the final solution here. The LRU is split to a hot (5/8ths) and cold (3/8ths) chain. Scans can only evict
216.
by morgo 12y ago
There's lots of development in this space all of the sudden. From Oracle there is also MySQL Fabric. Fabric also provides sharding + failover, but does so by using modified MySQL connectors rather than a proxy.
217.
by morgo 12y ago
I would say that the biggest risk with mysqldump is restore time. If a significant portion of the buffer pool is in memory, it may actually dump quite fast. MySQL 5.5 and above also have a modified LRU algorithm so that tablescans (i.e. my
218.
by morgo 12y ago
The non-standard TIMESTAMP behavior is deprecated btw, and can be disabled via --explicit_defaults_for_timestamp: http://dev.mysql.com/doc/refman/5.6/en/server-system-variabl...
219.
by morgo 12y ago
I also use the same handle as a conference. Me: @morgo Conference: @morgo_nz I impersonated them with flight of the conchords jokes in 2011: https://twitter.com/morgo/status/128488872165384195 Immature on my part.
220.
by morgo 12y ago
The presentation was probably this one: http://www.percona.com/live/mysql-conference-2014/sites/defa... Slides 16-18.
221.
by morgo 12y ago
ALTER TABLE .. TRUNCATE PARTITION should be much faster than deleting as well. At least it is in MySQL.
222.
by morgo 13y ago
Facebook has previously answered this question on Quora: http://www.quora.com/MySQL/Is-Facebook-considering-ditching-... Also from the WebScaleSQL FAQ: Q: Why didn't you base this on MariaDB, Percona Server, Drizz
223.
by morgo 13y ago
It would not be a stretch to call percona server a branch as well. It has rebased twice already (5.5 and 5.6) and tries to focus on high impact small changes. Not to make excuses, but oracles position of supporting releases for 8 years on m
224.
by morgo 13y ago
Row-level locking, mvcc, multiple write threads. Mysql (default Innodb engine) is actually pretty good at io scalability since ~ 5.5 (2010). We have google, percona, Facebook, oracle to thank for this.
225.
by morgo 13y ago
In response to your last point: MySQL 5.6 and above have online DDL.
226.
by morgo 13y ago
Yeah, the MySQL 5.6 defaults are quite different to the ones mentioned here. Also: - the query cache is now off by default (but 1M) - innodb_additional_mem_pool_size no longer applies. - some of the per-session variables can actually alloc
227.
by morgo 13y ago
My understanding is that it just implements the postgres protocol, its query planner is now completely different. "ParAccel's DBMS engine is built for analytics, initially based on PostgreSQL" -> http://en.wikip
228.
by morgo 13y ago
I wonder how true that is now. There are mergers and acquisitions in companies of all sizes, so to some degree no jobs are safe.
229.
by morgo 13y ago
This is the same problem as to why slaves were single-threaded for so long: Statement based replication (default) is tricky to apply in parallel, since you can't easily figure out ordering dependencies. In MySQL 5.6 replication is now
230.
by morgo 13y ago
Having a VPN service like this is useful for testing Geo-based features. I've used / can recommend https://wonderproxy.com/
231.
by morgo 13y ago
I am a database administrator. I somewhat agree with the conclusion which essentially says having your on premise DBA is on the decline. Hiring DBAs has always been tricky for employers anyway. It's a position of responsibility (to pr
232.
by morgo 13y ago
Totally agree. Even if this concept doesn't work perfectly, it puts pressure on couriers to give Amazon the discounts they want. Its kind of like how large enterprises use an Open Office support quote to bring down the cost of their M
233.
by morgo 13y ago
For context, the first NoSQL interface for MySQL was: http://yoshinorimatsunobu.blogspot.ca/2010/10/using-mysql-as... (It's appeared on HN a few times). The MySQL team decided to run with a more standard memc
234.
by morgo 13y ago
I recommend switching to Row-based Replication, which avoids this problem.
235.
by morgo 13y ago
What @kozlovsky said :) Only advice I would add, is SELECT .. FOR UPDATE doesn't (and shouldn't) use MVCC, so it guarantees you the most recent version of the row. In my comment I said the use case for this feature is not typical
236.
by morgo 13y ago
InnoDB uses MVCC, so the only operations that respect the SELECT .. FOR UPDATE queries are updates. As long as you use transactions, I wouldn't say this is a common application pattern. It was good for the article to mention gap locki
237.
by morgo 13y ago
AWS has a lot of managed infrastructure components like RDS, EMR, S3.
238.
by morgo 13y ago
I work for Oracle (MySQL team). It was going to a local meetup group in 2005 and giving a talk that eventually led me down this path.
239.
The slow decline of using MySQL read slaves(tocker.ca)
1 points by morgo 13y ago|0 comments
240.
by morgo 13y ago
Seems similar to Freshbooks.com fwiw.

More ›