Comments on: MySQL 5.7 By Default 1/3rd Slower Than 5.6 When Using Binary Logs https://www.percona.com/blog/binary-logs-make-mysql-5-7-slower-than-5-6/ Wed, 03 Oct 2018 21:00:16 +0000 hourly 1 https://wordpress.org/?v=6.5.2 By: Meral Sönmezer https://www.percona.com/blog/binary-logs-make-mysql-5-7-slower-than-5-6/#comment-10969772 Wed, 03 Oct 2018 21:00:16 +0000 https://www.percona.com/blog/?p=35923#comment-10969772 great post, thanks for this article.

]]> By: Rick Pizzi https://www.percona.com/blog/binary-logs-make-mysql-5-7-slower-than-5-6/#comment-10969316 Tue, 26 Jun 2018 16:19:41 +0000 https://www.percona.com/blog/?p=35923#comment-10969316 guys please mind the consequences of setting this to zero

]]> By: Davoice https://www.percona.com/blog/binary-logs-make-mysql-5-7-slower-than-5-6/#comment-10968899 Tue, 16 Jan 2018 09:06:15 +0000 https://www.percona.com/blog/?p=35923#comment-10968899 Thank you for writing this article, we didn’t know what was making the writes so slow in our system! In our case it was not 1/3x slower, but 1300x slower!!!

]]> By: Roel Van de Paar https://www.percona.com/blog/binary-logs-make-mysql-5-7-slower-than-5-6/#comment-10968031 Thu, 27 Apr 2017 02:53:33 +0000 https://www.percona.com/blog/?p=35923#comment-10968031 Hi Mannoj. Not sure I am following your reply. Please note that this is in the (what is upstream for us) MySQL server. It also makes sense to be able to configure these variables on/off for different configurations. Let me know if I missed anything.

]]> By: Mannoj https://www.percona.com/blog/binary-logs-make-mysql-5-7-slower-than-5-6/#comment-10968020 Mon, 24 Apr 2017 07:06:31 +0000 https://www.percona.com/blog/?p=35923#comment-10968020 But why you leave default settings for file sync variables. Would this be better if such important variables are hardcoded and not picked as default.

]]> By: Stephen N https://www.percona.com/blog/binary-logs-make-mysql-5-7-slower-than-5-6/#comment-10968019 Sun, 23 Apr 2017 23:16:57 +0000 https://www.percona.com/blog/?p=35923#comment-10968019 Thanks for this great post. We just upgraded from 5.6 to 5.7 and have been scratching our heads trying to figure out why its slower.

]]> By: Dave Holmes https://www.percona.com/blog/binary-logs-make-mysql-5-7-slower-than-5-6/#comment-10966638 Mon, 27 Jun 2016 21:19:41 +0000 https://www.percona.com/blog/?p=35923#comment-10966638 @Roel great analysis and your blog post and I think the title of the article was spot on, we were in an emergency situation and had to clone a MySQL slave server and had no choice but to throw an 5.7 box into production.

We saw the IO usage rise significantly but IO throughput remain low totally unexpected and not something we had seen with other 5.7 production servers.

As ever your information rocks guys! hope to make Amsterdam this year! #perconalive

]]> By: brandentimm https://www.percona.com/blog/binary-logs-make-mysql-5-7-slower-than-5-6/#comment-10966611 Thu, 16 Jun 2016 17:08:37 +0000 https://www.percona.com/blog/?p=35923#comment-10966611 It would be interesting to see the results with Performance Schema turned off. A quick investigation shows a 15% increase in the number of instruments enabled by default between 5.6 and 5.7.

]]> By: Roel Van de Paar https://www.percona.com/blog/binary-logs-make-mysql-5-7-slower-than-5-6/#comment-10966566 Sun, 05 Jun 2016 23:27:21 +0000 https://www.percona.com/blog/?p=35923#comment-10966566 @Scott & @Mark – In regards the (now removed) title “Binary logs make MySQL 5.7 slower than 5.6” – yes! It looks like our blog team edited the original. I have restored my original title (“MySQL 5.7 By Default 1/3rd Slower Than 5.6 When Using Binary Logs”) and introduction! Thank you for the heads up!

]]> By: Mark Callaghan https://www.percona.com/blog/binary-logs-make-mysql-5-7-slower-than-5-6/#comment-10966560 Fri, 03 Jun 2016 23:15:45 +0000 https://www.percona.com/blog/?p=35923#comment-10966560 I think the title for this blog post is misleading. 5.7 tests ran with sync_binlog=1, 5.6 ran with sync_binlog=0. You ran a test with different configurations. I do that all the time. I prefer that the default not get changed in the middle of the 5.7 lifecycle, but that needs a different title.

]]> By: Scott Klasing https://www.percona.com/blog/binary-logs-make-mysql-5-7-slower-than-5-6/#comment-10966559 Fri, 03 Jun 2016 22:24:22 +0000 https://www.percona.com/blog/?p=35923#comment-10966559 Seems a bit misleading to suggest 5.7.x is slower than 5.6.x since comparing sync_binlog=0 is not an apple to apple comparison of sync_binlog=1. Granted when both versions use sync_binlog configuration defaults it is slower, but the point is even with 5.6 by itself, 5.6.x is slower when sync_binlog=1 because it waits for the OS to push to disk. More importantly the discussion should highlight that yes systems are at risk when sync_binlog=0 and the database crashes from an OOM. All data in flight not persisted to disk will be lost, and often corrupts the database to such extent it is not repairable. Therefore a system should always default to safe values and let the administrator choose when to be at risk for performance reasons. This reminds me of relational databases back in the 1980’s where IBM’s Database 2 (DB2) defaulted to some of the worst locking configurations(isolation level and timeout) and it required a knowledgeable consultant to know the default values were bad.

]]> By: Singer Wang https://www.percona.com/blog/binary-logs-make-mysql-5-7-slower-than-5-6/#comment-10966557 Fri, 03 Jun 2016 17:33:47 +0000 https://www.percona.com/blog/?p=35923#comment-10966557 I’d argue that sync_binlogs isn’t needed for the technical definition of ACID.

]]>