Oracle have just made availble the Release Candidate for MySQL Cluster 7.4 (MySQL Cluster 7.4.3) – it can be downloaded from the development release tab here. Note that this is not a GA release and so we wouldn’t recommend using it in production.

The delta between this RC and the 7.4.2 DMR can be viewed in the MySQL Cluster 7.4.3 Release Notes

There are three main focus areas for this RC and the purpose of this post is to briefly introduce them:

• Active-Active (Multi-Master) Replication
• Performance
• Operational improvements (speeding up of restarts; enhanced memory …

The other day a customer asked me to do capacity planning for their web server farm. I was looking at the CPU graph for one of the web servers that had Hyper-threading switched ON and thought to myself: “This must be quite a misleading graph – it shows 30% CPU usage. It can’t really be that this server can handle 3 times more work?”

Or can it?

I decided to do what we usually do in such case – I decided to test it and find out the truth. Turns out – there’s more to it than meets the eye.

How Intel Hyper-Threading works

Before we get to my benchmark results, let’s talk a little bit about hyper-threading. According to Intel, Intel® Hyper-Threading Technology (Intel® HT Technology) uses processor resources more …

TokuDB offers high throughput for write intensive applications, and the throughput scales with the number of concurrent clients.  However, when the binary log is turned on, TokuDB 7.5.2 throughput suffers.  The throughput scaling problem is caused by a poor interaction between the binary log group commit algorithm in MySQL 5.6 and the way TokuDB commits transactions.   TokuDB 7.5.4 for Percona Server 5.6 fixes this problem, and the result is roughly an order of magnitude increase in SysBench throughput for in memory workloads.

MySQL uses two phase commit protocol to synchronize the MySQL binary log with the recovery logs of the storage engines when a transaction commits.  Since fsync’s are used to ensure the durability of the data in the various logs, and fsync’s can be very slow, the fsync can easily become a bottleneck.  A …

One of great innovation in MySQL 5.7 is memory summary tables в Performance Schema and corresponding views in sys schema

And as troubleshooting freak I have huge reason to greet this feature.

Before version 5.7 we had very limited abilities to diagnose memory issues in MySQL. We could use operating system tools, such as vmstat, top, free, but they only showed what MySQL server uses memory, but do not show how. In version 5.7 things changed.

Lets examine what can we study about memory usage by MySQL Server.

At first, this is total amount of memory, used by all internal MySQL structures:

mysql> select * from …

Thanks to everyone who attended and participated in last week’s webinar on 'Galera Cluster for MySQL vs MySQL (NDB) Cluster: A High Level Comparison'. If you missed the sessions or would like to watch the webinar again & browse through the slides, they are now available online.

In this webinar, Severalnines VP of Products, Alex Yu, who was part of the team at Ericsson who originally developed the NDB storage engine in 2001, gave an overview of the two clustering architectures and discussed their respective strengths and weaknesses: 

1. MySQL Cluster architecture: strengths and limitations
2. Galera Architecture: strengths and limitations
3. Deployment scenarios
4. Data migration
5. Read and write workloads (Optimistic/pessimistic locking)
6. WAN/Geographical replication
7. Schema changes
8. Management and monitoring …

In MySQL 5.6 we introduced a huge improvement in the way that index and table statistics are gathered by InnoDB and subsequently used by the Optimizer during query optimization: Persistent Statistics. Some aspects of the way that Persistent Statistics work could be improved further though, and we’d really like your input on that.

How much to sample?

The statistics are gathered by picking some pages semi-randomly, analyzing them, and deriving some conclusions about the entire table and/or index from those analyzed pages. The number of pages sampled can be specified on a per-table basis with the STATS_SAMPLE_PAGES clause. For example:

ALTER TABLE t STATS_SAMPLE_PAGES=500;

This way …

upload on 2014.12 [ten important tips of MySQL database design for better performance] Download this PDF

There are few easy ticks to see what is stuck inside MySQL instance. All these techniques will not give you whole picture, but might help to find queries that block performance. Let’s start from what exactly doing your MySQL server right now.

Which queries are running now?

This will give you an idea what’s running right now so you can find long running queries which slowing down MySQL and/or causing replication lag:

mysql -e "SHOW PROCESSLIST" | grep -v -i "sleep"

It is more convenient than just run “SHOW PROCESSLIST” as it’s hiding all connected threads in “Sleep” state so you’ll get a clean output. Also you can get same output but updating each second:

watch -n1 'mysql -e "SHOW FULL PROCESSLIST" | grep -v -i "Sleep"'

What to look for? This is complex output but you can start with Time and State columns. When you see a query running for more …

When I talk about troubleshooting I like to repeat: "Don't grant database access to everybody!" This can sound a bit weird having one can give very limited read-only access.

But only if ignore the fact what even minimal privileges in MySQL allows to change session variables, including those which control server resources. My favorite example is "Kill MySQL server with join_buffer_size". But before version 5.7 I could only recommend this, but not demonstrate. Now, with help of memory summary tables in Performance Schema, I can show how unprivileged user can let your server to use great amount of swap.

At first lets create a user account with minimal privileges and login.

$../client/mysql -ulj -h127.0.0.1 -P13001
Welcome to the MySQL monitor.  Commands end with ; or \g.
Your MySQL connection id is 10
Server version: 5.7.6-m16-debug-log Source distribution

…

High season is coming, how do you make sure that MySQL will handle the increased load? Stress tests could help with that, but it’s not a good idea to run them in a production environment. In this case Select_scan, Select_full_join and other MySQL counters could quickly give you an idea of how many queries are not performing well and could cause a performance degradation as the load goes up.

Select_scan from SHOW GLOBAL STATUS indicates how many full table scans were done since last MySQL restart. Scanning the entire table is a resource intensive operation. It also forces MySQL to store unnecessary data in the buffer pool, wasting memory and IO resources.

Full scan of a tiny table would be quite fast so missing indexes could stay invisible until the load rises or the dataset grows up. This could also be the case for developers who work with too small data sets on their dev boxes. To prevent performance issues all newly added …