The summer break seems to be over and the event season is heating up again! There is a number of conferences and events coming up in the next months — here is a quick summary of the events that I plan to attend.

This Friday I will attend an event here in Hamburg: the "Silpion Sommerfest", organized by Silpion (a local IT solutions provider which is a partner of Sun Microsystems as well). I will be there to network and talk about MySQL.

This coming weekend (2009-09-12/2009-09-13), there will be the PHP Unconference here in Hamburg, Germany . It will consist of two days of Barcamp-style sessions about …

The Facebook patch for MySQL has more performance counters to understand the sources of disk IO from InnoDB. The data below is from a slave that is idle except for the replication workload.

----------------------
BUFFER POOL AND MEMORY
----------------------
Total memory allocated 53455038988; in additional pool allocated 20971520
Buffer pool size   2949120
Free buffers       0
Database pages     2937077
Modified db pages  214123
LRU old pages      1101382
Read ahead: 3278457 random, 3550809 sequential
Pages dirty: 7.26 percent
Pending reads 0
Pending writes: LRU 0, flush list 0, single page 0
Total writes: 129932 LRU, 3860863 flush list, 0 single page
Write sources: free margin 129932, bg dirty 61259, preflush 3742640
Neighbor pages flushed: 3013232 from list, 66315 from LRU
Pages read 11168431, created 115089, written 3990795

The server has read ~11M pages and ~6.8M have been for …

Matt Asay has written an interesting post speculating that Oracle might use the delay caused by the European Commission investigation into its acquisition of Sun to drive the price down. Sounds reasonable enough to me.

In it, Matt makes a couple of statements, one I agree with: “Oracle… likely will prove to be a better manager of this asset than Sun was”; and one that I have real doubts about: “MySQL’s… doing just fine, thank you”.

MySQL might well be doing fine. Unfortunately Sun’s financial results don’t actually provide any evidence either way.

Billings for the MySQL/Infrastructure were up 51% to $313m in FY09, according to information presented with Sun’s financial results, with revenue hitting $100m (up 10%) in …

Andrey mentioned it long ago: MySQL Connector/C++ 1.1.0 makes use of some Boost components. The change impacts those who compile the MySQL driver for C++ from source. You need to have Boost 1.34.0 or newer installed on your build system. Users of a binary distribution do not need to take any action. Boost is not required when using the pre-compiled binaries. Nor is Boost needed to run client applications that make use of Connector/C++.

Boost is a highly regarded cross-platform library for C++ development used by many projects, for example OpenOffice.org. Some parts of Boost are included in the C++ Standards Committee’s Library Technical Report (TR1) and will be in the new C++0x Standard. Several other parts are either additions to TR1 or proposed for …

Andrey mentioned it long ago: MySQL Connector/C++ 1.1.0 makes use of some Boost components. The change impacts those who compile the MySQL driver for C++ from source. You need to have Boost 1.34.0 or newer installed on your build system. Users of a binary distribution do not need to take any action. Boost is not required when using the pre-compiled binaries. Nor is Boost needed to run client applications that make use of Connector/C++.

Boost is a highly regarded cross-platform library for C++ development used by many projects, for example OpenOffice.org. Some parts of Boost are included in the C++ Standards Committee’s Library Technical Report (TR1) and will be in the new C++0x Standard. Several other parts are either additions to TR1 or proposed for …

1. Go to Profiles
2. Click on Advanced
3. Highlight Active profile
4. Click on Menu and select Edit
5. SMS Text/Phone/MMS etc.
6. Select Tone
7. Ring Tone: Change as desired
8. Volume: High/Medium/Low
9. Click back button
10. Select Save

Testing Secure Replication

For the first test I used encrypted (SSL) replication and inserted 200,000 records using three 10 minute disconnection intervals per hour.

After several hundred thousand of inserts, deletes and updates on the SOA and RR tables simultaneously over a normally connected SSL replication channel I have the following results:

RR Update Test
Test 1:
RR Records Updated (Changes Set) = 129145
Elapsed Time = 14147.411601
RR Updates/Second = 9.13
RR Updates/Minute = 547.8
Europe MySQL Slave RR Update Counts (Changes Set) = 129145
Asia MySQL Slave RR Update Counts (Changes Set) = 129145
Test 2:
RR Records Updated (Changes Cleared) = 129145
Elapsed Time = 10168.638867
RR Updates/Second = 12.70
RR Updates/Minute = 762
Europe MySQL Slave RR Update Counts (Changes Cleared) = 129145
…

Testing Non-Secure Replication

A series of experiments were conducted to determine whether MySQL replication would prove to be reliable with SSL enabled. Please note that all tests were conducted using the MyDNS schema with includes the SOA and RR tables on MySQL 5.1.

The first experiment sets focused on replication operations and not on a predetermined set of Insert, Update or Delete patterns. So Inserts were used since they are the easiest to tag and verify. Again, the focus is on replication channel fault recovery.

Later iterations used simultaneous Inserts, Deletes and Updates against both SOA and RR tables from production snapshots of the mydns database, with request sets greater than 100,000 rows each, over an SSL replication channel.

The first test was done by closing the Slave server's outbound SQL port to the Master with iptables, the Slave was uaware it was no longer …

This is an old post I put on the old mysqlhow2.com that ended up on modphp.org. Both sites are run by people I know so it's coo. Anyway, I thought it should be here as well.

Lets look at the results of the query without returning all the results:

NOTE: Use EXPLAIN
- Explain will show the possible rows to be examined in your query. (The more rows the slower the return)
- Example of EXPLAIN: (we will be using the world database by mysql.com)

EXPLAIN SELECT * from City where CountryCode =’USA’ and Population > ‘1000′ order by District;
Results:
+—-+————-+——-+——+—————+——+———+——+——+—————————–+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra |
…

MySQL replication is a great feature: it's easy to setup and relatively easy to manage, it can provide scalabilty and availability, a master can serve dozens of slaves, etc. However, sometimes people have the problem that the slaves can't keep up; they get further and further behind, which is a major problem. So this post talks about why that happens (sometimes) and what you can do about it.

Causes:

• The most common reason that a slave cannot keep up with the master is the SQL Thread. There are 2 threads on the slave that handle replication: the IO Thread which connects to the master and pulls down the binary log events, and the SQL Thread which executes these events on the slave. But the master might have dozens of concurrent threads executing inserts/updates/deletes. So when they get to the slave, they are run in a single thread which may not be able to do the same throughput. There's a good reason for the …