Tuesday, April 20, 2010

Around the volcano and home from US in 20 hours

"Sir, we advise you to not board this aircraft". The plane was ready for departure but the empty gate seemed to underline what the ground staff just said. All other flights to Europe were cancelled. And there are certainly worse places to be stranded than San Francisco in spring. After speaking at the MySQL Users Conference my already re-scheduled via Frankfurt in two days was likely to be cancelled again. I had to try something to avoid being stuck in US for days if not weeks.

7 hours later cold air streaming in through the front door welcomed me to Reykjavik. At departure time in US no onward flights were scheduled from there. Eyjafjallajökull (see picture) was still erupting sending its ashes to Europe. Rather than being stuck in US I could be easily stuck on Iceland. But I wanted to see my family. And I wanted to see the volcano.

By the time of my arrival on Iceland individual transfers to Northern Europe opened, I got a seat to Oslo and arrived at a deserted ghost airport. The only flight for the day would bring me there to my hometown Stockholm. The ground crew in US wished "Good Luck" when boarding. 20 hours later I had dinner with my family. 2 days ahead of schedule and no extra net travel time.

Monday, February 01, 2010

MySQL Cluster 7.1.1 is there

A new version of MySQL Cluster 7.1 beta has been released and is available from our ftp directory at


The Cluster 7.1 beta code base is identical to Cluster 7.0 GA. While maintaining our GA quality throughout all core functionalities we added two of the most requested enhancements as independed features to 7.1 beta:
  • ndb$info with SQL level real-time monitoring of Cluster
  • an easy-to-use and high performance native Java interface and OpenJPA plug-in

ndb$info makes cluster status and statistics available on SQL level. Log into the MySQL Server and simply use SQL to retrieve configuration details, memory status or the node status and uptime:
mysql> SELECT * FROM ndbinfo.nodes;
| node_id | uptime | status | start_phase |
| 2 | 45678 | STARTED | 0 |
| 3 | 45676 | STARTED | 0 |
| 4 | 45678 | STARTED | 0 |
| 5 | 312 | STARTED | 0 |
4 rows in set (0.03 sec)

ndb$info is designed to add little to no overhead even for extensive monitoring of Cluster.

MySQL Cluster Connector for Java

Designed for Java developers, the MySQL Cluster Connector for Java implements an easy-to-use and high performance native Java interface and OpenJPA plug-in. Using the Java Persistence Interface of Cluster/J a primary key lookup is as simple as
Fish e = session.find(Fish.class, 4711);

The OpenJPA for Cluster implementation plugs into Apache OpenJPA and provides native access to Cluster. Many transactions can now be directly executed on Cluster, without the MySQL Server being involved. This allows much faster query execution and higher throughput.

You can read more about Cluster/J and how to use it here:

To learn more about and have the chance to meet the architect and developer behind Cluster/J you might want to attend the webinar. More info here: