Monday, January 24, 2011

Live Webinar: What MySQL replication cannot do. And how to get around it.

Thursday, January 27, 10 am PT/1 pm ET

MySQL built-in replication is the greatest component of the web economy, Its model of scaling out has allowed the growth of companies such as Google, Yahoo!, Facebook, Wikipedia, and countless more. However, this simple model has limitations. Some you can live with, and some will send DBAs in a quest for other tools, from which they often come back empty handed.

This webinar covers some of these limitations, and how Tungsten can overcome them:
  • Multiple source replication, also known as the headquarters that wants to synchronize data from its branches. Impossible to do with MySQL. An easy job for Tungsten.
  • Automatic slave promotion and demotion, both for failure recovery and maintenance. In MySQL it requires manual synchronization of data, and sometimes results in data loss. For Tungsten, it is child's play.
  • Seamless upgrade of replicated servers. You can do that in MySQL, but it's all manual and error prone labor. Tungsten simplifies the task and makes it safer.
  • Parallel replication. In MySQL, slaves must queue all their transactions into a single thread. Tungsten allows parallel replication per database, speeding up the data transfer way faster than native replication.

Wednesday, January 12, 2011

Live Webinar: MySQL operations made simple

Live webcast Thursday, 1/20 at 1 pm ET/10 am PT

The web economy requires prompt data availability and scalability. While MySQL is the backbone of web applications, its manageability in replicated clusters still requires manual labor, can lead to data loss, and it is error prone. DBAs want rapid deployment of new data nodes and even more rapid replacement of failing ones without data loss.

Unlike MySQL built-in replication, Tungsten provides easy administration, allowing the DBA to deploy and replace data nodes and whole clusters safely in a few seconds.

Servers fail all the time. Don't make your emergency plan based on assertions like "if the server fails". Every server will ultimately fail, and you either have a seamless fail over and fail back procedure, or you are in for some unpleasant task. Tungsten is that seamless procedure.

This webinar demonstrates what chores a DBA has to go through when promoting a slave and synchronizing the remaining unevenly updated nodes. We'll show how a well designed tool like Tungsten can do the job in seconds rather than hours.