My first real session of the conference was “An introduction to Database as a Service with an emphasis on OpenStack using Trove” by Amrith Kumar (Tesora) and Matt Griffin (Percona). I had originally planned on attending “MySQL and Openstack deep dive” by Peter Boros (Percona), but after speaking with him last night in the vendor hall I realized that this session could have more immediate applicability to my team.
The session started with an overview of the cloud landscape which showed AWS and Microsoft at the top of the cloud market for Enterprises. But there are a lot of other clouds that are getting evaluated and the market is growing fast and changing. 68% of enterprises have less than 20% of their application in a public cloud so a lot of room for growth.
Clouds provide a lot of value but this session concentrated on Speed, Low Cost, and Flexibility. They pointed out that most of the cost savings come in the early phases of a project/application because you don’t have to oversize your infrastructure. Database as a Service provides simplified database provisioning and the ability to manage at scale. OpenStack and Trove help companies provide Database as a Service.
Next they provided a good overview of OpenStack and the services it provides. Then they introduced Sahara (Hadoop as a Service) and Trove (Database as s Service). Since this was a session on Trove there was very little information on Sahara. Trove has to run ontop of OpenStack today, but there are discussions of breaking that dependency in the future. Trove supports multiple databases (MySQL, PostgreSQL, MongoDB, Counchbase, ….). It provides framework for higher level operations like backup and recovery, replication, clustering, HA, etc. They are adding new database platforms with just about every release (twice a year release cycle). The real value of Trove is that the configurations contributed are from industry experts. Trove does not remove the need for expertise, but it frees organizations to roll out quicker because you can get expertise as you go. This helps with dev ops because you don’t have to have a platform expert on each project.
Trove supports a number of different database platforms, but they are all at different levels of maturity. There is some loss of control because o the standardization of instances, but Trove is open source so you can modify it as needed. Configuration groups in Trove is one way you can modify the default behaviors. Trove will add Galera clustering in Liberty. Oracle support in Trove can be purchased. Biggest advantage provided by Trove and OpenStack is a common API to provision resources from private, to public, to hybrid cloud deployments.
All in all this looks like a good project that we should dive into to see how it can increase our effectiveness.
You must be logged in to post a comment.