Wednesday, January 29, 2014

Talking Trove with Baz #CloudOps @Lithiumtech

Last night I had an opportunity to hang out at the @Lithiumtech office for their San Francisco based #CloudOps meetup and learn all about an OpenStack project which is still in the incubation stages but has a really bright future.  Michael Basnight, Trove PTL from Rackspace came out fully decked in his three cat shirt and unicorn hat to spread the love and recruit more troops to bring the hopes and dreams of the project to life.



As a database administrator, your job time should be spent doing valuable things.  Installing a database server is not one of those valuable tasks.  If you have to install that database server repeatedly, that is definitely not time spent well.  I would equate the joy of installing a database server with going to the dentist for a triple root canal.  The intricacies of installing and configuring a database, setting up the proper security settings, initializing database instances and tables are all things that take a considerable amount of time.

Let's not even get into the complexities of change management or those pesky enterprise silo teams that put the brakes on even the most critical of projects.  There's nothing like bringing in the systems, storage, software, network, security and BI teams all together to thwart the efforts of business units and partners alike.

To put it bluntly, installing and configuring a database server sucks.  So wouldn't it be nice if you could click the easy button to install a database server (so you can leave early for beer)?

The Trove project is aimed at delivering Database-as-a-Service on top of OpenStack's Infrastructure-as-a-Service.  The idea is that, as a consumer of cloud services, instantiating a database server should be easy.  The cloud user should be able to select from a list of database servers such as mySQL, PostgreSQL, mongo, cassandra and, potentially, others.  Using a standard API the instance can be launched, managed and even auto-scaled in future implementations.

With OpenStack, these are all things that can be automated with Trove, abstracted away from the user, avoiding all of those pesky silos and change management forms.  Then all of those people whose jobs used to involve repetitive manual tasks can also do more important things (or leave early for beer with you).

At the moment, Trove is an extremely young project.  With great luck, Trove is on it's way to becoming an integrated project.  Unfortunately, as it currently stands with the OpenStack Foundation and the Technical Committee, Trove is confined to being incubated until it takes certain steps.

What will it take for Trove to become integrated and, eventually a core project?  You!   The Trove project needs people who are willing to work on the code base and have the technical experience to contribute.

Some of the things currently in development include:

Support for additional databases beyond mySQL
Support for redundant database instances
Integrated backup mechanisms for upstream databases that don't support backup
Support for provisioning with Heat

For a full list of Trove blueprints see: https://blueprints.launchpad.net/openstack/trove
To contribute code to OpenStack see: https://wiki.openstack.org/wiki/HowToContribute

It was a lot of fun listening to Baz talk about Trove and that's not just the beer talking.  Be sure to come out to Lithium for the next CloudOps meetup (http://www.meetup.com/cloudops) and join in on the excitement.

Oh, and you might win a tablet.

(Also, did I mention the free beer....)