« Return to event
Time: 2:30 PM - 3:15 PM
Track: Track 1
The DataBase on Demand (DBoD) service at CERN, was born out an idea to provide a MySQL backend to the CERN community. The service has been designed to allow scalability by means of automation, making a DBoD user a DBA of its own instance(s). The service is based on an open community SLA.
Nowadays DBoD provides about 60 single instances between MySQL and Oracle.
With a time to production of about 6 months, this presentation walks through the project architecture (it reuses many of the IT-DB group software projects), the functionality that we currently provide and the different options we are considering in order to provide some degree of High Availability.
The Backup and Recovery functionality relies strongly on the underlying storage, NAS type. This allows a good level of abstraction and eases some of the operations e.g. full backup duration reduced to a few seconds with almost no impact to a production database.
Top 3 things
1) Some design principles to achieve of a flexible DBaaS on cloud.
2) Effective use of storage snapshots (NFS access) for backup and recovery solutions.
3) Ideas to get on board High Availability with less cost possible.
- Ruben Gaspar Aparicio, CERN
Published: 25.01.2013 - This document is restricted to members