TroveVision

Trove Vision

Trove is an OpenStack Service which provides simple, intelligent, reliable, and scalable provisioning, monitoring, and management of both single and multi-node datastores.

1. OpenStack

  Trove is an integrated OpenStack project, built on top of OpenStack
  services (Nova, Heat, Swift, Cinder, et al), and uses OpenStack
  methodologies and infrastructure for testing and development.

2. Simple

  Trove has push button, easy API calls to provision both single
  instance datastores and datastore clusters. Trove takes care
  of any heavy lifting required for provisoning the appropriate
  datastore type.

3. Intelligent

  Trove intelligently schedules where and how the datastores get
  provisioned, based on what hardware and capabilities are available
  underneath.

  Examples of this include:
    - If availability zones are available trove automatically
      schedules each node of a cluster in different AZs
    - If attached storage is not available trove uses local storage
      for storing datastore data.

4. Reliable

  Trove provisions datastores reliably, so that users of the API
  don't shoot themselves in the foot.

  Examples of this include:
    - Trove blocks restores to flavors with insufficient disk-space.
    - Trove understands cluster requirements, and prevents users from
      inadvertantly provisioning a nonsensical configuration (eg. an
      even number of nodes in a galera cluster).

5. Scalable

  Trove makes it easy to scale-up and scale-down provisioned
  datastores based on usage. Trove provides ways to resize compute,
  memory, and disk space for datastores as well as way to resize the
  number of nodes in a multi-node datastore cluster.

6. Monitoring of datastores

  Trove provides APIs that make it easy to monitor and reports
  state of datastores that it has provisioned. Trove also actively
  takes actions based on the outcome of such monitoring.

  Examples of this include:
    - Trove promotes a slave to a master if it detects a failed master
    - Trove provisions a new slave if it detects a failed slace
    - Trove builds a new node for a cluster if it detects a failed
      cluster node

7. Management of datastores

  Trove provides APIs that make it easy to perform routine and
  scheduled maintenance tasks on the datastore like taking backups,
  tuning datastore configuration parameters, etc. Trove also provides
  APIs that allow users to seamlessly migrate from one version of a
  datastore to another version of the same datastore.

8. Both single and multi-node

  Trove provisions both single-node instances, and multi-node
  clusters. Trove provisions both homogenous clusters (i.e. all nodes
  behave the same, eg. Cassandra Cluster), as well as instance sets
  with heterogenous relationships (i.e. All nodes are not the same,
  eg. Async replication with mysql master-slave instances).

9. Datastores

  Trove works with both traditional SQL databases (like mysql, and
  percona) as well as NoSQL datastores (like cassandra, and mongoDB).