The Juno release of OpenStack made its debut on October 16, and as an active contributor to OpenStack, and to the Trove project in particular, I believe that this is a very significant milestone.
The success of a cloud platform fundamentally depends on the ability to easily deploy applications on that platform, and the Juno release builds on Icehouse in making OpenStack a much more compelling option for customers wishing to operate a private cloud within their own enterprise, or for service providers who wish to provide a public cloud service.
OpenStack’s stated mission is “to produce the ubiquitous Open Source Cloud Computing platform that will meet the needs of public and private clouds regardless of size, by being simple to implement and massively scalable.”
Since the launch, just over 4 years ago, OpenStack has grown from being a small open source project with a handful of contributors and participant companies to the fastest growing open source project in history, with major contributions from thousands of developers and well over a hundred companies in the past several releases. The accompanying chart illustrates this rapid growth. Today, beyond any reasonable doubt, OpenStack is the most widely used open source cloud computing platform available in the market.
OpenStack is now in use in production at a variety of high profile clouds in a wide variety of industries and verticals. Al Sadowski (the 451 Group) observed in May 2014 that 4,700 people attended the OpenStack Summit in Atlanta. He points out that the OpenStack community boasts members from 139 countries; only 89 countries were represented at the most recent Sochi Winter Games!
In May 2014, the Icehouse project delivered many improvements in the Compute and Storage projects (Nova, Swift and Cinder) that aimed at improving the ease of use and enterprise friendly operation of OpenStack including such things as rolling upgrades in Nova, improved replication in Swift and backend migration in Cinder. It saw considerable improvement to Neutron, the Networking service, and the introduction of Trove, the database as a service project.
In Icehouse, Trove provided support for multiple datastores including MySQL, Cassandra, Redis, and MongoDB. Users could provision a database instance with one of these databases, and Trove automated many capabilities like installation, configuration, and backup and restore. Furthermore, a user could launch a Trove instance based on a backup that had been taken earlier.
The Juno release further improved the core services including Ironic, the bare metal service and Docker support, through a StackForge project, additional storage backends for Cinder, and improved storage policies in Swift. Sahara, the Big Data processing service based on Hadoop and Spark makes its debut in Juno.
In Juno, Trove provided support for PostgreSQL and Couchbase, and added support for MySQL replication and MongoDB clustering. We (at Tesora) worked with the community to help define a framework for an implementation of both replication and clustering in a manner that would allow for extensibility to other datastores, and we contributed to the project all the code for MySQL replication in this release.
In our experience, enterprises operate complex data driven applications and invest a considerable amount of their time and resources on managing the database tier. This includes expense in such things as high-end hardware for their databases, and the many hours of DBA time spent in maintaining these databases. By some estimates, almost 80% of IT expenses are (in one way or the other) related to database related expenses!
OpenStack helps simplify the IT operations role by simplifying the activities of provisioning and deploying compute, storage and database capacity in the private cloud, thereby freeing up the administration resources to focus on more critical activities higher up in the stack. IT administrators can be confident that a newly provisioned compute instance from Nova will be exactly identical to an instance provisioned earlier, and a database administrator can be confident that a database instance provisioned with Trove will be exactly identical to one provisioned earlier.
OpenStack simplifies the workflow for requesting cloud services and the API’s allowing enterprises to easily develop in-house applications to improve the turnaround time in responding to requests. Integration with Ceilometer, the Telemetry project, and in-house applications that end-users have implemented help understand resource utilization and demand trends, and allow for accurate IT budgeting and forecasting.
With the Icehouse and Juno releases, OpenStack has clearly progressed from the “novelty” phase to the “worthy of serious consideration” phase for deployment in private cloud environments. At this point it contains all of the features and functions that would be required to operate it usefully in a private cloud environment and begin to use it for development and test use-cases in the vast majority of situations. While OpenStack is already widely used in a variety of production deployments, new capabilities that are planned for each of the significant projects in the upcoming releases will make OpenStack suitable for production deployment in the vast majority of situations.
Trove poses some unique challenges, not the least being the explosion in the number of combinations of operating systems, databases (both relational and non-relational), and OpenStack distributions. After all, in today’s data-driven world, an application can be only as successful as the data that it is able to manage and manipulate.
Amrith Kumar is a founder and chief technology officer of Tesora, a company focused on delivering a database as a service platform for OpenStack. By providing an enterprise distribution of Trove (in addition to our community edition), certified guest images for all supported databases, support and services, Tesora is working hard to help enterprises easily and cost-effectively migrate to OpenStack.
3 For some datastores, like MySQL.
4 Database Usage in the Public and Private Cloud; Tesora Infographic.