It was a big day for OpenStack as it marked the 10th full release in just 4 short years. Juno, the newly minted full release, was chalk full of exciting new features, updates, and bug fixes.
Here’s a little preview of what’s new, and I’m really excited to play around more with the “big data” options which are a part of the Sahara project, dubbed OpenStack Data Processing.
New Goodies in Juno
There really are too many updates to list here, and as you’ll read toward the end, I have posted a link to the entire release notes for a proper read. I just wanted to highlight some of the features that I have found to be interesting.
The word “enterprise” is showing up a lot along with OpenStack documentation and feature listings. This is a testament to the work being done by, and for large enterprise organizations who are looking to leverage the open cloud platform as a solution.
Lots of new features arrives for different hypervisors. Libvirt got a lot of love in this release which provides better support for general libvirt functions and performance increases, plus additional support improvements for LXC. Yes, there were containers before Docker 😉
Nova-network has new features also, which despite the predicted end to that portion of Nova, is well utilized in production environments, and continues to be a popular method for networking in OpenStack builds even today. For many organizations, the additional configuration and complexity with Neutron is not necessary according to their requirements.
The VMware driver received improvements which will help more folks make the jump to embracing OpenStack backed by VMware if they want to, without the need to go with the soon to be commercially released VIO (VMware Integrated OpenStack)
IPv6? Yes please!! It’s here now, with a fully supported IPv6 infrastructure. Many organizations found that the lack of IPv6 was an immediate limiter of the platform. Ask, and ye shall receive! Or in some cases, contribute to the core project and ye shall receive!
High Availability is the hot item in Neutron also with the addition of HA for the L3 agent. Yes, you heard that right, it’s a new feature to have HA. The Distributed Virtual Router is a new feature also and will help to cement Neutron as a powerful networking infrastructure which has caused some naysayers in the past to cast doubt on the viability of the scalability and resiliency of OpenStack networking. Some of that doubt was based in fact, but with Juno, those fears should be lifted.
Along with IPv6 and better HA, there were lots of plugin updates in this release:
That will bode well for the adoption of Neutron for sure. NFV gained a lot of focus also which you will hear more of in the video below.
Block Storage (Cinder)
Drivers galore is the theme of Cinder among much more that came in the Juno release. We have seen the addition of many new drivers including Fusion IO, Hitachi, Nimble, Pure, and many more. This is all good stuff, and you will be able to see if your current or potential storage provider has a driver, and their feature support here at the Cinder Support Matrix wiki page.
Cinder also got some new features such as Volume Replication, Volume Pools, Consistency Groups, and Snapshots for Consistency Groups. All of which have been hot items on the wish list for many administrators.
Image Service (Glance)
Image storage may not seem exciting, but if it didn’t work, people wouldn’t be happy. The addition of asynchronous processing is a nice feature, and other key work done with Juno includes the updated Metadata Definitions Catalog, and additional policy updates to make the service more aligned with traditional data center and cloud operations practices.
Lot’s of stuff. Really cool stuff. No, seriously, it’s all really cool. Each of the projects has updates for Horizon, including new projects, and there were more updates to the engines and frameworks driving the web UI. All in all, this was even more of a powerful update than the Icehouse release which saw major improvements.
Among the features, the one I am the most pleased about is the experimental Keystone-to-Keystone federation, or as many call it, the “it’s about time” feature. While extensibility within the environment has been good, imagine the situation where you have multiple OpenStack cloud environments to manage. It’s not that far out of reach since more organizations are make the move to the popular open platform.
Aside from that nifty feature, Keystone has been furnished many bug fixes and stability updates. Bad news for those who are on pre-Havana releases though, because one of the changes is that a rather non-clean deprecation of database mappings means that only Havana and Icehouse are directly upgradeable to Juno. Previous releases will need much more care and feeding.
Another fun deprecation, and by fun I mean entirely not fun, is the move from the nomenclature “tenant” to the alternative “project”. You may recall the early use of project which was then turned away in favor of tenant. As a result, there will be some confusion for a while but I’m sure that it will settle out over time.
Now in its second iteration as a core project, Trove saw many updates for additional DB frameworks, as well as feature updates for existing frameworks. Both on relational and NoSQL deployments, this was a rich update for the DBaaS platform which should bring more people to it, ultimately making the snowball of driving more features and stability updates as we go forward.
We do love our logs and performance metrics, so if you’re looking for the improvements here in the Ceilometer project, most are found on storage and handling of metrics and data. Much of the work was around partitioning data to enable a scalable platform to handle the growing amount of information that will be coming in as a result of increased adoption.
Tons of resource updates, and scalability improvements, and also recovery improvements in the event of failures during stack deployments. Heat is attracting more attention for sure, and the work that I’ve been seeing happen to tie Docker in with Heat is probably a big area where we will see a natural marriage of the two technologies.
Data Processing (Sahara) – ALL NEW!
Big Data here we come!! The updated project named Sahara has been. This warrants a post all to itself, but needless to say, as the words Big Data (yes, capitalized makes it more enterprise-friendly) dripping off of the tongues of CEOs and CIOs everywhere, we should see some strong innovation in this project.
If you want to get your Hadoop on, now it’s going to be just that much easier when you are running on top of your OpenStack cloud!
Object Storage (Swift)
You’ll find that Swift was looks a little different in the release cycle, and that is for a good reason. Swift is a relatively independent project and celebrates its updates and releases on a different schedule. Because Swift is a standalone product as well as a project, this can be confusing to some. Nonetheless, Swift did get a recent update to version 2.2.0 and you can read more about that here.
Updates to Swift included the new storage policies, updated Keystone support, and server-side account-to-account copy. Luckily the upgrade process for Swift is also one of most loved because it can be done in a zero-downtime manner by upgrading nodes and removing legacy nodes by simply distributing the data across the new nodes.
What is great as well is all of the documentation updates that have occurred as a part of this release. Not only are there updated install guides, but the Administrator Guide, Design Guide, Security Guide, and the now separately managed High Availability Guide are superb resources to get started on your OpenStack journey, or to learn more as an existing OpenStack user.
The full release notes are available here for your reading pleasure of course 🙂
If you’ve got 45 minutes to spare, and you want a great rundown of the features and some good conversation about how the new release was put together, this is a great presentation hosted by Mark Collier walking through OpenStack Juno. It’s a highly recommended listen if you ask me!
Congratulations to the OpenStack board, partners, contributors and all involved on another great milestone in the OpenStack platform!