Announcing Mirantis OpenStack 8.0 — the Most Stable OpenStack Distribution on the market

Interested in learning more about how Mirantis OpenStack 8.0 can help you stop worrying about your cloud and start innovating with it?

Join the What’s New in Mirantis 8.0 Webinar on Thursday, March 3.

Last September, when we announced Mirantis OpenStack 7.0, we called it the most stable OpenStack distribution yet — a claim we take seriously. We know how important it is for you to be able to meet, or even exceed, your internal customer SLAs, and we got tremendous validation of that stability from many of our largest customers. But of course, these are organizations who run thousands of nodes at a time, so perhaps it’s no surprise that they then asked for even more stability! So here we are with Mirantis OpenStack 8.0, which ousts 7.0 to take the crown as the most stable OpenStack distribution on the market.

Of course there’s more to Mirantis OpenStack 8.0 than just stability.  Among the many improvements over 7.0 are performance enhancements through bare metal deployment using Ironic, simplified web-scale operations, and more flexibility than any other commercial distribution available.

Stop worrying about the stability of your OpenStack private cloud

At the top of the list, however, is stability — which is no surprise, considering the size of some of our customers using Mirantis OpenStack in production, and the mission-critical nature of their deployments.  They can’t afford to have their clouds crash, and that means blunt feedback to us that in turn results in constantly improving testing and product quality.

To that end, with the 8.0 release we have introduced a brand new performance test suite. This test suite measures the performance of components such as networking, storage I/O and Ceph, and has helped fix a new class of bugs — for example, a race condition during create and delete operations of Cinder volumes.

This new test suite has also impacted our reference architecture. For even better performance and scalability, you can now deploy RabbitMQ on dedicated nodes. We also continue to increase the coverage of existing test suites. For example, we added new tests to try and break the Logrotate feature. Finally, we have increased the automation of our test suites to over 70%.

Netting it all out — we are the number #1 bug fixer in Liberty, with over 1,100 bugs fixed and an additional 300 or so bugs fixed in Mirantis OpenStack 8.0 (and destined for Mitaka), demonstrating our ongoing commitment to stability.

You can feel confident that you are offering your users a private cloud platform you can stand behind.

What have we done for your developers lately?

While stability will doubtless appeal to your operators, your developers will also find new features to love in Mirantis OpenStack 8.0, in the form of Ironic and new Murano and Kubernetes capabilities.

Ironic is OpenStack’s bare metal provisioning project, and enables your developers to provision a bare metal server with an API call or a click of a button, just like a VM. Up until the OpenStack Liberty release, we haven’t felt that Ironic was hardened enough to be put into production, but now it is, and we’ve integrated Fuel and Ironic so that Ironic can be deployed with the same simplicity as other OpenStack services.

Why is this interesting? Well, a lot of real-time or performance sensitive workloads, such as Cassandra, Hadoop, MySQL, deep machine learning or workloads that cannot be virtualized, need bare metal servers. Without Ironic, this has meant that even if you had an OpenStack private cloud, these particular workloads couldn’t be automated.  Now, with Ironic part of our 8.0 release, you can stop worrying about tickets, emails/phone calls/program meetings/spreadsheets… Developers can self-provision (or tear-down) their bare metal servers on-demand, through Horizon, just like a virtual machine.

For those who are more interested in scaling out than scaling up, Mirantis OpenStack 8.0 finally makes true hybrid cloud applications virtually effortless, adding the ability for the underlying IaaS to scale automatically in response to the load generated by the Kubernetes container orchestration platform through a new Kubernetes Murano package. What’s more, should local OpenStack resources run out, this package also enables workloads to burst to Google Cloud.

Finally, Cloud Foundry (CF) users get a tech preview of Murano services exposed into CF through Cloud Broker integration, which means that CF developers can use Murano services instead of the default CF marketplace, making applications a lot more portable to other PaaS systems.

Simplified web-scale operations

Mirantis OpenStack also enables you to spend less time fixing issues and more time on strategic topics with additional simplification and automation.

This release packs a number of features to simplify both “day-1” initial setup and “day-2” ongoing operations. MOS 8.0 now supports spreading OpenStack controllers across multiple racks using a layer 3 segmented underlay network, increasing both availability and scalability.

In addition, the Mirantis OpenStack deployment tool, Fuel, is now an official Big Tent OpenStack project, and the Fuel UI continues to evolve: a new equipment view provides you with an aggregate view of all your equipment. There’s also a tech preview of task based deployment, which parallelizes many tasks in Fuel and can speed up deployment by up to 2x. Finally, the settings tab is a lot better organized. Network setting are all consolidated, network groups can be added or deleted from the GUI, and IP ranges can be set for all network groups. Plugin settings are now reflected under this tab.

And while we’re on the subject of plugins, which provide so much flexibility to Fuel’s capabilities, Fuel now provides plugin endpoint info. For example, if you install Grafana, you can get the IP address for your Grafana dashboard from the Fuel UI. Also, due to popular demand, you can now see visual cues about which components have been tested together for compatibility. (These tags are also available through a new component registry.) Now, if a user chooses a combination of capabilities that aren’t known to work together, that’s flagged so the user can decide whether to go on.

For day-2 operations, the LMA (logging-monitoring-alerting) toolchain now has a new name – StackLight, a term borrowed from manufacturing, where physical lights offer visual and audible indicators of a machine state or process event. StackLight, available as a Fuel plugin later this month, will support clustering for HA and scalability, making it a true enterprise-class monitoring solution.

Fuel has also started a multi-release journey towards enabling configuration changes and the ability to add new plugins to an already deployed cloud. With 8.0, you will be able to do this in a limited manner — for example the Zabbix plugin can now be added to an already deployed cloud.  You can also change any option for keystone, nova, and neutron through the CLI.

Next, there is a new best practice guide to automate pushing of patches that eliminates the need to log into each node individually to apply patches.

Finally, a very important source of knowledge and expertise — the documentation — has some major updates. The Fuel documentation and the Mirantis OpenStack planning guide have been overhauled. There is a new Ceph operations guide as well. And with 8.0 you can further tap into our expertise through a brand new knowledge base.

More infrastructure choices: Pure-play vs. Co-engineered

As always, the latest release of Mirantis OpenStack is true to our “pure play” philosophy, which means that at the heart of our distribution we have a highly available infrastructure controller certified to orchestrate multiple compute, storage and networking platforms.

Of course, when you consider that OpenStack itself is intended to orchestrate multiple platforms, that’s not surprising — in fact, it’s by adhering to that OpenStack philosophy that we come to this point.

Still, it’s important to understand the distinction, because some competing distributions advocate co-engineering instead, prescribing the hypervisor and host operating system choices along with their OpenStack distribution; and often requiring other components, such as specific software-defined networking (SDN), storage, hardware, cloud management platform (CMP), platform-as-a-service (PaaS) solutions as well.

At Mirantis we believe the locus of the cloud is the cloud controller. This is where we invest our efforts to make sure OpenStack works well and interoperates with third party components. Co-engineering generally advocates that the anchor point be something else, like a Linux distro (in case of Red Hat and Red Hat Enterprise Linux), or hypervisor (in case of VMware and ESX) or the network (in case of Cisco). One road gives you infrastructure flexibility, while the other may lock you in and prevent you from meeting your customer requirements.

At the end of the day, Mirantis OpenStack 8.0 is the only distribution that orchestrates KVM on Ubuntu, RHEL7 KVM, XenServer and VMware vCenter! To see what other options are available, check out our complete list of technology partners.

And speaking of flexibility, if you are a cloud architect, you will love our new feature that enables you to dynamically build a Ubuntu bootstrap image with the drivers and patches of your choice, removing a major roadblock in discovering and supporting the latest hardware for Ubuntu KVM compute nodes. You should also know that the pool of Fuel deployment plugins continues to grow, with the current count being over 170. We continue to expand our Mirantis Unlocked partners program to officially certify and support plugins for our customers, and you can find these certified customer-ready plugins in our Fuel plugin catalog.

Your cloud needs are bound to evolve over time, or with one-time events such as mergers and acquisitions. More infrastructure choices means more flexibility in adapting your cloud to new requirements.

Consume the latest open-source innovation

Finally, everything we do is 100% open-source and upstream-first. In addition to fixing the greatest number of bugs, we were the #2 contributor to Liberty by reviews and #3 by commits.

Mirantis OpenStack 8.0 also includes Ceph Hammer, which provides performance improvements when accessing buckets with a large number of objects (as in millions) and in all-SSD clusters. The object interface also supports more of the Swift API (such as object versioning). We were the #3 contributor to Ceph Hammer by lines of code.

Where to get it

Meanwhile, download Mirantis OpenStack 8.0 and take it for a spin.  See how you can consume all of this cutting edge innovation without spinning up an engineering team of your own.

Learn more through additional videos:

For more details on what’s new and different, you can read the release notes or check out the what’s new page.

Latest Tweets

Mirantis and Ericsson on Edge Computing