OpenStack Kilo ships with identity federation, storage improvements, bare-metal service

OpenStack Kilo is out, but some challenges persist

OpenStack Kilo is out, but some organisations may think twice about deployment

The OpenStack community released the eleventh version of the open source platform this week, codenamed Kilo, which ships with loads of improvements including new management APIs, security improvements for NFV, and the first full release of the bare metal cloud service. But a number of challenges still conspire to make the platform difficult to implement for some organisations.

The organisation has added improvements across the board, including:

  • Nova Compute: New API versioning management with v2.1 and microversions, which makes it easier to write long-lived applications against compute functionality. Operational improvements include live upgrades when a database schema change is required and better support for changing the resources of a running VM.
  • Swift Object Storage: Erasure coding provides efficient and cost-effective storage, and container-level temporary URLs allow time-limited access to a set of objects in a container. The latest release also brings improvements to global cluster replication and storage policy metrics.
  • Cinder Block Storage: Major updates to testing and validation requirements for backend storage systems across 70 options. Users can now attach a volume to multiple compute instances for high-availability and migration use cases.
  • Neutron Networking: The load-balancing-as-a-service API is now in its second version. The community also added additional features support NFV like port security for OpenVSwitch, VLAN transparency and MTU API extensions.
  • Ironic Bare-Metal Provisioning: The first full release of the Ironic bare-metal provisioning project with support for existing VM workloads and Linux containers, platform-as-a-service and NFV.
  • Keystone Identity Service: Identity federation enhancements to support hybrid workloads in multi-cloud environments.

This is the latest version of OpenStack since Juno, which was first released in October last year. Liberty, the version currently under development, is due to be released in October this year.

The OpenStack foundation said deployments of OpenStack are growing, with production deployments accounting for about half of those. But the project still needs to improve in some areas – which reflects the maturity of the platform more than anything.

“One thing I would definitely call out when considering OpenStack versus something like AWS is that with the former we have the ability to take advantage of innovation more rapidly than would be the case with the latter,” Michael Yoon, chief technology officer of MiMedia, an OpenStack user, told BCN. “So specifically, things like erasure coding, SMR technology, kinetic drive and object storage at the drive level, these are all making a very serious impact when comparing one solution from another.”

“But the container strategy hasn’t really been there in Swift. Also upgradability is certainly a challenge – one that’s been improved upon since the early releases, but still needs a lot of work.”

One of the biggest challenges, Yoon explained, is that there aren’t enough vendor-agnostic best-practice and architectural guides out there today. So the overhead, in terms of the initial research required to stand up an OpenStack cloud in the first place, is high.

“There’s a decent amount of research you need to do if you’re looking to get into this, and the problem is every vendor has their version, stocked with their own IP, of what makes a performant OpenStack distribution; each has a set of best practices and there’s still a fair amount of having to wade through it all to make things less vendor specific,” Yoon added.

This challenge seems to be highlighted in the latest cloud research from 451Research. According to the analyst house’s latest Cloud Price Index (CPI) report the TCO of proprietary commercial cloud management offerings is less than that of OpenStack distributions because of the cost of hiring additional manpower to implement them.

“The proprietary offering’s TCO benefit is simply the result of the high cost of OpenStack engineers – the distributions themselves are priced lower than the proprietary offerings,” the report explained. “With OpenStack, migration should, on paper, be less expensive, but it will be made more difficult than necessary due to a lack of federation among providers and the numerous OpenStack reference architectures.”

Nevertheless, the newly announced features may help improve the attractiveness of the platform among organisations.

“OpenStack continues to grow, and features like federated identity and bare metal provisioning support make the platform more compelling for enterprise IT leadership and application developers who want a stable, open source alternative to proprietary options,” said Al Sadowski, research director at 451 Research and one of the lead authors on the CPI report.