When talking about the benefits of OpenStack, it’s easy to assume there’s an already-built cloud and so focus on how applications are deployed. But what about how you actually build an OpenStack-based cloud at day zero? Meaning, getting OpenStack up and ready to configure for service delivery, when you’re starting with bare metal in a rack? It turns out that this is a non-trivial exercise. A combination of blueprints and orchestration can help make day zero OpenStack provisioning more agile.
Why Worry About Day Zero OpenStack Provisioning?
Why does day zero provisioning matter to IT? After all, it’s after the hardware is configured, virtualized and the OpenStack distribution is installed that the really interesting “as a service” stuff, happens, right? That’s true, but we’re still at the nascent stages of private cloud, and there is a race on against public cloud.
Guest article by Alex Henthorn-Iwane, QualiSystems
Today, the main reason why more applications and data aren’t put into the public cloud is security concerns—but in many other ways public clouds are beating the pants off private clouds because they’re just so easy. It takes minutes and a credit card to add public cloud infrastructure capacity. It can take weeks to months to purchase, receive, provision and turn infrastructure into a private IaaS cloud. That’s not terribly competitive.
It’s estimated that 25% of IT spending is now happening in the shadows, outside of IT control. A discussion we had with the CIO of a SMB banking firm was revealing. He said he’d let his open source private cloud team attempt to build what he called a “dirt road through the jungle”, but that he really intended to just shift everything to the “eight-lane highway” AWS cloud as soon as practical.
OpenStack’s success is to a large degree predicated on the success of private clouds. If private clouds don’t increase in competitiveness versus public clouds, then at some point security issues get easier and, based on current market-share trends, AWS, Azure and possibly VMWare-based public clouds end up ruling the day. Okay, that’s a bit doomsday in tone, but the faster that OpenStack private clouds can be brought up, the better.
Day Zero Infrastructure Provisioning Challenges
Day zero provisioning today is fairly slow for pretty much any form of datacenter infrastructure. The reason is that there are many design decisions, and thus provisioning dependencies to map across compute and storage configurations, networking topology and connectivity, etc.
Of course, one could argue that traditional IT compute, storage and networking products are a significant issue, since they were designed before the era of “software-defined everything” that is characterized by API-enablement and ease of programmability. Nonetheless, even if you accept that you can use white box, software-defined hardware in every category, all those design and provisioning dependencies require expertise, and the down and dirty provisioning steps must be carefully performed in the right order to not mess things up.
Most day zero provisioning is still done manually and it is typically at least a week-long affair, if not longer. With OpenStack the complexity can be accentuated by the fact that there is such dynamism in the standards and in terms of more programmable, software-defined infrastructure solutions that are in the big picture and still emerging from a market share point of view.
Converged Infrastructure, Reference Architectures, and Blueprints…Oh My!
One answer the industry has come up with to simplify the initial integration of all the infrastructure components to deliver private clouds is the concept of converged infrastructure. There are a few different categories underneath this idea:
- Converged infrastructure typically means prescriptively designed sets of hardware and software, provisioned, QA’d and certified by a single vendor, such as VCE
- Reference architecture typically means a looser matrix of compatible vendor/product combinations
- Hyper-converged appliances boil the prescriptive model down to an appliance level. These appliances are designed to scale out private cloud infrastructure simply by adding more appliances.
Today, most of the above categories of products and architectures are built using VMWare, but the principle is still important. Create a blueprint of actual hardware products that are tightly specified, including firmware revision levels, and have someone stand behind that blueprint as being able to deliver interoperable function and performance.
The Role of Day Zero Provisioning Orchestration
The only issue with all those blueprints is that it doesn’t change the intricacy and complexity of the provisioning steps themselves. Of course, scope matters. If you’re provisioning at the level of a hyper-converged appliance, it’s easier than if you’re provisioning at the rack or pod level. This is where day zero orchestration comes into play. I choose the word orchestration in particular because the intricacy of day zero provisioning requires automation that can coordinate and control many different task-level automation processes to take a rack of bare metal devices from PXE boot to an OpenStack private cloud deployment that is functionally ready for service configuration and delivery (Day One and beyond).
Some helpful ingredients for effective OpenStack day zero orchestration include:
- Ability to abstract bare metal, virtualization, software, and logical concepts such as network addresses, licenses, etc.
- Modeling of arbitrary topologies—networking becomes trickier and more important as you scale up the size of the infrastructure
- Orchestration authoring that is ideally visual so non-programmer IT personnel can contribute meaningfully to cloud automation practice
- Object-based task automation that is sustainable and maintainable
- Open southbound interfacing, so that there is flexibility of what type or mix of languages is used
When Day Zero provisioning orchestration is employed, provisioning times can easily be reduced from days to hours, even without implementing optimizations such as parallelism in orchestration workflows. In addition, orchestration ensures greater accuracy by eliminating simple human errors during the provisioning process.
Day Zero OpenStack orchestration is an important enabler for the success of OpenStack private clouds. It brings greater agility, speed and accuracy to the initial delivery of infrastructure. Manufacturers, distributors, systems integrators and IT organizations can leverage Day Zero orchestration to make their end-to-end OpenStack private cloud more competitive as part of an IT as a Service strategy.
Read the Aberdeen report Effective Private Clouds Phase in Richest Features First
Alex Henthorn-Iwane joined QualiSystems in February 2013 and is responsible for worldwide marketing and public relations. Prior to joining QualiSystems, Alex was vice president of marketing and product management at Packet Design, Inc., a provider of network management software, and has 20+ years of experience in senior management, marketing, and technical roles at networking and security startups. Through his roles at QualiSystems, Packet Design, CoSine Communications, Corona Networks and Lucent Technologies he has acquired expertise in cloud computing, software defined networking and network function virtualization, DevOps, ITaaS, and IT automation and orchestration. He has written for Embedded Computing, Virtual Strategy Magazine, Datamation, SDN Central, Datacenter Knowledge and InformationWeek. These days, Alex focuses a lot of his writing around the intersection of new, programmable infrastructure technologies (cloud, SDx, NFV), DevOps and the orchestration and automation enablement needed to make all of them a reality, in both the enterprise and service provider/carrier space. Connect with him on LinkedIn, Twitter, or SlideShare.