Given the current widely held view that a hybrid cloud approach is the right way forward for many larger enterprises, much of the private segment focuses on OpenStack as the cloud-stack platform.
OpenStack is an open source cloud platform that consists of 30-plus projects, each of which delivers a specific service to the overall stack. This approach allows services to evolve independently and enables additional service projects to spawn easily.
In many ways, OpenStack is a Hobson's choice: the only show in town. OpenStack benefits include that it is well-supported, with numerous companies that add value, mainly in deployment and management tools. Many large enterprises have OpenStack in prototype, and it is in production with both enterprises and academia.
Though many experts predicted OpenStack use would soar, that hasn’t happened. One challenge has been to find admins with the skills to deploy a complex environment, which has proven tricky to tune and to stabilize even in the sandbox. Predictably, companies, such as Red Hat and Dell, have risen to the challenge with deployment tools and pre-built configurations. Still, with 30 modules, it’s hard work to make the right decisions.
The range in development of various OpenStack projects has also dulled OpenStack's use in production. OpenStack started off small, with just a few modules, but it has grown considerably. These newer modules range in maturity. Users perceive this variable spectrum as instability, with repetitive work and occasionally the need for a forklift update with the introduction of major new features.
In many ways, OpenStack is similar to a Linux project -- lots of command-line interface and scripts, heavy manual interaction to configure and manage the cluster, and little automation. That's fine for a simple environment, but more sophistication is likely required for OpenStack given the demands of multiple VMs and the agility required to manage them. The platform needs automated management, which is finally beginning to reach the market.
As with much of IT, the initial focus in OpenStack -- beyond the creation of basic storage and networking services -- was on app management and server orchestration. The result is that OpenStack requires some catch-up on the software-defined networking front. Neutron, the networking project, has had challenges with scaling deployments as it continues to evolve.
Storage is a different matter. Ceph, another open source object storage stack, has effectively torpedoed OpenStack's own object storage project, Swift. Ceph is well-accepted and, while it continues to evolve and improve, has recognized the need for interface standards compatibility with public clouds. In an effort to distinguish itself from Ceph, Swift veers away from industry practices and APIs to the extent that it is likely to become obsolete.
Why would you use OpenStack?
There are five reasons why a business would choose OpenStack.
The first and most important is that there is no real alternative; no other low-cost, open source stack exists. In fact, there is no other comprehensive stack for private clouds at all. There are open source alternatives on the market in Eucalyptus, CloudStack, Nebula and others, but OpenStack blows them away for scope. None of these has the broad vendor support -- from the likes of Dell, Hewlett Packard Enterprise, IBM, Red Hat and Mirantis to a host of smaller companies -- that there is with OpenStack.
Second on the list of OpenStack benefits, the key modules get the attention needed to correct past errors and achieve production-level maturity. The base modules are all production-ready today. OpenStack.org recognizes the need to improve and publishes a project lexicon with maturity levels shown to set expectations for different elements. It presents working sets of modules to ease startups for beginners.
The third reason is the popularity of OpenStack, both in the prototype stage and mainstream production. As a result, there is a wealth of information and experience that can be found online.
The fourth of the OpenStack benefits is an increase in the market in toolkits -- from monitors to dashboards -- that automate and ease work. It's a bit chaotic, but it's also an indication that some first-class tools will surface within the year.
Finally, major public cloud vendors are willing to partner with OpenStack, which aims to unify scripts and ease cross-cloud operations. There is still work to do, but common solutions for the major clouds should fall into place, allowing for a single control structure.
The containers approach is disruptive. Containers bring the promise of a dramatic uplift in effective performance, which makes them one of the hottest items in IT. However, reality has lagged behind the hype, and there are plenty of questions around containers. Does OpenStack sit on top of Kubernetes? Or does Kubernetes sit on top of OpenStack?
Containers are so new that it would take several months to sort out several fundamental issues. However, there are enough interested participants that we will get an answer.
Public clouds -- the real competition for OpenStack
In the next few years, expect the battle for IT to focus in on public versus hybrid clouds. The public clouds have so many inherent advantages in connectivity, scale, cost models, data services and pace of evolution that they are already formidable adversaries. That begs the question of whether private in-house IT has a long-term future.
To make the issue more complex, some public cloud service providers (CSPs) now look to drop their own stacks into the private cloud market, similar to Azure's expansion with Azure Stack. This would give CSPs two key advantages, both derived from lock-in.
First, they would define operating processes for the private cloud in their own terms, which greatly eases a transition from private to hybrid cloud operations. Second, CSPs would create a path for the private cloud to be assimilated into their public space. Both of these lock-ins are likely to be costly for users.
A better alternative would exist if cross-cloud operation was streamlined, with scripting structures supported by all the cloud stacks from OpenStack through Azure to AWS. A move to cloud-agnostic management tools might achieve this.
OpenStack has warts, but it looks to be the best way to avoid vendor lock-in and work in a dynamically evolving environment. OpenStack benefits include its price (as open source), which is attractive, and that it will soon have graduated from early adoption into mainstream production. These are turbulent times in IT, and it's important to take some risks to gain the huge advantages of business innovation. The one option that doesn't exist is to wait and see.
What does the future hold for OpenStack?
Build a hybrid cloud with OpenStack
Choose between OpenStack and VMware for private cloud