There is a chasm between the public cloud operating model and how enterprises roll out their private cloud programs. Let’s review three crucial differences:
|Foundational cloud principle||Public Cloud Approach||Private Cloud Approach|
|Closed-loop automated operations||Public cloud vendors don’t roll out regions or services by hand. They invest heavily in internal IP to provision, monitor, troubleshoot and upgrade these services.||Enterprises run private clouds with a combination of people, open-source tools or integrating multiple vendor tools via ad-hoc scripting.|
|Cloud platform Engineering at scale||Public clouds engineer massively scalable shared multi-tenanted cloud control planes. This is treated as a product and IP development effort and is accomplished by hiring 100s of highly talented software engineers who have worked at VMware, Google, Facebook, et al.||Enterprises treat private clouds as a one-off IT project and cobble together teams from various groups and consultants. Few enterprises can afford 100s of platform engineers, or attract enough of them at the talent bar.|
|Operational reliability and high feature velocity||Public clouds get better every year in both operational reliability and new features. This can only happen beyond a tipping point of operational efficiency and scale.||Due to constraints, enterprise teams cannot evolve fast enough to keep pace with modern cloud-native stacks. They continue to build up technical debt making it challenging to take advantage of the latest capabilities that are coming to market, leading to rapid obsolescence.|
Operating costs and complexity continue to be top concerns about working with the private cloud. Enterprise customers want an automated, self-service infrastructure with the same level of agility, ease-of-use, and scale of the public cloud.
However, most enterprises need to make significant changes to their private cloud operating models and prepare themselves for the complexity and scale of the transformation
This just begs the question: What if private clouds were 10x easier to deploy, manage, and scale and required no specialized skills or know-how?
How to replicate the public cloud operating model in a private cloud?
It turns out that there is an architectural approach that will achieve that result. It’s possible to apply the public cloud operational model to your own cloud initiative to dramatically improve the odds of success.
Engineering a Cloud Control Plane
The heart of the solution lies in developing a scalable multi-tenanted cloud control plane. This is a centralized master arbiter to implement robust SLAs and automate cloud operational services including deployment, HA, monitoring, upgrades, alert management, backup/recovery to simplify operations and remove manual processes. Indeed, this will require significant engineering effort and software IP to realize, but the benefits of architecting this once will provide standardization across the enterprise and will avoid the cost and overhead of snowflake initiatives. The business outcomes will be dramatically better:
- Operationalize private clouds in days, not months
- Developers can deploy applications into a private cloud environment in 5 minutes using a self-service model
- Private clouds can be delivered with little added staffing. Furthermore, less time is spent on low-level infrastructure tasks and special skills
Enterprises need to create a “platform engineering” team and should treat this as a product development initiative.
Leveraging Open Source Frameworks
Under the hood, public cloud vendors leverage Open Source technologies and then wrap it around an operational service that customers can easily consume.
Open Source cloud projects like Kubernetes & OpenStack are a great way to build out private clouds. Arguably, open-source projects can innovate faster than public clouds leveraging the scale and skills of the larger communities and ecosystems. Combined with an operationally efficient cloud control plane, open-source innovation can provide the velocity to deploy the latest and emerging capabilities, thus avoiding rapid obsolescence.
Abstracting disparate infrastructure
IT teams in large enterprises deal with mind-boggling diversity and the complexity of their existing infrastructure footprint. They routinely have to manage multiple data centers in geographically distributed locations, heterogeneous server, storage, and networking SKU’s. New infrastructure and colocations continue to be added over time. Again, leveraging open-source operating systems such as Linux, hypervisors like KVM, broad integrations and plugins available for OpenStack and Kubernetes, and building in alerting and monitoring infrastructure, it is possible to provide an abstraction layer that can partially overcome this complexity.
Using these architectural patterns, you can roll out a private cloud and benefit from significantly lowered operations costs compared to public clouds, higher DevOps productivity, and more easily address privacy and regulatory compliance for your workloads that need to stay on-premises.
If your plans include private clouds, you could establish an engineering and operations team and then build out all the platform components and architect the cloud using the above principles yourself. This can take many months, even stretching into years for large enterprises, and could cost a fortune. If having a private cloud is strategic to you, but managing and scaling it is not, then you are better off considering a managed option.
Platform9’s SaaS platform has been built with these principles from the ground up and can significantly reduce your operational burden and get your private cloud up and running in a matter of days. You could get private clouds that are 10x easier to run and require no specialized skills or know-how.
Platform9 was founded with a mission to enable freedom in cloud computing. Platform9 enables five freedoms in deploying, managing, and scaling private cloud environments.