Bring In the “New” Infrastructure Stack

Jim Sangster - April 19, 2017
Today, Mirantis has announced Mirantis Cloud Platform 1.0, which heralds an operations-centric approach to open cloud.  But what does that mean in terms of cloud services today and into the future?  I think it may change your perspective when considering or deploying cloud infrastructure.

When our Co-Founder Boris Renski declared Infrastructure Software Is Dead, he was not talking about the validity or usefulness of infrastructure software, he was talking about the delivery and operations model for infrastructure software.  Historically, infrastructure software has been complicated as well as being notorious for challenging in terms of lifecycle management.  The typical model encompassed a very slow release model comprised of very large, integrated releases that arrived on the order of years for major releases (1.x, 2.x, 3.x...) and many quarters for minor releases (3.2, 3.3, 3.4…).  Moving from one to the other was an extremely taxing process for IT organizations, and combined with a typical hardware refresh cycle, this usually resulted in the mega-project mentality in our industry:

  1. Architect and deploy service(s) on a top-to-bottom stack

  2. Once it is working - don’t touch it (keep it running)

  3. Defer consumption of new features and innovation until next update

  4. Define a mega-project plan (typically along a 3 year HW refresh)

  5. Execute plan by starting at #1 again

While virtualization and cloud technologies provided a separation of hardware from applications, it didn’t necessarily solve this problem.  Even OpenStack by itself did not solve this problem.  As infrastructure software, it was still released and consumed in slow, integrated cycles.

Meanwhile, many interesting developments occurred in the application space.  Microservices, agile development methodologies, CI/CD, containers, DevOps — all focused on the ability to rapidly innovate and rapidly consume software in very small increments comprising a larger whole as opposed to one large integrated release.  This approach has been successful at the application level and has allowed an arms race to develop in the software economy: who can develop new services to drive revenue for their business faster than their competition?

Ironically, this movement has been happening with applications running on the older infrastructure methodology.  Why not leverage these innovations at the infrastructure level as well?

Enter Mirantis Cloud Platform (MCP)…

MCP was designed with the operations-centric approach in mind, to be able to consume and manage cloud infrastructure in the same way modern microservices are delivered at the application level.  The vision for MCP is that of a Continuously Delivered Cloud:

  1. With a single platform for virtual machines, containers and bare metal

  2. Delivered by a CI/CD pipeline

  3. With continuous monitoring

Our rich OpenStack platform has been extended with a full Kubernetes distribution which together enables the deployment and orchestration of VMs, containers and bare metal together, all on the same cloud.  As containers become increasingly important as a means of microservices development and deployment, they can be managed within the same open cloud infrastructure.

Mirantis Cloud Platform diagramMirantis will update MCP on a continuous basis with a lifecycle determined in weeks, not years.  This allows for the rapid release and consumption of updates to the infrastructure in small increments as opposed to the large integrated releases necessitating the mega-project.  Your consumption is based on DriveTrain, the lifecycle management tool connecting your cloud to the innovation coming from Mirantis.  With DriveTrain you consume the technology at your desired pace, pushed through a CI/CD pipeline and tested in staging, then promoted into production deployment.  In the future, this will include new features and full upgrades performed non-disruptively in an automated fashion.  You will be able to take advantage of the latest innovations quickly, as opposed to waiting for the next infrastructure mega-project.

Operations Support Systems have always been paramount to successful IT delivery, and even more so in a distributed system based on a continuous lifecycle paradigm. StackLight is the OSS that is purpose-built for MCP and provides continuous monitoring to enable automated alerts with a goal of SLA compliance.  This is the same OSS used when your cloud is managed by Mirantis with our Mirantis Managed OpenStack (MMO) offering where we can deliver up to 99.99% SLA guarantees, or if you are managing MCP in-house with your own IT operations.  As part of our Build-Operate-Transfer model, we focus on operational training with StackLight such that post-transfer you are able to use the same in-place StackLight and same in-place standard operating procedures.

Finally!  Infrastructure software that can be consumed and managed in a modern approach just like microservices are consumed and managed at the application level.  Long live the new infrastructure!

To learn more about MCP, please sign up for our webinar on April 26. See you there!