Maintenance in the modern cloud environment

Sergey Goncharov - August 09, 2022
Mirantis Container Cloud

We recently delivered Mirantis Container Cloud 2.19, which along with the previous 2.18 release, includes several features aimed at simplifying the life of administrators tasked with the ongoing maintenance of their infrastructure.  While maintenance is perhaps not the “coolest” part of a cloud native offering, it is extremely important to both Mirantis and our customers.   Thus, we thought it would be good to take a moment to reflect on how we think about system maintenance from the perspective of Mirantis Container Cloud.

Ever since ENIACs in the 1940’s through today’s most modern enterprise data centers, the concept of maintenance is an ever-present challenge for both business decision makers and operators to evaluate and tackle.  Some readers may remember those endless checklists, including some of my personal favorites like, security, networking, and checkdisk, that were compulsory to “tick away” before even considering the start of a maintenance window, which typically happened late at night in order to prevent negative impacts on the business. In most enterprises, teams that did maintenance typically consisted of multiple individuals from different teams and backgrounds with a deep and complex understanding of the steps needed to ensure that nothing was “left behind.”

All of that hard work was done for one purpose: making business applications function with the least possible disruption. 

Cloud native data centers have bridged the gap between operations teams, hardware and software. What mainly differentiates cloud native in the context of operations and maintenance is the “cloudification” of cluster-based applications and the proper “fluidization or dispersion” of underlying hardware-OS dependencies. 

But are we truly maintenance-free? I’m afraid there is no simple answer to this complex question. However, we can definitely claim that we are making major steps to take us there.

In certain contexts, we look upon the hyperscalers, such as AWS, Azure, and Google Cloud, as a solution to eliminate downtime, reduce maintenance, and abstract business critical applications from the underlying infrastructure. But the elephant in the room lies in data placement, data security and geo allocation. Governments across the globe are putting more and more pressure to regulate and protect citizen/customer data, leading many enterprises to adopt multi-cloud scenarios to achieve compliance.

So in a multi-cloud world, how do you deal with the issue of maintenance?  How do you plan and manage it?

With Mirantis Container Cloud, we strongly believe that maintenance should be done in a predictable and frictionless way independently of whether you choose an on-premises, hyperscaler or multi-cloud deployment architecture.      

Maintenance made easy

In Mirantis Container Cloud we have introduced a set of features related to the predictable and frictionless maintenance paradigm described above. We already provide the direct ability to put machines into maintenance mode.  However, we feel that maintenance is not only about planning either cluster or individual machine outages, and so we are looking into making our upgrade/update processes as seamless as possible by introducing a feature that we’ve named the “Upgrade sequence,” that allows upgrades for prioritized machines to go first. During a machine or machine pool creation, you can use the Container Cloud web UI Upgrade Index option to set a positive numeral value that defines the order of machine upgrades during a cluster update.

Another major topic besides host maintenance and control plane upgrades is dependency on the actual data storage technologies and registries.   In Mirantis Container Cloud 2.18, we introduced a feature that would enable our customers to attach custom Docker registries to child clusters. This eliminates unpredictable registry downtime or registry maintenance.

As for data storage, we are introducing new Ceph features, including a brand new user interface (UI) for health status, a detailed summary and placement options.  These enable our bare metal customers to have full control and visibility into their underlying storage solution for better planning.

Regardless of whether you choose on-premises, public cloud or multi-cloud deployment architecture, Mirantis Container Cloud presents you with a single pane of glass and a unified management layer to plan, execute and report on maintenance activities for both underlying individual hosts or upgrading your system to the latest version.

Technical innovations

In almost every release, we add technical innovations “under the hood” which might not make the top features list or get highlighted in release blogs like this one.  However, we think it is important that our customers understand the ongoing work and innovation that we produce to continually improve Mirantis Container Cloud.  Below are several examples of these enhancements which we’re implementing “behind the scenes” of Mirantis Container Cloud.

  • Several usability upgrades for Ceph, including CephFS and disk mapping. In a continuous evolution of the Ceph storage solution, we are adding several requested features that elevate the Day1 and Day2 experience for Operators. For a full list, please see the release notes below. 

  • Ubuntu 20.04 on VMware, which should help customers using the VMware virtualization platform to reduce dependencies on paid Linux distributions, while maintaining a high level of security, performance and innovation.

  • MITM proxy support for bare metal and OpenStack is a common request from our large enterprise customers in financial services, in order to achieve deep traffic scanning with custom trusted CA’s. This enables proper inspection of security-sensitive traffic and elevates customers’ security standards.

  • Enhanced user creation via API allows customers to broaden their automation and integration patterns for third-party appliances, to create and manage service user accounts and much more. 

  • IPSec traffic encryption for Kubernetes workloads, including embedded solutions like the Mirantis OpenStack control plane (Technical Preview). 

As always, for more details, please refer to the release notes:

We love to hear how our customers are taking advantage of the technology we deploy and what changes or enhancements you would like to see in future versions.  We welcome your input and feedback so that we can continue to provide you with the best possible products and services to allow you to meet your infrastructure goals.

If you’d like to test drive Mirantis Container Cloud, you can get started here.


Sergey Goncharov

Sergey Goncharov is Sr. Product Manager at Mirantis.