The HPE Container Platform is a key component of HPE’s hybrid cloud solutions – and the company’s overall as-a-service strategy – to help customers accelerate their digital transformation initiatives with edge-centric, cloud-enabled, and data-driven applications.
Containers and Kubernetes are Game-Changers
Containers have transformed how modern applications are developed, packaged, and deployed. Benefits include speed, efficiency, consistency, and perhaps most importantly, the portability to write once and run anywhere. Both the BlueData and MapR teams have seen this transformation evolve in the artificial intelligence (AI) and analytics space. Containers are now pervasive and represent the standard deployment model for machine learning, deep learning, and advanced analytics. As one analyst recently told me, it’s no longer a question of “why” deploy on containers – rather, it’s a question of “why not” deploy on containers.
Over the last few years, the Kubernetes open source project has quickly become the de facto orchestrator of choice for managing containerized application workloads and services. While it has gained broad-based support across a range of applications, the primary use cases are in the areas of cloud-native microservices applications and CI/CD pipelines. The industry has also seen growing adoption of Kubernetes for AI and analytics applications – and again, both the BlueData and MapR teams are actively involved in the intersection of these technologies.
Containers and Kubernetes are game-changers. The pace of innovation in the Kubernetes ecosystem is astounding. However, it is important to note that Kubernetes is not a product, it is an open source project. The design goal of Kubernetes is to promote pluggability and extensibility from the applications/services layers to the infrastructure layers such as compute, storage, and networking.
Guiding Principles for the HPE Container Platform
HPE believes in the power of Kubernetes and the critical role it will play in the company’s quest to help customers on their hybrid cloud journeys and digital transformation initiatives. Armed with extensive input from enterprise customers, as well as from industry analysts and partners, HPE has converged on a few key guiding principles that form the basis of the HPE Container Platform.
- Deliver business outcomes:
Developing and deploying net new applications is insufficient. Organizations want a platform that can support the whole lifecycle of applications – especially for data-driven applications. Applications produce data and analytics provide insights that are injected back into the application. This data is produced and stored in multiple places – in data centers, in the cloud, and increasingly, at the edge.
Enterprises need a comprehensive container platform to support a diverse range of application workloads and use cases (cloud-native, app modernization, AI/ML, and analytics) and user personas (app developers, DevOps, data scientists) across any location with secure multi-tenancy, lifecycle management, and support for day 2 operations.
- No app left behind:
The majority of enterprise applications are monolithic. Cloud-native is a key focus for net new application development, and enterprises are re-architecting a subset of their existing applications to be cloud-native. But that still only represents a relatively small portion of enterprise applications in a typical large organization.
Enterprises need a solution that can provide the benefits of containerization to traditional, non-cloud-native monolithic applications, in addition to cloud-native microservices applications. To achieve this goal, a solution must address the enterprise-grade security, networking, and storage needs of these monolithic applications – with a focus on persistent data storage.
- Simplicity and speed:
The benchmark for this vector is the public cloud and the services offered by the public cloud providers. In the public cloud, developers and administrators can easily spin up one or more cloud-based Kubernetes clusters with fully integrated storage and networking; they can also deploy their tools and apps with just a few clicks in a matter of minutes.
Enterprises need a turnkey platform running either on-premises or in any public cloud that can deploy one or more Kubernetes clusters with pre-integrated storage and networking – along with one-click deployment for pre-defined tools and apps. Ultimately, the goal is to accelerate application development and help developers release more code faster no matter where the cluster is located.
- Bare-metal containers:
In the AI and analytics space, the standard for containerization is bare-metal. Machine learning, deep learning, and advanced analytics workloads are deployed on containers running on bare-metal servers – to ensure optimal performance and efficiency. But this isn’t the case for other use cases and workloads; the majority of containers are run on virtual machines, either on-premises or cloud instances.
Enterprises need the option to run Kubernetes clusters on premises on bare-metal servers, for all of their application workloads. This approach delivers significant benefits not only in terms of performance but also in better utilization and cost savings since it reduces the number of virtualization layers while increasing container density.
With the new HPE Container Platform, HPE’s objective is to address each of these requirements.
Innovations of the HPE Container Platform
The HPE Container Platform is a turnkey Kubernetes-based software solution for deploying and managing containerized applications on any on-premises infrastructure (either bare-metal servers or virtual machines), on any public cloud, and at the edge. It addresses the aforementioned enterprise requirements through the following capabilities and innovations.
- Multi-tenant multi-cluster management:
This component is based on the BlueData platform, and it provides a control plane that allows administrators to deploy and manage the lifecycle of one or more managed Kubernetes clusters with secure multi-tenancy. Available as a web-based UI, REST API, or CLI, it is optimized for bare-metal deployments to achieve cloud-like simplicity, speed, and flexibility – as well as greater efficiency and lower costs.
Different development groups can access one or more Kubernetes clusters across different versions. This model enables enterprises to separate different use cases such as CI/CD, databases, and analytics running on their own clusters. Each may have its own operational lifecycles and SLAs that can be challenging to achieve on a single Kubernetes cluster.
- Integrated data fabric and persistent storage:
Unique to the HPE Container Platform, this pre-integrated component powered by the MapR filesystem can store and manage data ingested from applications and/or external sources, as well as the default container persistent storage for the Kubernetes clusters via the MapR CSI.
This scale-out data fabric allows the buildout of rich, closed-loop solutions informed by data and analytics, including edge computing and IoT use cases. In addition, this proven filesystem provides unique capabilities for stateful applications, in particular, for non-cloud-native monolithic apps that require fault-tolerant and distributed persistent storage.
The HPE Container Platform also provides connectivity to external storage and enables the use of other storage products (HPE Cloud Volumes with HPE Nimble Storage or other software-defined storage) that offer a container storage interface (CSI).
- Open source Kubernetes and KubeDirector:
The Kubernetes clusters managed by the HPE Container Platform are 100 percent open source versions of Kubernetes. This feature enables customers to be future-proofed, with the ability to deploy a wide range of Kubernetes ecosystem products and tools (e.g. application monitoring).
The HPE Container Platform control plane auto-deploys KubeDirector, an open source custom controller developed by the HPE BlueData team, on every Kubernetes cluster. KubeDirector brings a consistent application onboarding process for non-cloud-native apps and delivers cloud-native agility and efficiency without re-architecting them.
Popular Kubernetes packaging and deployment mechanisms such as Helm, Operators, and YAML files are a perfect fit for deploying stateless and cloud-native applications. With KubeDirector, developers have a flexible and simple controller framework to onboard non-cloud-native monolithic apps (e.g. complex, stateful applications).
- App store with on-demand deployment:
The HPE Container Platform provides a library of pre-built Kubernetes application artifacts such as KubeDirector, YAML, and Helm Charts to accelerate deployments. Examples include data analytics frameworks (such as Apache Spark via KubeDirector), CI/CD tools (such as Jenkins via a Helm chart), and many more.
As a result, these innovations extend the ability of the HPE Container Platform to address large-scale enterprise Kubernetes deployments across a wide range of use cases – from machine learning and edge analytics to CI/CD pipelines, application modernization, and much more.
If you’re at KubeCon I CloudNativeCon this week, stop by the HPE booth (P28) to meet the experts and talk to our team. You can also visit the HPE Container Platform web page, and watch this brief demo video to see how it works:
About the author:
GuestBloggerCDI
HPE Software-Defined and Cloud Group
Hewlett Packard Enterprise
twitter.com/HPE_HybridCloud
linkedin.com/showcase/hpe-cloud
hpe.com/composable
