PRODUCT REVIEW: HPE Container Platform
June 17, 2020
AI app development has grown up in a world of open source software. There are HUNDREDS of software options across all the phases of a pipeline, from collecting data to making decisions based on the analysis. Essentially, too many options for any person or team to master all of them—AND they are all different.
In my previous post, “Containers Driving Transformation”, I noted that Robert Christianson, HPE’s Chief Cloud Strategist, articulated three benefits to containers. He said that they are a more efficient development environment compared to virtual machines and are appropriate for a multi-cloud and/or hybrid IT environment.
451 Research has recently published a survey and analysis that lend Christianson’s position some statistical support. Between full adoption today and piloting within the next 24 months, 94% of the 464 enterprises in the survey have or will adopt containers, and 80% of those enterprises will use Kubernetes for orchestration.
HPE released its “HPE Container Platform” in March of this year, which integrates BlueData and MapR into its “HPE Container Platform” (HCP). HCP provides a structured, secure, compliant, standardized framework for any company to properly orchestrate, manage, and govern its Linux and container-based AI application development and execution infrastructures. The line-up provides a range of benefits, among them:
- single view of and access to the software stack—all data, storage, and compute
- single security model with a single point of control
- the ability to set and verify compliance with standards, and
- flexibility to customize the infrastructure (compute, storage, and networking) to the differing needs of different applications.
These benefits work to shorten time to market, reduce costs, and increase competitive positioning.
But from a technical perspective, MapR brings two important features to the integrated solution: (1) global namespace (with all that this implies, including stateful data), and (2) connection to existing storage (or data repositories or lakes) via a “data tap”.
By connecting to existing data repositories, AI developers can read data that already exists, avoiding the expense and complexity of copying that existing data from original (non-AI) source to a new (AI storage) location. This is especially helpful during development, when businesses typically are trying to prove functionality at a low initial cost.
After data scientists prove the value of the data, setting up a production environment with a global namespace is a significant positive. It allows applications to read data no matter where it originates or resides, whether stored in multiple locations around the world or in the primary production application environment.
The addition of BlueData software provides management capabilities to the HPC. Enjoying a single point of control, virtualized resources that can be used with high utilization (offering the same benefits that VMWare offered for P to V), reduced cost and increased velocity, setting standards, monitoring compliance, and an overall reduction in time to market, are all attributes that make the business case for the HCP very strong.
By: John Duffy, Chief Technologist and Strategist