Ubuntu 20.10 with Raspberry Pi Support

For any Linux admin who's been looking to deploy single board computers for various purposes, there's a new (while at the same time old) player in the Raspberry Pi mix – Ubuntu 20.10. Groovy Gorilla is the first official Ubuntu release to not only be optimized for the Raspberry Pi as a server distro, but as a full-blown desktop as well.

To make this even more appealing, Ubuntu 20.10 will include the likes of LXD 4.6 and MicroK8s for the easy deployment of resilient micro clouds, small clusters of servers providing virtual machines, and Kubernetes on demand at the edge.

Any Raspberry Pi 4 board with 4GB or 8GB of RAM can be deployed with Ubuntu Desktop or Server. And this isn't a stripped-down version of the platform; it's the full monty. Canonical has put in a ton of work to optimize Ubuntu for Raspberry Pi. According to the Ubuntu PR machine, "With this release, Ubuntu is optimized for Raspberry Pi, giving users of all levels and capabilities the access to Linux and microcloud technologies."

Mark Shuttleworth, CEO of Canonical, said of this iteration, "In this release, we celebrate the Raspberry Pi Foundation's commitment to put open computing in the hands of people all over the world." Shuttleworth continues, "We are honoured to support that initiative by optimising Ubuntu on the Raspberry Pi, whether for personal use, educational purposes or as a foundation for their next business venture."

Downloads will be made available starting October 22, 2020 from the official Ubuntu Raspberry Pi download page.

SaltStack Acquisition Brings More Automation to VMware

After reading about the VMware purchase of SaltStack, a thought occurred to me that this acquisition might have well been pursued for a single purpose – to compete with Kubernetes. Although VMware is the ruler of the virtual machine space, as is, they didn't have the technology to compete with the cluster management found in Kubernetes.

So we asked Purnima Padmanabhan, VP and GM, Cloud Automation, Cloud Management Business Unit, VMware, about this very thing. Let's take a look at some of her responses.

Padmanabhan said, "First off, instead of competing with Kubernetes, VMware is committed to helping customers build, run, and manage their software on Kubernetes. For years, VMware has contributed to the Kubernetes community and related open source projects."

Padmanabhan continued, "Our goal is to help our customers accelerate their journey to modern, containerized applications, and, with our Tanzu portfolio, we are well positioned to make that happen." She then gave us the key to unlock the door to this mystery when she said, "In addition, VMware's acquisition of SaltStack furthers the goal, by helping our customers automate their complete application to infrastructure stack." She added that, "Through automation, we help customers shrink the time it takes them to release new applications to the market; continuously expand their application scale, scope and business impact; and dynamically adapt their application and cloud stack to meet their changing needs."

So it looks as though VMware plans on leveraging the power of SaltStack to help empower automation and (probably) CI/CD.

New Storage Model Could Replace POSIX

POSIX has been around for a long time, but within the realm of enterprise computing, it has its issues which limit the scalability of compliant systems. This is especially true for systems that make use of deep learning, artificial intelligence, and other data-intensive use cases.

That's where object storage comes in. This particular model doesn't require a hierarchical data structure. Instead, object storage makes use of flat pool data (where each piece of data is defined by its associated metadata). This type of storage has no scalability limitations, which could be a boon for large-scale applications.

A group of engineers (including Henry Scott Newman, Chief Technology Officer at Seagate Government Solutions) created mmap_obj(), which makes it possible for object storage systems to process data within memory. The mmap_obj() protocol is a software abstraction for low-latency access to files. Although not yet in production, this breakthrough could mean accessing data in filesystem storage would no longer be a challenge for systems that must work with very large amounts of data.

The mmap_obj() model creates a new mapping in the virtual address space and would greatly benefit low-latency hardware storage hardware, such as NVMe over Fabrics (NVMeoF) and Storage Class Memory (SCM).

To read more about mmap_obj(), check out the paper written by John Michael Bent, Ujjwal Lanjewar, Nikita Danilov, Scott Hoot, and Henry Scott Newman, entitled "User-level low-latency access via memory semantics to objects in object storage" (https://priorart.ip.com/IPCOM/000263656).