SBC Clusters — Beyond Raspberry Pi

By

-

July 12, 2018

Cluster computers constructed of Raspberry Pi SBCs have been around for years, ranging from supercomputer-like behemoths to simple hobbyist rigs. More recently, we’ve seen cluster designs that use other open-spec hacker boards, many of which offer higher computer power and faster networking at the same or lower price. Farther below, we’ll examine one recent open source design from Paul Smith at Climbers.net that combines 12 octa-core NanoPi-Fire3 SBCs for a 96-core cluster.

SBC-based clusters primarily fill the needs of computer researchers who find it too expensive to book time on a server-based HPC (high performance computing) cluster. Large-scale HPC clusters are in such high demand, that it’s hard to find available cluster time in the first place.

Research centers and universities around the world have developed RPi-based cluster computing for research into parallel computing, deep learning, medical research, weather simulations, cryptocurrency mining, software-defined networks, distributed storage, and more. Clusters have been deployed to provide a high degree of redundancy or to simulate massive IoT networks, such as with Resin.io’s 144-pi Beast v2.

Even the largest of these clusters comes nowhere close to the performance of server-based HPC clusters. Yet, in many research scenarios, top performance is not essential. It’s the combination of separate cores running in parallel that make the difference. The Raspberry Pi based systems typically use the MPI (Messaging Passing Interface) library for exchanging messages between computers to deploy a parallel program across distributed memory.

BitScope, which is a leader in Pi cluster hardware such as its Bitscope Blade, has developed a system with Los Alamos National Laboratory based on its larger BitScope Cluster Module. The Los Alamos hosted system comprises five racks of 150 Raspberry Pi 3 SBCs. Multiply those 750 boards by the four Cortex-A53 cores on each Pi and you get a 3,000-core parallelized supercomputer.

The Los Alamos system is said to be far more affordable and power efficient than building a dedicated testbed of the same size using conventional technology, which would cost a quarter billion dollars and use 25 megawatts of electricity. There are now plans to move to a 4,000-core Pi cluster.

Most clusters are much smaller 5-25 board rigs, and are typically deployed by educators, hobbyists, embedded engineers, and even artists and musicians. These range from open source DIY designs to commercial hardware rack systems designed to power and cool multiple densely packed compute boards.

96-core NanoPi Fire3 cluster shows impressive benchmarks

The 96-core cluster computer recently detailed on Climbers.net is the largest of several cluster designs developed by Nick Smith. These include a 40-core system based on the octa-core NanoPC-T3, and others that use the Pine A64+, the Orange Pi Plus 2E, and various Raspberry Pi models. (All these SBCs can be found in our recently concluded hacker board reader survey.)

The new cluster, which was spotted by Worksonarm and further described on CNXSoft, uses FriendlyElec’s open-spec NanoPi Fire3.

The open source cluster design includes Inkscape code for laser cutter construction. Smith made numerous changes to his earlier clusters intended to increase heat dissipation, improve durability, and reduce space, cost, and power consumption. These include offering two 7W case fans instead of one and moving to a GbE switch. The Bill of Materials ran to just over £543 ($717), with the NanoPi Fire3 boards totaling £383, including shipping. The next biggest shopping item was £62 for microSD cards.

The $35 Fire3 SBC, which measures only 75x40mm, houses a powerful Samsung S5P6818. The SoC features 8x Cortex-A53 cores at up to 1.4GHz and a Mali-400 MP4 GPU, which runs a bit faster than the Raspberry Pi’s VideoCore IV.

Although the Fire3 has only twice the number of -A53 cores as the Raspberry Pi 3, and is clocked only slightly faster, Smith’s benchmarks showed a surprising 6.6x times faster CPU boost over a similar RPi 3 cluster. GPU performance was 7.5x faster.

It turned out that much of the performance improvement was due to the Fire3’s native, PCIe-based Gigabit Ethernet port, which enabled the clustered SBCs to communicate more quickly with one another to run parallel computing applications. By comparison, the Raspberry Pi 3 has a 10/100Mbps port.

Performance would no doubt improve if Smith had used the new Raspberry Pi 3 Model B+, which offers a Gigabit Ethernet port. However, since the B+ port is based on USB 2.0, its Ethernet throughput is only three times faster than the Model B’s 10/100 port instead of about 10 times faster for the Fire3.

Still, that’s a significant throughput boost, and combined with the faster 1.4GHz clock rate, the RPi 3 B+ should quickly replace the RPi 3 Model B in Pi-based cluster designs. BitScope recently posted an enthusiastic review of the B+. In addition to the performance improvements, the review cites the improved heat dissipation from the PCB design and the “flip chip on silicon” BGA package for the Broadcom SoC, which uses heat spreading metal. The upcoming Power-over-Ethernet capability should also open new possibilities for clusters, says the review.

Hacker board community sites are increasingly showcasing cluster designs — here’s a cluster case design for the Orange Pi One on Thingiverse — and some vendors offer cluster hardware of their own. Hardkernel’s Odroid project, for example, came out with a 4-board, 32-core Odroid-MC1 cluster computer based on an Odroid-XU4S SBC, a modified version of the Odroid-XU4, which won third place in our hacker board survey. The board uses the same octa-core -A15 Samsung Exynos5422 SoC. More recently, it released an Odroid-MC1 Solo version that lets you choose precisely how many boards you want to add.

The Odroid-MC1 products are primarily designed to run Docker Swarm. Many of the cluster systems are designed to run Docker or other cloud-based software. Last year Alex Ellis, for example, posted a tutorial on creating a Serverless Raspberry Pi cluster that runs Docker and the OpenFaaS framework. Indeed, as with edge computing devices running modified versions of cloud software, such as AWS Greengrass, cluster computers based on SBCs show another example of how the embedded and enterprise server worlds are interacting in interesting new ways using Linux.

Join us at Open Source Summit + Embedded Linux Conference Europe in Edinburgh, UK on October 22-24, 2018, for 100+ sessions on Linux, Cloud, Containers, AI, Community, and more.

The Return of Spectre

By

ZDNet

-

July 12, 2018

The return of Spectre sounds like the next James Bond movie, but it’s really the discovery of two new Spectre-style CPU attacks.

Vladimir Kiriansky, a Ph.D. candidate at MIT, and independent researcher Carl Waldspurger found the latest two security holes. They have since published a MIT paper, Speculative Buffer Overflows: Attacks and Defenses, which go over these bugs in great detail. Together, these problems are called “speculative execution side-channel attacks.”

These discoveries can’t really come as a surprise. Spectre and Meltdown are a new class of security holes. They’re deeply embedded in the fundamental design of recent generations of processors. To go faster, modern chips use a combination of pipelining, out-of-order execution, branch prediction, and speculative execution to run the next branch of a program before it’s called on. This way, no time is wasted if your application goes down that path. Unfortunately, Spectre and Meltdown has shown the chip makers’ implementations used to maximize performance have fundamental security flaws.

Container Adoption Starts to Outpace DevOps

By

Container Journal

-

July 12, 2018

The survey of 601 IT decision-makers conducted by ClearPath Strategies on behalf of the Cloud Foundry Foundation (CFF) finds that 32 percent of respondents have adopted containers and are employing DevOps processes. But the number of respondents who plan to adopt or evaluate containers in the next 12 months is 25 percent, while 17 percent are planning to adopt or evaluate DevOps processes. Overall, the survey finds that within the next two years, 72 percent of respondents either already are or expect to be using containers. That compares to 66 percent who say the same for DevOps.

The total number of organizations that have broadly adopted containers stands at 18 percent, while another 40 percent have deployed containers on a limited basis. Another 40 percent said they are still in the early stages.

Becoming a 10x Developer

By

Kate Heddleston

-

July 12, 2018

When I was first learning to play water polo, a coach told me something I’ve never forgotten. He said, “Great players make everyone around them look like great players.” A great player can catch any pass, anticipating imperfect throws and getting into position. When they make a return pass, they throw the ball so that the other person can make the catch easily.

A 10x engineer isn’t someone who is 10x better than those around them, but someone who makes those around them 10x better.

Over the years I’ve combined my personal experience with research about building and growing effective teams and turned that into a list of 10 ways to be a better teammate, regardless of position or experience level. While many things on this list are general pieces of advice for how to be a good teammate, there is an emphasis on how to be a good teammate to people from diverse backgrounds.

10 Ways to be a Better Teammate

Create an environment of psychological safety
Encourage everyone to participate equally
Assign credit accurately and generously
Amplify unheard voices in meetings
Give constructive, actionable feedback and avoid personal criticism
Hold yourself and others accountable
Cultivate excellence in an area that is valuable to the team
Educate yourself about diversity, inclusivity, and equality in the workplace
Maintain a growth mindset
Advocate for company policies that increase workplace equality

Developer Recruitment Drives Open Source Funding

By

The New Stack

-

July 12, 2018

The latest 2018 Open Source Jobs Report points to several ways employers can help developers. For the study, the Linux Foundation and Dice surveyed over 750 hiring managers involved with recruiting open source professionals.

Due to the survey’s subject, it is not surprising almost half of hiring managers (48 percent) say their company decided to financially support or contribute open source projects to help with recruitment. Although this sounds incredibly compelling, it is fair to question how much hiring managers actually know about open source management. Since 57 percent of hiring managers say their company contributes to open source projects, a back-of-the-envelope calculation says that 84 percent of companies that contribute to open source are doing so at least in part to get new employees.

The New Stack and The Linux Foundation have teamed up to survey the community about ways to standardize and promote open source policies programmatically. We encourage readers to participate.

Certification Plays Big Role in Open Source Hiring

By

Esther Shein

-

July 11, 2018

With high demand for Linux professionals and a shortage of workers with these skills, it’s small wonder that employers are willing not only to train their staff but also to help them get certified. Forty-two percent of employers report having trained existing workers on new open source technologies this year to meet their needs, compared to only 30 percent in 2017, according to the 2018 Open Source Jobs report.

The report, produced by Dice and The Linux Foundation, also found that 38 percent of companies are less likely to rely on outside consultants, compared with 47 percent in 2017. Consequently, they are turning to training to keep up in a fast-paced, ever-changing tech environment. Sixty-four percent of hiring managers say their employees are requesting or taking training courses on their own – the exact same percentage as last year.

Why? There is a strong belief that IT certifications are a reliable predictor of a successful employee, according to IT trade association CompTIA. In its own research, CompTIA found five reasons why 91 percent of employers believe IT certifications play a big role in the hiring process:

Certifications help fill open positions
Most companies have IT staff who have certifications
Certified IT pros make great employees
IT certifications are increasing in importance
Training alone is not enough

Certification as an incentive

Forty-two percent of employers are using training and certification opportunities as an incentive to retain employees, up from 33 percent last year and 26 percent in 2016, this year’s Open Source Jobs Report found. Underscoring the importance employers place on certifications: Nearly half (47 percent) of hiring managers say employing certified open source professionals is a priority for them, essentially the same number as last year.

The same percentage say they are more likely to hire a certified professional than one without a certification. An increasing number of companies are willing to pay for certifications, with 55 percent that reported they helped to cover the costs of certifications this year, up from 47 percent last year and 34 percent in 2016. Only 17 percent say they would not pay for certifications, a decline from 21 percent last year and 30 percent in 2016.

Certifications is a benefit that can be used as a recruiting tool, and employers that offer certification courses for full-time employees should mention it in job postings, the report stresses. Similarly, professionals seeking this benefit should make clear during the interview process their desire to continue their education and become certified while employed.

However, there continues to be debate over the value of certifications versus on-the-job experience. There are many seasoned tech professionals who claim years of experience is more important, yet the average certification now represents a 7.6 percent premium on an IT pro’s base salary, according to research firm Foote Partners, which publishes an annual IT Skills and Certifications Pay Index. Specifically, gains were seen in networking and communications and applications development and programming language certifications, the firm says.

A significant majority (80 percent) of open source professionals say certifications are useful to their careers, up slightly from 76 percent in the previous two years. The main reasons cited are that certifications enable employees to demonstrate technical knowledge to potential employers (stated by 45 percent of respondents), and certifications make professionals more employable in general (33 percent). Forty-seven percent of open source professionals plan to take at least one certification exam this year, up from 40 percent in 2017.

Vendor neutrality matters

Employers increasingly want vendor neutrality in their training providers, with 77 percent of hiring managers rating this as important, up from 68 percent last year and 63 percent in 2016. Almost all types of training have increased this year, with online/virtual courses being the most popular. Sixty-six percent of employers report offering this benefit, compared to 63 percent in 2017 and 49 percent in 2016. Forty percent of hiring managers say they are providing onsite training, up from 39 percent last year and 31 percent in 2016; and 49 percent provide individual training courses, the same as last year.

Additionally, employers say they increasingly see benefits from sending employees to conferences. Fifty-six percent of hiring managers said they pay for employees to attend technical conferences, up from 46 percent in 2017.

Download the complete Open Source Jobs Report now and learn more about Linux certification here.

A SysAdmin’s Guide to Network Management

By

OpenSource.com

-

July 11, 2018

If you’re a sysadmin, your daily tasks include managing servers and the data center’s network. The following Linux utilities and commands—from basic to advanced—will help make network management easier.

In several of these commands, you’ll see <fqdn>, which stands for “fully qualified domain name.” When you see this, substitute your website URL or your server (e.g., server-name.company.com), as the case may be.

Ping

As the name suggests, ping is used to check the end-to-end connectivity from your system to the one you are trying to connect to. It uses ICMP echo packets that travel back to your system when a ping is successful. It’s also a good first step to check system/network connectivity. You can use the ping command with IPv4 and IPv6 addresses.

Continuous Integration and Delivery Tool Basics

By

HPE Blog

-

July 11, 2018

CI/CD tools are key to today’s agile, container-driven software production cycle. This “explain like I’m 5” overview helps you get started.

Once upon a time, back when waterfall was the primary software development methodology, you could spend months or years working on a single software project. With the help of agile methods, programming cycles were reduced to days or weeks.

How quaint all that looks now! Weeks? Ha! In today’s DevOps-driven world, we use continuous integration, continuous delivery, and continuous deployment (CI/CD) to update programs in time frames measured in days or hours. For example, with its CI/CD pipeline, Netflix takes less than half an hour to move its programs from code check-in to multi-region deployment.

Automation is a key part of the CI/CD process, so naturally there are several tools that aim to help you manage the tasks. To set the stage, let’s review the basics of CI/CD before I give you an overview of some popular CI/CD tools.

How VMware Manages Open Source Code and Compliance

By

eWeek

-

July 11, 2018

VMware is an active member of the open-source community, contributing its efforts to multiple projects and organizations. Helping to organize and lead VMware’s open-source initiatives is the company’s chief open source officer, Dirk Hohndel.

In a video interview with eWEEK, Hohndel explains where open-source technologies are being used by VMware and how the company contributes to development efforts. Hohndel joined VMware in 2016, after spending 15 years as the chief Linux and open source technologist at Intel.

Open source at VMware has multiple aspects, including code contributions, software projects and code used within VMware’s commercial products. Open-source licenses often include compliance requirements that organizations like VMware need to track. For example, VMware’s vSphere virtualization platform includes open-source components, with license compliance and reviews conducted by Hohndel’s team.

“On the engineering side, we have open-source teams in pretty much every business unit that are working with open-source components and contributing to projects,” he said.

Resilient RDMA IP Addresses

By

Jason Perlow

-

July 10, 2018

Oracle Linux kernel developer Sudhakar Dindukurti contributed this post on the work he’s doing to bring the Resilient RDMA IP feature from RDS into upstream. This code currently is maintained in Oracle’s open source UEK kernel and we are working on integrating this into the upstream Linux source code. 1.0 Introduction to Resilient RDMA IP The Resilient RDMAIP module assists ULPs (RDMA Upper Level Protocols) to do failover,…

Click to Read More at Oracle Linux Kernel Development