New features of multiple Raspberry Pi Compute Modules 4 brought to new ClusBerry series

Accompanying the release of ModBerry 500-CM4 and AI GATEWAY 9500-CM4, we present to you a cluster version of the device, called ClusBerry 9500-CM4. Main difference between standard Gateway and ClusBerry is the possibility to include multiple Raspberry Pi Compute Module 4 in one device, as well as the intended suitable amount of wired and wireless interfaces, suited for the project.

Fully configurable devices are something desirable in the IoT market, where high performance and low cost is a key factor to success of implementation. TECHBASE’s Industrial IoT Ecosystem gives the opportunity to adjust ordered devices with certain resources and cut unnecessary I/Os, lowering the total cost of the device. 

Raspberry Pi Compute Module 4 Cluster

Reason for use of Raspberry Pi CM4 cluster in ClusBerry 9500-CM4

Various implementations must have guaranteed high hardware performance to react fast enough in real time. For this purpose, the arrays of processor blocks are constructed to be assigned to individual tasks. For several years now, attempts have been made to use various types of SBC for this purpose, including, of course, Raspberry Pi. However, the practical effectiveness of such solutions so far has not been of interest for several reasons. First of all, these solutions were most often associated with many mechanical limitations and the structure of the matrix itself required excessive wiring, preventing failure-free operation and the cost of the entire maintenance of the structure.

Raspberry Pi Compute Module 4 Cluster

This is where Raspberry Pi Compute Module can shine, but due to the hardware speed limitations of the buses in this module, it was not completely effective and was rather a development platform. Altho the introduction of new Compute Module 4 has opened the possibility to construct and maintain effective hardware matrix solutions with the use of both PCI-Express buses and 1GBps Ethernet. Therefore, the ClusBerry 9500-CM4 opens up completely new capabilities of utilizing cluster solutions for Industrial Automation and server applications.

Wide range of ClusBerry modules

ClusBerry 9500-CM4 supports up to 8 cluster modules and comes with a variety of interchangeable modules to choose from, including:

  • Standard 9500-CM4 cluster module with Compute Module 4 and chosen configuration:
    • I/O Controller with range of DI, DO, AI, 1-Wire, RS-232/485 and CAN interfaces
    • Communication Gateway with wired (1/2x Ethernet, Serial Ports) and wireless interfaces (LTE-cat.M1, 4G, 5G, LoRa, ZigBee, Z-Wave, Wireless M-Bus)
    • AI Gateway with 1x Coral Edge TPU via PCIe M.2, introduced in December 2020: or up to 4x Coral Edge TPU via USB3.0
  • NAS File Server with 2x SSD SATA III and RAID support, managed with Nextcloud or ownCloud software
  • USB3.0 Hub for 5G communication, Modems, AI Cluster and peripherals
  • Gigabit LAN/WAN Router with additional 2.5GbE network card as an independent switch/router shielded from the mainboard cluster network
  • SuperCap / Power management module for backup power supply (supercapacitors / Li-Ion battery) and sleep mode management aided with ESP32-module
  • Additional expansion cards, with resources suited for the installation (DIO, AIO, Serial Ports and dedicated sensor cards, detailed below)
Raspberry Pi Compute Module 4 Cluster

ClusBerry 9500-CM4 with available expansion cards 

ClusBerry 9500-CM4 can be equipped with multiple expansion cards, e.g. serial RS-232/485 ports, range of digital and analog I/Os, USB, HDMI and Ethernet. Interfaces can be expanded with additional I/Os and opto-isolation, relays, Ethernet, 1-Wire, CAN, M-Bus Master and Slave, accelerometer and many more features like TPM Security Chip & eSIM. The device can also be equipped with additional SuperCap backup power source for continuous work and safe boot/shutdown in case of emergency.

ClusBerry 9500-CM4 series also offers a standard PCI module support for various wireless communication protocols, such as:

  • GSM modem (4G/LTE and fast 5G modem)
  • economic NarrowBand-IoT technology
  • LoRa, ZigBee, Z-Wave, Sigfox, Wireless M-Bus
  • secondary Wi-Fi/Bluetooth interface or Wi-Fi Hi-Power
  • custom wireless interfaces
Raspberry Pi Compute Module 4 Cluster

Software cluster management with Docker and K3s Lightweight Kubernetes

With use of Docker-based and Kubernetes solutions, installation and management of ClusBerry 9500-CM4 is easy and backed with a large community for further support and development. Kubernetes is a portable, extensible open-source software platform for managing containerized tasks and sites that enables declarative configuration and automation. The Kubernetes ecosystem is large and dynamically developing. Kubernetes services, support and tools are widely available.

Kubernetes provides:

  • Detection of new services and traffic. Kubernetes can balance the load and redirect the network traffic to ensure the stability of the entire installation.
  • Kubernetes data storage management enables you to automatically mount any type of storage system – on-premises, from cloud providers and others.
  • Automatic deployment and rollback. You can describe the expected state of your installation with Kubernetes, which will take care of bringing the actual state to the expected state in a controlled manner. For example, with Kubernetes, you can manage your cluster modules at ease, boot modules from one to another, upgrade firmware crosswise and provide safe operation of each module
  • Automatic management of available resources. ClusBerry 9500-CM4 provides a cluster of modules that Kubernetes can use to run tasks in containers. You determine the CPU power and RAM requirements for each container. Kubernetes arranges containers on machines in such a way as to make the best use of provided resources.
  • Self-healing Kubernetes reboots containers that have stopped working, replaces them with new ones, forces disabling containers that are not responding to certain status queries, and does not announce their availability until they are ready to run.
  • Managing confidential information and Kubernetes configuration with TPM Security Chip allows you to store and manage confidential information such as passwords, OAuth tokens and SSH keys. Secured data and configuration information can be provided and changed without having to rebuild the container image and without exposing sensitive data in the overall software configuration.
Raspberry Pi Compute Module 4 Cluster

ClusBerry 9500-CM4 availability

First prototypes are being developed, since Compute Module 4 is already available for the purchase. Delivery time for various configurations of ClusBerry will be approximately 2 months, depending on the CM4 supply on the market and chosen expansion cards. For more information contact TECHBASE’s Sales Department via email or Live Chat here.

Coral Dev Board - a new Raspberry Pi-like platform from Google

The Coral platform launched by Google is expected to contribute to the easier creation of systems in the field of the Internet of Things. During CES 2020, the manufacturer will devote a lot of attention to her.

Google presented a new, miniature accelerator module for the Coral platform. The sensor has dimensions of only 10×15 mm, so it is smaller than the American one-cent coin. Coral is a Google platform created to facilitate the construction of equipment compatible with the Internet of Things technology, equipped with artificial intelligence.

Components produced by Google have already been used to create many intelligent systems used, for example, in healthcare, agriculture or technologies supporting the functioning of cities. Coral works well also in offline mode and with limited connectivity. The new module is expected to increase its functionality.

The Coral Dev Board together with the USB and PCIe accelerator went on sale in 2019, and is now being sold in 36 countries. We can expect many innovations related to this technology during CES 2020. Google announces that it has prepared various demonstrations of the new module capabilities for this meeting.

Google’s Coral Dev Board

Coral Dev Board

Unlike popular single board, Raspberry Pi or Raspberry’s Chinese competitors, according to Google, Coral Dev Board is to be a specialized computer – it is primarily targeted at developers who want to have a device on which algorithm learning is possible. The computer is running Mendel Linux built on Debian, we will also find Tensor Flow Lite libraries with ready-to-compile models. No internet connection is required in the learning process – the entire process is done locally.

From the hardware side, the most important here is the Edge TPU coprocessor – a Google tensor system designed to be used together with the Tensor Flow library used in machine learning. In addition, the board got the ARM 4-core NXP i.MX 8M processor, GC7000 Lite Graphics, as much as 1 GB of RAM (LDPRR4) and 8 GB of eMMC memory. Coral connects to other devices via Wi-Fi oral USB Accelerator and Bluetooth 4.1.

Therefore, it cannot be said that, compared to other SBC Coral, it is hardware limited. What’s more, it has a microSD card slot, two USB type C sockets (one for power supply, the other for data transfer), minijack output, HMDI output, GPIO pins. The device can also be connected to the Internet via a cable with an RJ-45 plug. It is also possible to buy a camera useful in machine learning image processing algorithms.

Industrial use of Raspberry Pi-like development boards

Introduced in November 2017, the ModBerry M300 series, based on NanoPi NEO revolutionised the economic segment of Industrial IoT devices and proved, that automation and monitoring can be done effectively with low expenditure on industrial installations.

ModBerry M300 O1 based on OrangePi Zero Plus features Allwinner H5 (Quad-core Cortex-A53) SoC, moderate 512MB RAM, storage memory option with microSD slot, USB and Gigabit Ethernet port. The wireless communication is supported with onboard Wi-Fi module.

Offering much higher performance and wider feature range, the ModBerry M300 O2 features same SoC as M300 series, but thanks to OrangePi Zero Plus2 means, the device is equipped with onboard 8GB eMMC, extra microSD expansion slot as alternative and wired/wireless interfaces, e.g. HDMI, Wi-Fi, Bluetooth 4.0.