Scaling IoT Analytics from Data Center to Fog Server to Network Edge

February 19, 2019


Scaling IoT Analytics from Data Center to Fog Server to Network Edge

Current data analytics architectures are full of challenges. Networking and storage costs, latency concerns, and security issues are present at and between each tier of such systems.

IoT systems filter, pre-process, and aggregate heaps of data at the network edge. This data is typically transmitted to higher level fog or cloud platforms that perform further analysis and make decisions based on the information at hand. These decisions are then communicated back to the edge, where they are put into action in the physical world.

However, this data analytics architecture is full of challenges. Networking and storage costs, latency concerns, and security issues are present at and between each tier of such systems (Figure 1). All of this infrastructure must also be managed, which can be complex, costly, and detract from a company’s core business objectives:

  • Developers must ensure that there are no disruptions caused by limited or intermittent connectivity.
  • Latency and throughput delays cannot impact the accuracy of data aggregation and analytics, or corresponding control functions.
  • The end-to-end system must ensure the anonymity of sensitive and private information.
Figure 1. Data analytics architectures that comprise edge, fog, and cloud tiers can quickly become mired in cost, complexity, and security issues (Source: ERPINNEWS).

For growing organizations, these obstacles dictate that more and more data analysis be moved to edge devices. Doing so, however, results in tradeoffs for device designers who must juggle the size, power consumption, heat dissipation, and cost envelopes of embedded systems with the compute, storage, and networking requirements of advanced data analytics.

That is, unless you bring the data center to the IoT edge.

Enterprise Analytics at the Edge

To support the growing need for IoT analytics at the edge, Tyan Computer Corporation’s Tempest server motherboards bridge the compute, storage, and networking performance of the enterprise with the compact form factors, wide operating temperatures, and long lifecycles of embedded systems. Based on Intel® Atom®, Core™, and Xeon® processors, the scalable portfolio enables IoT organizations to position analytics capabilities in on-premise servers, local gateways, or on the edge device itself.

Take, for example, the Tempest EX S5550-EX single-socket server motherboard targeted at on-premise appliance development. The 9.6" x 9.6 micro-ATX board supports chipsets up to Intel Xeon E-2100 series processors, providing four to six cores of multi-threaded performance at a maximum thermal design power (TDP) of only 95 watts.

Figure 2. The Tyan Tempest EX S5550-EX single-socket server motherboard equips chipsets up to the Intel® Xeon E 2100 series to deliver powerful IoT data analytics compute at only 95 W.

The onboard Xeon® processors integrate a suite of technologies that optimize data analytics in mixed-criticality environments like the Industrial IoT edge. Intel® Virtualization Technology (Intel® VT-x), Intel® VT-x wit Extended Page Tables (EPT), and Intel® Virtualization Technology for Directed I/O (Intel® VT-d). These features help partition data analytics workloads from other mission-critical functions, while also ensuring the security, reliability and performance of memory and peripherals.

The Tempest EX S5550-EX brings these capabilities out via up to four 1000Base-T LAN ports, two PCIe Gen.3 x8 + 1 PCIe Gen.3 x4 slots, eight SATA 6G ports for storage, as well as audio and display interfaces. For intensive analytics operations that require frequent memory accesses, the server motherboard supports up to 64 GB of fast 2666/2400 MHz DDR4 DRAM with ECC.

But more important for IoT analytics applications is the platform’s seven-year lifecycle guarantee and wide operating temperature of 0-55 ºC.

Next, the Tempest EX S5555-EX single-socket workstation board allows IoT engineers to take analytics even closer to the edge. Based on an 8th generation Intel® Core™ i3, i5, or i7 processor, the EX S5555-EX brings many of the same virtualization and networking features as the S5550-EX and also shares the micro-ATX form factor. Only here it’s worth mentioning that IoT applications catering to huge data traffic for deep learning-centric analytics can benefit from the inherent performance per watt (PPW) advantages and energy efficiency features enabled by Core® processors.

The 9.6" x 9.6" EX S5555-EX board for workstation and IoT applications also supports wide operating temperature of 0-55℃ and EMC Class B.

Figure 3. The four DDR4 dual in-line memory module (DIMM) slots in Tempest EX S5555-EX further boost the embedded system’s storage credentials. (Image: TYAN)

Consistent IoT analytics

The embedded systems chronicled above are only two of the ten solutions in Tyan’s scalable Tempest family of products, yet they demonstrate how standards-based platforms powered by robust processors efficiently meet the demands of enterprise analytics compute at the IoT edge. And how the x86-based embedded systems provide the flexibility to run high-performance analytics and conventional software elements on a single processor.

Leveraging embedded platforms the Tyan Tempest line also allows developers to efficiently reuse hardware and software components from edge to cloud, thus reducing costs and minimizing time to market.

Visit the TYAN booth at Embedded World in Nuremberg, Germany on 26-28 February 2019 to learn how scalable compute solutions can bring high performance analytics to the lowest tiers of your IoT infrastructure.

Charles Chen is the Director of Product Planning & Marketing at Tyan Computer Corp.