Earlier in August, I attended my 7th VMworld US (13th if we count VMworld EMEA) and first as a non-VMware employee. VMworld always been an event full of industry defining innovations that influence technical architectures for enterprise customers globally delivering business transformations. This year’s event was no different with many major announcements across multiple areas such as multi-cloud, software-defined data center & security. One of the major advancements discussed was the re-architecting of VMware vSphere to deeply integrate and embed Kubernetes. This evolution will provide an application centric infrastructure for both developers and IT teams. VMworld is also an event that showcases innovations in areas setting technology trends of the future. One such area has been innovations at the edge with expanding new frontiers for virtualization based on efforts around ESXi on Arm.
VMware unveiled their ESXi on Arm efforts first at VMworld US 2018 in a keynote demo of a 2-node ESXi cluster running on a very cost & power-effective 64-bit Arm powered Marvell ARMADA 8040 Macchiatobin reference board. This cluster was hosting mission critical workloads inside a wind turbine representing an edge scenario. With these virtualized instances, fault tolerance can now be achieved for mission critical sensor collection workload and making it easier and safer to update devices in the field. It also allows these edge nodes to be managed the same way as servers in data centers.
ESXi on Arm cluster running in a wind turbine
One of the use cases of pushing compute closer to the edge is to process the data generated by increasing number of connected IoT devices locally. Arm Neoverse outlines this vision of powering next generation of edge to cloud infrastructure. Arm experts have predicted more than one trillion devices by 2035. This explosion in data changes the traffic patterns of today and data will flow into the network with increasing need to extract and analyze information at the edge. The need for more data analysis at the edge could be for various reasons such as latency sensitivity, bandwidth requirements, security and real time processing to name a few. It is also driving machine learning and artificial intelligence at the edge. The Arm Neoverse ecosystem is at the forefront of this technological shift and providing real solutions for customers.
IoT driven Data Explosion
As an illustration of bringing compute closer to the data source, the VMware Office of the CTO showcased ESXi running on an Arm-based Mellanox BlueField SmartNIC hosting an Ubuntu VM to process weather sensor data locally. Weather sensor data is ephemeral in nature and processing it in a local database to perform machine learning on it.This is becoming an increasing trend in leveraging edge computing for IoT deployments.
ESXi on Arm processing Weather Sensor data on a SmartNIC
VMware has showcased multiple use cases for ESXi on Arm on both Broadcom Stingray & Mellanox BlueField SmartNICs. These use cases include management and/or data plane offload, security via tenant isolation, high-throughput analytics, running low-latency applications, storage offload, network offload and providing energy efficiency for workloads. The VMworld US 2019 keynote included a demo running four hypervisors concurrently on a single x86 server without nesting any of the hypervisor instances. Seemingly impossible, it included three ESXi on Arm instances running on SmartNICs hosting different workloads such as Snort NIDS (logging suspicious traffic to a local database) and a pair or SD-WAN virtual network function (VNF) VM in fault tolerant (FT) mode.
Multiple ESXi on Arm-based SmartNICs
This use case is very interesting as a true Network Functions Virtualization (NFV) networking appliance managed by vCenter in a consistent manner with ESXi on Arm running on these SmartNICs providing ability to run applications. In addition to ESXi, running NSX-T on a SmartNIC was showcased which enables use cases such as freeing up costly x86 cycles, providing isolation & security and running bare metal and containerized workloads.
There are additional virtualized edge use cases that require specialized platforms. One such platform showcased as an Arm-based stackable edge computing platform is HiveCell. It is a small, low-energy and inexpensive server that is physically stackable via connectors providing power and networking connectivity between units. This form factor opens up interesting use cases in areas such as ruggedized environments (Oil Rigs, Military deployments), retail, manufacturing, weather, data science etc.
Hivecell Cluster
For edge computing, security is also one of the most challenging and important factors customers need to consider given that the threat models at the edge are very different than those in data centers. Also, there is significant fragmentation of devices and platforms at the edge and variations in standards as well. To overcome this challenge, VMware showcased early prototype work for implementing Unified Secure Enclaves which explores a unified Trusted Execution Environment (TEE) framework for virtualized environments. This project included an Arm-based prototype implementation as shown below.
Security for Edge
In addition to these technology innovations for edge/IoT, VMware also unveiled an Early Customer Evaluation Beta program for Arm-based workloads on VMC with following criteria:
Early customer Evaluation for VMC on Arm
As part of Arm and VMware’s joint efforts, we are looking to engage with partners and customers to gather feedback and requirements on how virtualization on Arm at the edge and in the cloud benefits your current and future business needs. If you are visiting Arm TechCon, please visit the VMware team showcasing the innovations at the Infrastructure Zone (Booth 143). In addition, Arm will be exhibiting these innovations jointly with VMware at VMworld 2019 Europe.
[CTAToken URL = "mailto:Bhumik.Patel@arm.com" target="_blank" text="Contact Bhumik Patel" class ="green"]