In the light of this year’s circumstances, AWS’ annual re:Invent conference had to be held virtually, primarily on the basis of video conferences. This, however, did not compromise the event’s impact and significance. If anything, the fact that this year’s conference was free for participants and spanned over a content-packed three weeks, has only enhanced its informative value.
To provide you with a short recap of re:Invent’s 21 days of announcements, keynote talks, and updates, we’ve summarized some impressions, gathered by AllCloud team members at the conference. Below are the highlights they valued most.
This year’s Network and IoT announcements
Shachar Carmel, Head of Solution Architects at AllCloud
Among everything AWS has to offer, my personal favorites are services related to Network and IoT. Fortunately, the last three weeks have been packed with announcements regarding these fields. Here are some of the highlights I liked the most:
Network Services
VPC Reachability Analyzer
While some of the network-related releases, such as AWS Network Firewall and the Gateway Load Balancers, have already been announced a few weeks before re:Invent, one of the interesting announcements during the conference itself was the VPC Reachability Analyzer feature.
This feature provides an easy way to perform network connectivity tests from resources within a VPC to destinations in the same VPC, as well as in other VPCs. This helps with the troubleshooting of network issues in simple or complex network structures.
A test or an Analyze Path can include resources such as EC2 instances, ENIs, Gateways, as well as required ports and protocols (TCP/UDP). In case that a destination is not reachable during the test, a comprehensive report is generated, detailing the reasons that caused traffic to fail. This makes it easy to understand and diagnose problems. If the test does succeed, a graphical report is generated, displaying the network path from the resource to the destination.
IoT
IoT Defender ML
One of the biggest concerns in any IoT workload is device security. Right from the beginning, when the device is being provisioned and registered, you need to make sure that it has a unique certificate that allows it to connect to the IoT cloud endpoint. Then you need to make sure that you grant the device the right permissions to publish and subscribe to specific topics only.
Once this is all set, the device should be able to connect to the cloud and send its messages. But how can you protect your connected devices? How can you make sure that the device is acting normally and that it is only sending messages it is supposed to send and not spamming your endpoint, or sending information that might jeopardize your workload or IoT application?
To address these concerns, AWS introduced the AWS IoT Device Defender, back in 2017. AWS IoT Device Defender is a fully managed service that helps with protecting connected devices by setting up rules and alarms that match your devices’ behavior. However, devising these is often a difficult task, since it requires a deep understanding of anticipated device behavior. You’d have, for example, to anticipate how many messages a device should be sending or receiving, what a standard message size should be, and other metrics of this kind.
IoT Defender ML makes this task much simpler as you can now build your rules and alarms based on your device’s historical behavior. If you’re connecting a new device, Defender will gather information from the device for 14 days, and use it as training data for the ML model. Subsequently, it will keep and retrain the model with new data every day. When an anomaly is detected, built-in mitigation actions can be taken automatically, such as revoking the device certificate. IoT Device Defender ML can be enabled on a single device or on a group of devices and also provide reports that show the device’s state.
A few other interesting IoT announcements included:
IoT Panorama – a machine learning appliance that makes it easy to deploy computer vision technologies on-prem using existing cameras.
IoT Monitron – an end-to-end solution for detecting abnormal behavior in industrial machinery.
SageMaker and QuickSight Q
Carsten Riggelsen, Leader Data/AI DACH at AllCloud
I found the keynote talk by Andy Jassy really exciting. A couple of main themes struck me as being interesting in particular. I’ll focus here on SageMaker and Quicksight, and a couple of extra features that will be brought to market by Amazon over the next couple of months and in early 2021.
SageMaker is an all encompassing machine learning platform on AWS. Sagemaker Data Wrangler, is a new extended feature of Sagemaker that helps to pre-process and prepare data before you actually develop your models.
Data preparation and data munging is something that may consume around 80 to 90 percent of the time of any Machine Learning (ML) project. Data Wrangler will make this much more efficient, while also removing the need to rely on ad-hoc solutions from 3rd paties.
Another add-on to SageMaker is called SageMaker Feature Store, which is a one-stop-shop solution for dealing with so-called ML features. The simplicity which comes about with this store will greatly benefit data science teams and machine learning engineering teams using the SageMaker platform. With SageMaker Feature Store you won’t have to constantly reinvent the wheel and will have to spend less time doing the same things (i.e. dering the same feature over and over again from raw data) that have already been done by your teammates.
My next favorite takeaway caters to a slightly different audience, namely the users who use QuickSight for business intelligence. The users of this solution are often challenged by having to pose technical queries to get the insights they need. Instead, they want to interact with an environment, using natural language. And this is exactly what QuickSight Q will allow for.
Analytical questions can be posed in natural language and will be converted behind the scenes to queries using Amazon’s intelligent techniques effectively yielding the results that you’re looking for. We’re not dealing here with pre-canned questions and sentences, This way of interacting with data certainly lowers the entry bar for obtaining insight from data – it is a game-changer and beyond anything we’ve seen until now.
Services on the AWS Marketplace
Eric Crump, SVP, AWS North American Practice Lead at AllCloud
During re:Invent Amazon announced the expansion of its AWS Marketplace, which from now on will also include professional services by consulting partners such as AllCloud. This allows customers to purchase software and related services through one centralized platform.
AllCloud is one of the first AWS Consulting partners to offer services on the AWS Marketplace, among them are the AllCloud Security Hub Deployment Accelerator.
The Security Hub Deployment Accelerator is a packaged professional services offering, created to help organizations modernize their security toolsets and capabilities, and accurately and quickly move to AWS next-generation cloud-native security services, AWS Control Tower, and AWS Security Hub.
On Manufacturing Announcements
Traditionally, the Manufacturing Sector has been known to be reluctant when embracing the Enterprise Cloud. This has had several reasons. Most notably, the Legacy technology employed in most manufacturing environments used to be, and often still is, difficult to migrate, offering no clear path to the cloud. But probably most importantly, IoT aside, most manufacturers did not find compelling business cases that would have justified the overhead entailed by a large scale migration project.
The new releases by AWS, announced at re:Invent 2020, specifically tackle these two challenges and clear the way for manufacturers to embrace the enterprise cloud.
Legacy Technology
AWS has been working to provide best-in-bread design patterns and migration approaches for both Mainframe and Midrange computing. Working with partners such as AllCloud, AWS has developed logical ways to methodically modernize legacy technologies, which supersede common rip-and-replace approaches. As part of this effort, AWS introduced a new competency class at re:Invent, focused on Mainframe Migration.
Compelling Business Cases
AWS has increased its commitment to the manufacturing sector, and has invested heavily to tackle its sector-specific problems. As part of this effort, AWS has recently announced three significant developments at re:Invent 2020.
Amazon Monitron – Predictive Maintenance
With the announcement of Amazon Monitron, AWS has launched an end-to-end Predictive Maintenance solution, Amazon Monitron incorporates IoT Monitron sensors that feed prebuilt ML models to AWS. With this, Manufacturers can now leverage ML and AI to proactively model and address plant equipment maintenance.
Amazon Lookout
Amazon Lookout comprises a series of APIs and prebuilt ML Models that use data from existing equipment sensors to detect abnormal behavior, so action can be taken before machine failures occur.
AWS Panorama
Another area for anticipated growth, which has up until now been beyond the reach of most manufacturers, are Computer Vision (CV) technologies. CV employs smart cameras and specialized equipment to detect anomalies and to perform predictive analytics. AWS Panorama simplifies this process. Installed at the plant site, AWS Panorama leverages already existing Camera IP technology to provide CV analyses with predictive capabilities at the fraction of the traditional overhead and costs.
These announcements are part of Amazon’s Connected Factory vision, which aims to help manufacturers to harness the power of the enterprise cloud. Together with AWS’ innovation at the IoT front, and the launch of AWS outpost in 2019, the recent announcements attest to AWS’ commitment to the manufacturing sector.
These are just a few of the many new services and announcements launched by AWS at re:Invent 2020. Contact us for questions and stay tuned to our expert blog for more updates, playbacks and resources from the online mega event.