This is part two of our round up of the announcements from Andy’s AWS re:Invent 2019 keynote! You can find part one here.
“I can’t move from on-premises”
Let AWS help you with that.
Sometimes, it can be really hard to move from on-premises. It may be due to legacy support (see the announcement of the AWS EMP service for Windows which may help you). Often, though, the workload (such as a SCADA system) or the user experience demands low latency connectivity. Some new announcements may help to break down those barriers.
- AWS Outposts is now generally available. Outposts is a fully supported hardware solution that you can install in your own data centre. The hardware is managed by AWS but you interact with it as if it was in the AWS Cloud using familiar APIs. Support has now been added for RDS, ECS and EKS;
- AWS Local Zones places AWS compute, storage and database closer to large population centres where no AWS Region exists today. AWS Local Zones will allow you can run latency-sensitive applications close to end-users and resources in a specific geography. It is generally available in Los Angeles today with other locations to be announced;
- Finally, AWS has announced Wavelength, directly targeted at 5G mobile connectivity. In an initial partnership with Verizon, AWS has embedded compute and storage services at the edge of 5G networks. AWS Wavelength minimises network hops and latency between a 5G device and AWS. It’s ideal for gaming and live video streaming applications where low latency is critical. When will we see it in the Telstra 5G environment?
How do you make sense of all this data? How do you streamline your user experiences and turn into a proactive not reactive organisation? Machine Learning has been hot news for a few years now but it’s still not trivial. AWS has many offerings in this space from building your own models through to abstracted AI services such as Rekognition.
AWS is consistently trying to make it easier to put machine learning in the hands of every builder. Here are some new announcements that do just that.
- SageMaker Studio is the world’s first IDE for machine learning. Builders can develop code, track experiments, visualize data, and debug and monitor all within a single GUI. It will significantly lower the barrier to entry to the ML world;
- SageMaker Notebooks allow you to create and share Jupyter notebooks without having to manage infrastructure. You can also seamlessly switch instance sizes without having to reload models and data;
- SageMaker Experiments allows you to organise and manage thousands of jobs in one console, easily comparing results from different input parameters;
- SageMaker Debugger does what is says on the tin. It automatically collects data and inspects models to speed the sometime complex analysis of training issues. It provided recommendations to optimise training times and improve model quality;
- SageMake Model Monitor helps to detect concept drift in training models that can result in less reliable outcomes. It will visualise issues that could be affecting the model;
- Our favourite announcement is SageMaker AutoPilot, an end to end service for machine learning. It takes and transforms training data, selects an algorithm, trains a bunch of models and then inspects and compares the models. You can then deploy your chosen model with the click of a button (or an API call). It takes away a lot of the work in trying to decide on the algorithm and then having to set the parameters for the model to maximise accuracy. It will make machine learning simpler and more accessible than ever!
Amazon Kendra (a Kendra is an ancient English meaning a knowledge bearer) is a new enterprise search service that uses machine learning.
Kendra brings a new approach to Enterprise Search, not relying on keyword matches. It provides a natural language search capability. Kendra can search across content from file servers, SharePoint, intranet sites, file sharing services and more. It comes with a bunch of pre-built connectors. Of course, if you need to, you can build your own connector. Once you have configured your Kendra environment, you can add it to your own Intranet with a few lines of code! We at RedBear could make great use of Kendra!
Amazon Connect is one of AWS’s fastest growing services. It is a call centre as a service solution that can be up and running in a few minutes. AWS have been steadily adding new features since its release in March 2017.
Now AWS had further enhanced it with Contact Lens. It automates the integration to Transcribe and Comprehend amongst other enhancements. It will enrich the data available on calls and enable organisations to be proactive in their call centre engagements with customers.
Managed Fraud Detection
Building on the IAM and S3 Access Analysers announced last week, AWS is continuing to role out new security services. At last count, there are now nearly 30 dedicated security services. After the AWS re:Invent 2019 keynote, there were a few new ones to add!
First up is the Managed Fraud Detector service. This is huge! Building and managing an effective fraud detection service is a lot of work. Traditional solutions rely on keeping business rules up to date as the behaviour of fraudsters changes. That’s no trivial task.
This new service is aimed at online transactions. It uses machine learning and 20 years of fraud detection experience from Amazon to identify potentially fraudulent activity. No machine learning expertise is required. You upload your historical fraud dataset and select from one of the pre-built models. From there, the Fraud Detector builds a custom model, enriching data, choosing an appropriate algorithm, training and tuning the model. The final step is to enable business rules around the model predictions. That’s it.
To use the Fraud Detector service, make an API call from your online application to receive real-time fraud predictions and take action based on your configured detection rules! We can’t wait to try this one out with some of our Financial Services customers.
Amazon Detective is a new service that offers SIEM like capability for the AWS Cloud. It provides a simple way to investigate and quickly identify the root cause of potential security issues or of suspicious activities. Amazon Detective automatically collects log data from AWS resources. It uses machine learning (along with statistical analysis and graph theory) to generate a linked set of data. Security teams can then make use of this dataset, speeding up the investigation.
It analyses events from many AWS sources (such as VPC Flow Logs, CloudTrail and GuardDuty) as well as integrating with partner security products.
This is another key announcement for us and we will be getting hands on it very soon!
Data centres need power and lots of it. Let’s not avoid the issue, just because you have moved to the Cloud, doesn’t mean that your applications aren’t consuming power.
As the world warms due to the climate emergency, we all need to do our part within our organisations and our personal lives to be part of the solution. AWS has been investing in sustainable source of power for a number of years and is only growing that commitment. As a result, it is aiming to be running off 80% renewable energy source by 2024, 100% by 2030 and to be net zero carbon across the Amazon business by 2040.
That just doesn’t mean running off clean power. It also means across transport, its building, its entire operations. Not only is it investing in clean power infrastructure, it is also looking at innovative transport solutions.
As advocates of clean air, clear soil and clean water, we congratulate AWS on this commitment!
Stay tuned for more from re:Invent
That might be it for the AWS re:Invent 2019 keynote wrap ups (see part one here), but the week has only just started. We will be sharing more updated over the next few days so stay tuned on your favourite channels!