Here we are halfway through February already (all be it a bit belated… again) for another AWS Weekly Wrap-up.

As always, This list is a summary of the recent releases and announcements made by AWS and is far from exhaustive. I simply try to summarise the dozens of weekly announcements to those most sought after by our customers. If your interested in taking a look at all of AWS’s recent announcements, you can follow the “What’s New” section of their website (available here). If you would like to talk to somebody about how you might be able to leverage some of these new technologies and services, please feel free to reach out using the contact link at the top of the page.

We’ve got a lot of smaller updates to go through this week across several different areas, including:

  • AWS Serverless Lens for the AWS Well-Architected Tool
  • Content filtering for Amazon EventBridge
  • Amazon Personalize can now use 10x more item attributes for recommendations
  • Amazon Lex announces support for Alphanumeric slot types
  • ECS has added support for canary deployments
  • VPC Flow Logs now support 1-minute Aggregation Intervals

AWS Serverless Lens for the AWS Well-Architected Tool

By now most people should be aware of the AWS Well-Architected Framework and the associated tool. These have been developed to help people ensure they are building their solution in line with AWS best practices. What a lot of people may not beware of is the Lens that focus on specific deployment scenarios, going deeper into the specifics of these types of deployments. Until recently there have only been to available, one for High-performance computing (available here) and one for the Internet of Things solutions (available here). Well, now we have a third… AwS have just released a Well-Architected lens focused on serverless deployments. While not a small read (73 pages) anybody designing, deploying or working with a serverless based solution should give it a read. Available here, it goes through some example designs, and walks through the issues and additional things you should be thinking about when building these types of solutions. you can also find the official announcement on the AWS website.

For those who may not be aware, Kloud also conducts Well-Architected Reviews for our customers. These are typically week-long engagements where we can work with your team to ensure you are leveraging best practices across your environment. Please feel free to reach out if you are interested in speaking to somebody about getting a Well-Architected review conducted on your workload.

Content filtering for Amazon EventBridge

Amazon EventBridge now provides additional content filtering options for your event-driven architecture. Not everybody will be familiar with EventBridge, but it’s a really useful service for those wanting to integrate SaaS applications back into their AWS environments. EventBridge provides a way to connect AWS Services, your own application and SaaS application together. Imagine you want to consume ZenDesk or PagerDuty events and trigger actions based on certain event types. An example might be that you may want to trigger a lambda function when a ZenDesk ticket is closed. While you could direct it to an API endpoint, streaming all your events to a single common endpoint simplifies the solution and reduces the integration effort required. However, until recently you needed to perform much of the filtering (is this an event I care about) within your application… reducing the overall effort savings.

With this new announcement, you are given a wide variety of matching patterns you can use to filter your event stream. Taken from the official announcement “The new functionality supports matching on numeric ranges, checking for the existence or absence of attributes, comparing string prefixes, and ‘anything-but’ comparisons for strings. Rules can include multiple filters to create complex combinations in a single event pattern”. By filtering traffic within the EventBridge we can simplify our downstream services and potentially reduce our operating costs.

We’ll be looking to do an upcoming article on EventBridge as well as an in-depth look on where each messaging/streaming service (EventBridge, Kinesis, SNS, SQS, CloudWatch Events) makes sense. In the meantime, if you are interested in EventBridge the official announcement can be found here.

Amazon Personalize can now use 10x more item attributes for recommendations

Amazon Personalize was originally announced at Re:Invent 2018 and released Jun last year. It’s a machine learning service that allows you to personalize a users experience within your application. The easiest way to think about it is… “Other products you might be interested in” feature on most e-commerce websites. What’s really nice about this service is that it makes getting started with Machine learning a very simple process. You can simply provide Personalize access to your dataset, provide the interactions between the data and your halfway home. You then need to create a solution which includes defining the model you wish to use and training it against your data.

Until recently you could only have 5 item attributes within your dataset for Amazon personalize to leverage. While this was enough for most tasks, it limited the insight that could be pulled from your data. With this announcement that limit has now been raised to 50 attributes which should be enough for almost any requirement.

For those looking to get started with Amazon personalize, it actually features in my upcoming series on building a customer experience platform so keep an eye out. In the meantime, you can check out the official announcement here.

Amazon Lex announces support for Alphanumeric slot types

Lex is getting a lot of love this year, particularly for us Australians. With it landing on our shores last month and then getting Connect integration, we are seeing a lot of interest in this service. One of the issues when working with language is all of the nuances that come with it. When somebody says “Dog one two three” we know that’s a sentence (all be it one that doesn’t make sense), but “123DOG” is most likely a license plate. Previously you could assist Lex in making the determination between the two by defining custom slot types and filling them with sample values. However, now that Amazon Lex supports Alphanumeric values this process becomes a whole heap easier. And the best news is that this new functionality is in Sydney already.

When writing this article I realised that I’ve never actually published an article on Lex but don’t worry, it’s featuring heavily in my upcoming customer experience tutorial series coming out soon (just like personalize).

ECS has added support for Canary Deployments

The one has a bunch of our team quite excited. We’ve always been able to perform blue/green deployments using ECS and CodeDeploy, but what happens when you want to migrates users across slowly? Well that what canary deployments are for, they give you the ability to gradually cut users across from the old environment (blue) to the new (green). This staged approach allows you to validate the health and performance of your new change while not putting 100% of your production workload at risk. While you do need to take this type of deployment into consideration when designing and building your application (can you handle some transactions being processed on the old platform and some transacting on the new) the benefits of this method are well worth it.

Darren Ball and Kia Fallahi have written a fantastic blog post on how you can leverage this new functionality in your ECS workload and I highly advise people to go and check it out (available here). You can also take a look at the official announcement here.

VPC Flow Logs now support 1-minute Aggregation Intervals

And finally, for this weeks wrap up we have shortly aggregation internals for VPC flow logs. While not something that’s going to be relevant to most people, for those with traffic monitoring requirements this greatly increases the granularity of your traffic flows. There are two major benefits from this shorter aggregation internal. Firstly, anybody who has ever scrolled through traffic logs before know that fewer logs are best… the smaller the haystack the easier life becomes. This reduction in total logs makes it easier and quicker to find what you’re looking for and identify potential issues. The second and slightly more obvious benefit is that with a smaller interval we get feedback and alerts quicker ensure faster response time when responding to events. It’s important to note however that to truly benefit from these faster times, you should already have automated event triggering via cloud watch or another monitoring tool. Not much benefit between 1 minute and 5 if your responses are human intervention… this is why automation is so important. For those interested, you can check out the announcement here.

And that will do it for our AWS update for Wednesday the 12nd of February 2020 (yes it’s being released on the 13th). Also, continue to keep an eye out for our continuing series on Amplify, and an updated series on Amazon Connect. We post updates every Wednesday as well as detailed tutorials and deep dives on products throughout the week. In the meantime, If you’d like to see something on the Kloud blog, please feel free to comment below.

Amazon Web Services, Cloud Infrastructure