April 20, 2022
Welcome to the seventh edition of the Azure World Newsletter in 2022.
For the last week or two, I’ve been working on a complete re-filming of my AZ-900 course. And before that, I was intensely working on creating some labs. And I must say, I really can appreciate all of the subtle (and not so subtle) improvements that Microsoft has made to Azure over the last year.
When I am in the Virtual Machine creation tool, I notice some new options here and there. In small ways, the tool has improved. Same for storage accounts, web apps, and application gateways. I’m sure there are 10 changes that I didn’t notice for every change that I did notice.
Shout out to the team responsible for the Azure Portal at Microsoft. It gets better and better, and I can’t think of any part of the UI that feels like it’s gotten worse. For me, at least. Keep up the good work.
Thanks so much for being a subscriber! The unsubscribe link is at the bottom if you want to stop receiving these emails.
ONE.
A couple of weeks ago, Microsoft rolled out a new Azure Front Door service and a new Azure CDN service. The existing services have been renamed “Classic”.
The Azure Front Door service was first introduced only a few years ago in 2019. Its utility was clear from the start. It’s a combination of a load balancer, web application firewall (WAF), content delivery network (CDN), and caching service that operates at a global scale.
If you were hosting your web applications in several regions, putting Front Door “in front” of them made great sense. You could increase your applications’ availability, performance, and security while reducing the resources used on each server (potentially saving you money).
As can happen, Microsoft has been hard at work over these last couple of years working on the V2 of this service, and at the end of March, they made it available to customers.
Some features include new DNS options to validate domain names, new DevOps friendly command-line tools, new analytics capabilities, and pushing more application logic to the “edge” closer to the clients. So if you need to route users to a specific set of servers, that decision can be made closer to them geographically instead of being centralized.
The new services also include a simpler cost model, making it easier to understand the costs of using Azure Front Door and Azure CDN.
More details can be found at this blog link.
TWO.
One of the problems that you don’t really think about with Machine Learning is the amount of duplicated effort across teams when manipulating the data that comes into the model. If you have a large enough organization, the chances are high that one team will be spending significant hours developing a feature to add to a dataset that another team has already developed.
If you’re not aware, a feature in Machine Learning is like the column of a data table. Sometimes, the columns from inbound data are not used as-is and need to be transformed and aggregated for a specific purpose.
So imagine that one of the ML teams in your organization has done significant work on an inbound data source to make it more valuable in terms of the usefulness to machine learning models. How do they share that work broadly? How do other developers even know that this improved data source even exists?
The answer is “feature store”. This is effectively an abstraction layer between a machine learning model and the data. Like a “data layer” in an old n-tier architecture diagram. It can standardize how data is accessed and can even be used to make small data transformations, like changing data formats into a common format.
LinkedIn uses a feature store called Feathr, and Microsoft is now releasing this as an open-source tool on GitHub. Now, more organizations can save time and money by using one team’s improvements to their data features across all teams.
Source: https://azure.microsoft.com/en-us/blog/feathr-linkedin-s-feature-store-is-now-available-on-azure/
AZURE PLATFORM UPDATES.
The following announcements were made in the last two weeks:
- New Ebs/Ebds v5 series VMs have 300% faster storage throughput
- Capacity reservation support for AKS, in preview
- Azure Backup for Blobs has metrics and metric alerts, in preview
- Storage Tables now support Azure AD authentication and RBAC
- Azure Backup for VMs can now backup to archive tier storage
- Azure AD Graph retirement date pushed to December 31, 2022
- Azure Databricks now supports Delta Live Tables
- User-Defined Routing (UDR) now supports service tags
- New recommended alert rules for Azure VMs, in preview
- Azure Monitor can monitor activity logs for changes to Azure resources and resource groups in a subscription, in preview
- Azure Monitor has a new result set grid layout
Check out the Azure Updates page if any of these affect you.
https://azure.microsoft.com/en-us/updates/
COMING UP FOR ME.
I’ve been finishing up the recording for the new AZ-900 course on Udemy. Completely recorded from scratch and based on the new requirements that go into effect on May 5. If you plan to take the exam in May or later, my course will be up-to-date for you. Grab it here… https://sjd.ca/az900
WHERE TO FIND ME.
And that’s it for issue 3.07. Thanks for reading this far. Talk to you again in two weeks.
What is your favorite platform to be on? Perhaps we can connect there.
Facebook Page: https://www.facebook.com/getcloudskills/
LinkedIn: https://www.linkedin.com/in/scottjduffy/
Instagram: https://www.instagram.com/getcloudskills.ca/
Twitter: https://twitter.com/scottjduffy
Udemy: https://www.udemy.com/user/scottduffy2/
LinkedIn Learning: https://www.lynda.com/Scott-Duffy/1993589682-1.html