Everything You Need to Know About Azure Infrastructure – August 2020 Edition
This monthly update comes to you from COVID-19 lockdown central in Ireland. The county that I live in has been “locked down” for the last 3 weeks after isolated clusters in several factories were identified. Remote working has never been more important – and the cloud plays a big role in that. So, let’s see what Microsoft Azure did to improve things for us in the last month.
How do you find out about an outage (service incident) in Azure? For some, stuff doesn’t work right and you check the Azure status page. For others, they go onto some social media platform and observe their uneducated friends crowing “this is why I will never do cloud” (and become redundant). Microsoft wants to improve that experience for Azure customers using 5 pillars:
- Speed: The goal is to inform customers within 15 minutes of an outage. As you can imagine, with something as large and complex as Azure, there are frequent issues, most of which only affect a few customers. Confirming that there is an issue and then communicating that incident to a customer can take time. Microsoft is using artificial intelligence (“AIOps”) to detect, engage and mitigate outages. In the last quarter, AIOps handled over half of the outage issues. A core change is that issues will be communicated through Azure Service Health in the Azure Portal – issues unique to the customer will be presented privately here through the authenticated Portal.
- Granularity: By using the Azure Portal, incident information for specific resources can be shared privately with the customer. Customers can create general or specific alerts triggering actions such as an email or even a Logic App or Function.
- Discoverability: Service Health alerts can be viewed in the Azure Portal, but not all necessary people will have access to the Portal. You can (and this is recommended) set up alerts to inform all necessary people directly or via third-party systems.
- Parity: All people should know that messaging for all kinds of events affecting platform availability will be in Service Health; this includes outages and other events such as planned maintenance. Some organizations use Azure DevOps and those engineers might rarely use the Portal. Therefore, an Azure DevOps status page has been set up. Microsoft is also working on making the Service Health experience consistent across different clouds, such as Azure, M365, and Power Platform.
- Transparency: In my experience, being open and honest about an outage and the subsequent investigation, what Microsoft calls the root cause analysis or post-incident report (PIR), earns trust with the customer (internal or external). Microsoft is working on improving the process, but they must find the right balance between fact and speculation because some organizations will make mid-incident decisions based on these communications. PIRs for major incidents are shared within three days. PIRs for smaller incidents can be requested through Service Health in the Azure Portal.
As you can see Service Health is the place to go – allegedly – during/after an incident.
A Big Month for Azure Backup
Several announcements were made by the Azure Backup team:
- Configure backup for Azure file shares directly from the file share blade
- Selective disks backup for Azure Virtual Machine is in public preview
- Encryption of backup data using customer managed keys is in public preview
- Improvements to Azure VM backups are now available
The announcement that will affect the fewest of us is this one: Azure Backup for SAP HANA databases running on RHEL is now generally available. In my entire career, I have one 1 employer that use SAP (and it was a pain in the you-know-where). This story reminds me of the weekly meetings I used to have with Microsoft as a managed partner. They’d come in, we’d talk about sales pipes, what we were doing to develop the business, and then they’d ask us “how about SAP on Azure – is there any work there”. And every week I would respond with no – and there won’t be. But they kept on asking me the same question, week after week.
SAP on Azure is like those fancy high-end brand stores in the airport. You know the ones – they’re in a high foot-fall location, there are 2 well-dressed sales assistants, lots of pricey items with no price tags, and no customers. But week after week, month after month, year after year, that store is still there paying rent to the airport. Why? Because it only takes 1 customer every once in a while to make a profit.
Backing up SAP on RHEL on Azure sounds like it will affect a tiny percentage of customers. But those few customers will consume a LOT of storage and fund further improvements of Azure Backup. And that makes a “I don’t care” feature like this something that those of us who are using Azure Backup should care about.
Other Announcements from Microsoft
Here are other Azure IaaS headlines from the past month:
- SQL Server FCI on shared disks for SQL Server on virtual machines
- Backup storage cost savings for Azure SQL Database and Managed Instance
- Configurable backup storage redundancy option for Azure SQL Managed Instance
- New GPU NCas T4 v3 VMs are in public preview
- Release of new Azure CDN (Microsoft Standard) capability
- Update your Traffic Analytics queries by 31 August 2022
- Azure Security Center – News and updates for July 2020
- Azure Policy compliance scan action for GitHub workflows is in public preview
- Audit Logs of Azure Monitor logs queries now available
- Line numbers in Log Analytics query editor
- Azure Cloud Shell can now run in an isolated virtual network (public preview)
- PowerShell support in Durable Functions is in public preview
- Update rollup 49 for Azure Site Recovery
- GitHub Actions integration is now Generally Available
- AzCopy v10.6 released with updated sync features and larger blob size support
And Now for Something Different
I had my crazy times with phones when I paid for a Lumia 1020. Now I am a safe and boring person with an iPhone. I was intrigued by the Surface Duo, but I will not buy it. What I want to focus on is apps – not the apps running on the phone but the apps running in Azure that are being used on the phone.
When Microsoft announced the availability of the Duo there was a comment that it would pair up nicely with Windows Virtual Desktop. Some in the media wondered how that could ever work – how could you use a desktop on a phone screen? The answer is: you wouldn’t.
Remote Desktop Services (RDS) has long allowed you to publish an app. This means that the app runs on the server but appears in its own window on the client device. If that app isn’t too big, it might run just perfectly fine on the Duo using the Android Remote Desktop app, complete with touch and gesture support. In the sales or field first-line worker world, a stylus might be useful. But that all assumes that there is a signal, but that’s a conversation best left for another time.
Will the Duo be a success? I would not touch the Duo with a barge pole if I was an enterprise IT admin. It reminds me too much of tech like the Band and other long-dead devices from Microsoft. I am not convinced that Duo will be a supported brand in 3 years. Think of those airlines that used Lumias for in-cabin payments. What are they using now? Yup – iPhones. I think that Microsoft will have tough resistance with the Duo, and don’t be surprised to hear of a partnership with Dell or HPE to try to get some traction with large customers (at huge discounts).
I don’t want to be a total Debbie Downer. I hope that I am wrong and that the Duo evolves into something new, that I don’t envision, just like the Surface Pro did.