Microsoft Workloads on AWS

Year in review 2023: AWS for Microsoft workloads

Celebrating 15 years of running Microsoft workloads in the cloud, longer than any other major cloud provider

Before we wrap up 2023, I wanted to capture some of the highlights of the year—so I reached out to my esteemed colleague, Rajesh Rathod. Rajesh is an Amazon Elastic Cloud Compute (Amazon EC2) product leader who has been instrumental in the product direction, launches, and events relevant to customers running their Microsoft workloads on Amazon Web Services (AWS).

In the following interview, I ask Rajesh about: the # of Microsoft-related launches this year; a few key launches that excited our customers; our 15-year anniversary; trends and shifts he’s seeing with our customers; how to get started on AWS; and, of course, generative AI.

Let’s get to it.

Rajesh, please introduce yourself. How long have you been with AWS and what is your current role?

Hi there. I’ve been here at AWS for five and a half years, and I’m a product leader for Amazon EC2 commercial software services. My team builds products for Amazon EC2 dedicated compute, SQL Server workloads running on Amazon EC2, the management of commercial software licenses, and the migration of commercial software workloads to AWS.

This has been a busy year for your team. How many launches did your team release to help customers who are running Microsoft workloads on AWS?

My team covers many different products. For the past few years, I’ve been keeping track of how many major product features and new products we’ve launched. And for this year, I found that we launched 108 product features. That’s pretty exciting! And we built these features to help our customers assess, migrate, manage, and modernize their Microsoft workloads on AWS.

That’s fantastic! I like your point about building for our customers. We often talk about the fact that 90% of what AWS builds is driven by customers. So, I have to ask: Out of those 108 features, how many of them were actually driven by customer requests?

I would have to say the majority of them. Yes, more than 90% of them were driven by customers. Would you like me to give you a few examples?

Of course!

Great. One example is from our customers running Microsoft workloads on Amazon EC2 Dedicated Hosts. That’s our dedicated compute product. It allows them to bring their own licenses on a physical server fully dedicated for their use.

Over the past few years, customers have been asking us to provide a capability where they can mix different instance sizes on the same host. For example, let’s say I’m running an r6i Dedicated Host. I might have a few instances that are small and maybe a few that are large and extra-large—and I want to fit all of them onto this one Dedicated Host so that I can make the most optimal use of it.

This was one feature I would say almost all our Dedicated Host customers needed. It was important for them because when they use any licensed software for a workload on their Dedicated Host, they have to always license the entire host because all CPUs on that host need to be licensed. So, if they are not able to optimally use their Dedicated Host, there is a lot of capacity lying vacant and it’s expensive for them because of the wasted licenses.

By providing this feature (i.e., mix different instance sizes on the same host), one of our customers told us this year that they could save $6 million USD by just using this one feature for their products.

Wow! That’s significant savings. So, let me ask this: Before this feature was available, did this customer have a lot of Dedicated Hosts they couldn’t consolidate?

Yes. The customer had a lot of older generation Dedicated Hosts that they were not willing to upgrade as there was no compelling benefit for them. Once we launched this feature in the latest generation of Dedicated Hosts, they were thrilled to upgrade all their infrastructure, reduce the number of hosts they actually needed, and ultimately, save a lot on licensing costs.

You said you had another example?

Yes, another good example is the launch of Amazon Relational Database Service (Amazon RDS) Custom for SQL Server. There was a huge demand for this product by customers who needed a managed SQL Server service for applications that require customization of the underlying operating system and database environment. RDS Custom allows customers to use their own SQL Server installation media when creating a managed database instance. By using Bring Your Own Media (BYOM), customers can also leverage their existing SQL Server licenses with Amazon RDS Custom for SQL Server.

Based on customer feedback, we have launched many other features, such as increased pod density limits for Windows containers, flexibility to scale IOPS separately from storage on Amazon FSx for Windows File Server, and self-managed Active Directory support for Microsoft Office on Amazon EC2.

So what other launches were customers most excited about?

While we align our roadmap with what customers need, there are a few features that, depending on the customer use case, have an outsized impact on their business. One such feature that we launched this year was licensing optimization using AWS Compute Optimizer. And we started with SQL Server licensing optimization.

The way it works is AWS Compute Optimizer provides proactive recommendations to customers on where they can save on their licensing costs and how they can optimize their Amazon EC2 instances or the SQL Server editions they are using. So, without compromising the performance and availability, Compute Optimizer shows customers how they can save on costs.

This service is a very good example of our customer obsession. Our customers love it because we have always prioritized reducing prices and reducing costs for our customers. And with this one feature this year, we were able to provide proactive recommendations to customers and collectively help them save $500 million USD on AWS! We’re super excited and committed to helping customers reduce their spend on both unused infrastructure and commercial software licenses.

So, let’s look at the other 10% of features that our customers don’t necessarily ask for, but we develop them so that we can innovate on their behalf and delight them. What “10 percent” features did you release this year?

Yes, the 10%. So, one thing we always look at in our customer conversations is what pain points our customers are currently experiencing. They might not be asking us to develop a feature or build a new service, but it’s important that we identify their pain points and try to address them.

One of these 10% features I want to talk about here is the artificial intelligence-powered recommendation engine we built for breaking monolith applications into microservices. Our AWS Microservice Extractor for .NET team did an amazing job of envisioning, building, and launching this feature. Breaking up monoliths and identifying potential candidates of microservices is a laborious and error-prone task when it is done manually.

The team decided to use artificial intelligence with a machine learning model. The new AI-powered recommendations engine uses a machine learning model to scan the source code in your project. Once Microservice Extractor finishes the analysis, the tool will automatically group your classes to form candidates for a new microservice. This feature improves developer productivity and helps our customers modernize their applications because those monoliths are hard to maintain and iterate for innovation. This is a good example of invent and simplify to delight our customers.

This was also a big year for us to celebrate. We have been running Microsoft workloads in the cloud for 15 years, longer than any other major cloud provider. Off the top of your head, do you know how many Microsoft-focused services and features we’ve launched since 2008?

I’ve actually lost count of the number of features we’ve launched. But I do know that we have launched more than 25 products to help our customers with Microsoft workloads who want to either build new applications on AWS or migrate their existing applications and modernize once they migrate. We’ve continued to iterate on these products by launching many, many features to support our customers over the past 15 years.

But what is even more interesting is the experience that we have gained in these 15 years working with our customers. We understand what they really need for running their mission-critical Microsoft applications on AWS. Our customers have trusted AWS for all these years. We have thousands of customers who started running their Microsoft applications on AWS in 2008, long before other cloud platforms were even launched. And they are still running and scaling those applications with AWS today. Through this experience, we have innovated the products that we provide to our customers. And they get a lot of value by using those products.

But it’s not just the products we have developed. We have also built a strong partner network. As AWS started supporting Microsoft workloads in the cloud, a lot of our partners were also building products to help customers. And now our customers are using those as well. For example, there’s CloudEndure, which AWS acquired in 2019. They were early in identifying what customers needed for migrating to AWS and built a product. There are many other partner products that customers use to assess, migrate, manage, and modernize Microsoft workloads on AWS.

What trends or shifts have you seen with customers over the years, especially as the cloud has continued to mature and evolve?

I’m seeing three trends in the migration of Microsoft workloads, although many of these apply to other workloads as well. First is around the momentum of migrations. Customers still have a lot of workloads running on premises, so they are constantly migrating. But over the past couple of years, I have seen most of our large customers migrate more workloads to the cloud or have plans to accelerate their pace of migration. They are actively planning, starting the migrations, training their teams, and building operational muscle for the cloud. So that’s the momentum I’m seeing—and it’s strong for migrations.

The second shift I’ve seen is in customer confidence. In the early days, when cloud computing was new, customers were starting with less-critical applications, like their dev/test workloads, or using the cloud as a backup solution. Now we’re seeing customers run their mission-critical workloads on AWS. For example, Nasdaq has completed migration of the first U.S. Options Market to AWS. They use AWS Outposts and Amazon EC2 Dedicated Hosts to deliver a low double-digit microseconds latency for end-to-end and order-to-trade processes, in addition to a 10% performance improvement in round trip latency. Then there’s Capital One and Netflix who are running their cloud workloads here and have virtually exited all their data centers. We also have major airlines, vaccine makers, and organizations who are dealing with very critical business functions. They are all using AWS.

Which leads me to a third trend: migration methodology. Early on, many customers were mainly interested in vacating their data centers—and just doing a lift-and-shift of their applications. Now I’m seeing two patterns. First, customers are doing a thorough assessment of their workloads. They’re identifying which applications are best suited for lift-and-shift, which applications can be easily replatformed, and which applications can be refactored to run with cloud native services. There’s a lot of replatforming and modernization happening along with migrations.

And the second pattern I’m seeing is around what triggers the need to migrate from on premises or other cloud providers to AWS. Initially, customers were thinking about cost savings and scalability, which still remain relevant and useful triggers. But now I see more customers planning cloud migrations to improve their staff productivity, security posture, and the pace of innovation to meet the needs of their end customers. It’s about driving business agility. And those are the reasons why I see more customers migrating their Microsoft workloads to AWS.

You mentioned that you’re seeing more customers do a thorough assessment of their workloads. As you know, AWS and our partners have a lot of assessments to help our customers. Where should a customer begin? Is there a particular assessment you would recommend here?

We advise our customers to always do the Optimization and Licensing Assessment (OLA) before they start large migrations to AWS. The OLA helps in three different ways. First, it helps you identify what your target environment could look like—like what type of Amazon EC2 instance or storage or licensing you should use. This helps in making your migration plans more predictable. Second, it provides you the most optimal Total Cost of Ownership (TCO) for running those workloads. And third, it gives you the “art of the possible” in terms of what you can modernize before migrating, while migrating, or after migrating.

So, from that perspective, the OLA is a default option that most customers with Microsoft workloads should use. And there are many tooling options that AWS provides. We have our own first-party tooling, and our partners have their own tools to do this assessment as well.

If a customer is new to AWS, getting an Optimization and Licensing Assessment (OLA) sounds like a solid first step. How about existing customers? Should they do an OLA? What else do you recommend for those who have been using AWS?

Yes, the OLA is good for both new and existing customers. And once you are using AWS, modernization is going to give you the best return on investment. We recommend using the AWS Migration Hub Strategy Recommendations service, as well as AWS Compute Optimizer. Compute Optimizer is going to give you more optimization recommendations while running the same software stack, like SQL Server and Windows Server. It helps answer the question: How can you minimize your spend and optimize your workload?

With our Migration Hub Strategy Recommendations product, you’ll be given many options, including modernization—like whether you can move to containers, use serverless, migrate to an Amazon Aurora database, or transform your .NET applications to run on Linux. We continue to add new strategy recommendations to the Migration Hub.

Switching gears here. I would be remiss if I didn’t ask about generative AI—specifically as it applies to Microsoft workloads.

Ha! You cannot complete any conversation without generative AI these days. But, honestly, generative AI and machine learning, in general, are touching all aspects of business, both for our customers and for AWS. Our customers have started using it, and Amazon has been using machine learning for more than 25 years.

But for Microsoft workloads, in particular, I’m excited about the recent re:Invent announcement of Amazon Q and how it will help customers replatform their .NET applications to run on Linux. Customers who have built their .NET applications with .NET Framework are stuck running those applications only on Windows Server. And Windows Server licenses are expensive and restrictive. These customers actually want to run their .NET applications on Linux operating systems, which is, as you know, more innovative and less expensive. To do this, they need to refactor their applications from .NET Framework to .NET 6+. And that refactoring part can take a while if you do it manually.

But now with generative AI, Amazon Q is going to radically improve developer productivity for refactoring a .NET Framework-based application to .NET 6+. And once it is on .NET 6+, you can run it on Linux. You can even use AWS Graviton processors for additional cost savings and better price performance. So that’s one thing I’m super excited about.

Last question. Can you share anything about what’s on your team’s roadmap for 2024?

Well, all I can say is that we have our plates full with a lot of features we are developing to make it easier for customers and make AWS the best place for running Microsoft workloads. While I won’t ruin the surprise of what those launches are, I can definitely say that direct customer feedback keeps influencing our roadmap to build more than 90% of those features over the next few years.

AWS has significantly more services, and more features within those services, than any other cloud provider, making it faster, easier, and more cost effective to move your existing applications to the cloud and build nearly anything you can imagine. Give your Microsoft applications the infrastructure they need to drive the business outcomes you want. Visit our .NET on AWS and AWS Database blogs for additional guidance and options for your Microsoft workloads. Contact us to start your migration and modernization journey today.

Tamara Dull

Tamara Dull

Tamara Dull is a principal product marketing manager at Amazon Web Services based in Los Angeles. She focuses on the migration, optimization, and modernization of Microsoft workloads running on AWS. In non-tech mode, you’ll find Tamara helping shelter dogs find their forever homes with Outta the Cage, an L.A.-based nonprofit she cofounded in 2016.