

Picture a manufacturer that moved its ERP back end to Amazon EC2 three years ago. SQL Server runs on Windows Server instances in an Always On availability group. Two domain controllers sit in the same VPC and replicate with the sites still on premises. Email, Teams, and SharePoint live in Microsoft 365, and every sign-in runs through Entra ID. The Microsoft Unified Support renewal lands on the IT director’s desk in February.
The renewal quote assumes the support model the company bought when everything sat in its own data center. The environment no longer looks like that. So the practical question is which support arrangement fits the environment the team runs today, and who picks up the phone when something breaks between AWS and Microsoft.
This post walks through how I would evaluate a Microsoft Unified Support alternative for AWS customers, one workload at a time, before anyone signs a renewal.
When Microsoft workloads ran in your own data center, the support chain was short. The hardware vendor owned the box, Microsoft owned the software, and your team sat in the middle. On AWS, a third party owns the hypervisor, the storage, the network fabric, and in some cases the management layer of the database itself. A slow query might be a missing index. It might also be an EBS volume that ran out of burst credits. Your support model has to handle both answers.
This kind of mixed environment is now normal. Flexera’s 2026 State of the Cloud survey of 753 cloud decision-makers found that 73% of organizations run hybrid estates, up three points from the prior year. In the same survey, 83% of respondents run some or significant workloads in AWS, and 79% do the same in Azure. Flexera also observed that many organizations end up with complex environments “by happenstance rather than by design.” An acquisition brings in a second cloud. A department buys a SaaS tool. A few years later, the renewal team inherits the result.
Those figures describe cloud adoption broadly. They don’t tell you what your Microsoft dependencies look like on AWS, and no survey can. One AWS customer runs a handful of Windows file servers. Another runs a large SQL Server estate, Exchange Server on EC2, and hybrid identity through Entra Connect. Both would describe themselves as “on AWS,” and they need very different support coverage.
That is why an AWS-centered environment deserves a workload-specific renewal review. The infrastructure decision and the Microsoft support decision were probably made by different teams in different years. They need to be evaluated together now.
Start with an inventory. A spreadsheet is fine. For each Microsoft dependency, record five things:
Cover the obvious ones first: Windows Server, SQL Server, and Active Directory. Then keep going. List the .NET applications and the IIS servers that host them. List Microsoft 365 and Entra ID. Look for the integrations people forget, such as an Entra Connect sync server running on EC2, an SSIS package that pulls from Amazon RDS, or a file share on Amazon FSx for Windows File Server that a line-of-business application quietly depends on.
Pay close attention to where SQL Server runs. Self-managed SQL Server on EC2 means your team, or your support provider, owns the operating system, the SQL Server instance, patching, backups, and high availability. Amazon RDS for SQL Server shifts host maintenance and much of the backup work to AWS, and in exchange it limits what you can change at the instance level. Those are different support problems. A deadlock investigation looks about the same in both places. A Windows Server Failover Cluster that won’t come online is your problem on EC2, while the Multi-AZ failover mechanism on RDS belongs to AWS.
The licensing column matters more than it looks. As the next section shows, AWS treats License Included and BYOL software differently when it comes to support. Some products only come one way. AWS’s FAQ states that it sells only Windows Server and SQL Server licenses for use on EC2, so Exchange Server, SharePoint Server, and other Microsoft server products running on EC2 are customer-licensed by definition.
AWS does support Microsoft software, and it has for a long time. The AWS and Microsoft FAQ says AWS Support has supported Windows-based EC2 instances since 2008 and that AWS holds a Premier Support agreement with Microsoft. For customers on the Business or Enterprise support tier, AWS Support can work directly with Microsoft support engineers on issues involving Windows Server, SQL Server, or Windows desktop through Amazon WorkSpaces. AWS also says there are no regional restrictions on that benefit.
That is real coverage, and it has boundaries. The FAQ is specific about four of them:
There is one more wrinkle. The FAQ still refers to the “Business or Enterprise tier,” but AWS has restructured its plans. According to the AWS News Blog, customers on Developer Support, Business Support (classic), and Enterprise On-Ramp keep their current level of support through January 1, 2027. The current lineup is Business Support+, Enterprise Support, and Unified Operations. If your Microsoft coverage depends on your AWS tier, get written confirmation that the Microsoft escalation path carries over to the plan you will be on after that date.
With those rules in hand, separate four kinds of incidents, because each one routes differently. Use the worksheet below with your actual plan, not the plan you think you have.

That last row deserves attention. The AWS FAQ discusses Microsoft 365 only in the context of licensing Microsoft 365 Apps on WorkSpaces. It says nothing about support for Exchange Online, Teams, SharePoint Online, or Entra ID. For most enterprises those services carry more daily business risk than any single EC2 instance.
At 9:40 on a Monday morning, users of an order-management application hosted on EC2 start reporting timeouts. The application log is full of SQL timeout exceptions. Some users also can’t sign in, while others can. The help desk opens a ticket, and four teams start looking.
That is two unrelated problems and one outage. The storage issue causes the timeouts, and the time drift causes the sign-in failures. Each team can prove its own piece. Without one owner, each team also has a reason to wait for the others to finish.
Somebody has to own four jobs from the first minute:
This kind of tangle is becoming the norm. In its 2026 Annual Outage Analysis, Uptime Institute expects failures to increasingly come from “complex interactions between systems, including software, networks, and external dependencies.” The cost is real. In Uptime’s 2025 annual survey, 57% of respondents said their most recent major outage cost more than $100,000, and one in five said it cost more than $1 million. Those figures cover IT and data center outages broadly, not Microsoft workloads on AWS. They still show what an hour of finger-pointing can cost.
The ownership matrix below is the in-text visual for this section. Fill it in for your own top five workloads before the renewal conversation.

Add two items to your runbook now. First, pre-approve AWS to share case details with Microsoft, since AWS may ask for documented authorization before it does, and that should not hold up a Sev 1 at 2 a.m. Second, write down who can open a severity 1 case with each vendor, along with account numbers and contract IDs.
US Cloud supports Microsoft workloads and Microsoft–AWS integrations. Its engineers use the same support practices for Microsoft software hosted on AWS as they do on Azure. Customers keep their Microsoft licensing when they switch to the partner-delivered support model. For an AWS buyer, the first step is to identify which part of each workload the provider can troubleshoot.
Support tiers come up in every one of these conversations, and terminology varies by provider, so ask each one to define its terms. A common breakdown looks like this:
The question for any provider is how fast a severity 1 reaches L3, and what happens when it needs L4. Test US Cloud, or any Unified Support replacement, against these requirements:
Then ask the provider to demonstrate the path. Hand over the hypothetical scenario from the previous section and ask for a walk-through: who answers, what evidence they request, when they tell you to open the AWS case, and when they bring in Microsoft. A credible provider will tell you which parts it owns, which parts belong to AWS, and how it keeps the case moving across both. A provider that claims to own all of it, EBS included, is overpromising.
When Windows 11 desktops in Amazon WorkSpaces stopped detecting monthly and feature updates, US Cloud engineers traced the problem through Configuration Manager (SCCM) logs, Windows Update settings, and Group Policy. They identified the policy blocking updates and guided the customer through a correction, service restarts, and fresh scans. On October 22, 2026, the customer confirmed successful tests on several devices and asked to close the case so its team could carry the fix forward.
A Microsoft Defender for Cloud case took a different path. When onboarding AWS accounts failed across several regions, US Cloud investigated the CloudFormation failures and coordinated escalation to Microsoft. The customer found a workaround that enabled onboarding, and Microsoft confirmed a design limitation.
Some teams need design advice before an incident occurs. For a customer planning how Microsoft Active Directory Certificate Services and AWS Private CA would coexist, US Cloud advised on certificate trust, domain-controller certificates, and how to validate authentication in stages. The engagement covered migration planning.
Yes, if documented coverage fits your workloads. AWS states that its expanded agreement with Microsoft does not replace a direct Microsoft support agreement, so AWS Support alone won’t cover issues inside your Microsoft applications that are separate from AWS services. You still need a Microsoft support relationship. That can be Unified or an independent provider, as long as the provider’s written scope covers what you run. US Cloud supports the Microsoft software in AWS-hosted workloads and can escalate Microsoft product issues. Your AWS contract covers the route for underlying AWS service faults.
Yes, within defined limits. Microsoft products bought through AWS as License Included are covered under your AWS Support agreement. For BYOL software, AWS focuses on the AWS-service side of the problem. Products past Microsoft’s end-of-support dates are excluded from the joint AWS and Microsoft offering. The AWS FAQ does not cover support for Microsoft SaaS services such as Exchange Online or Entra ID.
It depends on the route. If the case starts with AWS and you are on a qualifying support tier, AWS Support can file a case with Microsoft when it finds the problem is due to a Microsoft product or driver, with your permission. Under Unified, you open the case with Microsoft. With an independent provider, the provider escalates under its contract. US Cloud can bring a Microsoft product issue to Microsoft directly. If the underlying fault is in AWS, the customer opens the AWS case and US Cloud can coordinate with both vendors. Whatever the route, confirm who owns the case after it reaches Microsoft.
Verify six things in writing: supported products and versions on AWS, which AWS deployment types are in scope, exclusions, the Microsoft escalation route and who owns the case afterward, response and severity commitments, and the access model for your AWS environment. Then confirm your AWS plan after January 1, 2027, and check whether any workloads run Windows Server 2016 or other products near end of support.
Support cost cannot be evaluated apart from support performance. If a lower-cost option can’t protect the workloads and escalation paths you mapped, it isn’t a better option. Once coverage is confirmed, cost comparison gets much simpler, because you are comparing equivalent scope.
Finance teams increasingly review technology spend as one portfolio. The FinOps Foundation’s State of FinOps 2026 reports that 90% of practitioners manage SaaS spend or plan to in the coming year, 64% manage licensing, and an emerging 28% are beginning to include labor costs. Support contracts belong in the same review. That is my application of the research; the survey didn’t measure support contracts. Flexera’s data points the same direction: 85% of its respondents call managing cloud costs their top challenge, and 78% rank managing software licenses third.
Build the comparison line by line:
| Cost line | Current arrangement | Proposed arrangement | Cash or labor |
|---|---|---|---|
| Microsoft Unified Support fee | Cash | ||
| Independent Microsoft support fee | Nicht zutreffend | Cash | |
| Retained AWS Support plan (after January 1, 2027) | Cash | ||
| Additional Microsoft escalation or per-incident costs | Cash | ||
| Transition work: onboarding, access setup, runbook updates | Cash and labor | ||
| Internal coordination time during incidents (estimated hours) | Labor |
A few rules keep the comparison honest:
Contract terms count as part of cost. Over nine years of tracking, Uptime found that third-party providers, including cloud and internet giants, telecommunications firms, and colocation companies, accounted for about two-thirds of publicly reported outages. Uptime says that shift may still require adaptation at a contractual and financial level. For a support buyer, that means response and escalation commitments belong in the signed agreement, where you can enforce them.
Work backward from two dates: your Unified renewal and your budget lock. If the renewal falls in the first quarter, both January dates above land in the middle of your evaluation. Run the five steps below in order.
Open cases trip up more transitions than anything else. Decide before cutover whether each open Microsoft case will be closed out under the old agreement or handed to the new provider with full history. Ask the onboarding team to document the handoff for each case before you sign.
The fastest way to start is the ownership matrix. Fill it in for your five most critical Microsoft workloads on AWS, and the gaps usually show up in the first hour. If you want a second set of eyes, US Cloud can review the matrix against your current coverage and walk through how a severity 1 case would move in your environment.
