Infrastructure usually stays out of the advertising conversation until something stops working. Then it becomes a question about campaign delivery, advertiser commitments, and revenue.

On November 12, 2026, Microsoft 365 experienced widespread availability problems. Users observed connection failures, timeouts, and server errors affecting Outlook and other Microsoft 365 services.

That disruption does not tell us how any particular ad platform performed. It does provide a timely reason to examine the dependencies behind business-critical services, including the technology delivering your advertising.

The October 2025 AWS incident offered another reminder. In its post-event summary, AWS described a DynamoDB DNS failure in Northern Virginia that affected services dependent on it.

For publishers and retail media operators, the useful question is what happens to live campaigns when part of the infrastructure fails.

From AdButler’s perspective, your vendor should be able to explain that clearly. Here are five questions to bring to the conversation.

At a glance: What to ask your ad tech vendor

• Who hosts the platform? Understand the providers and shared dependencies supporting delivery.
• Where do ad decisions happen? Distinguish regional serving capacity from creative distribution.
• What survives a dashboard outage? Establish whether existing campaigns can continue and which operations become unavailable.
• How does failover work? Ask about detection, traffic routing, recovery time, and testing.
• What does uptime measure? Review service definitions, historical results, and incident communication.

1. Who hosts your infrastructure, and where are you dependent on a single provider?

Most hosted ad tech platforms rely on external infrastructure somewhere in their stack. Knowing the provider’s name is a starting point. Understanding its role tells you much more.

Ask where ad decisioning, campaign configuration, reporting, creative hosting, and DNS operate. Then ask whether those components share a dependency that could interrupt several services at once.

Using one cloud provider does not automatically mean a regional incident will take down the entire platform. Equally, listing several providers does not establish that live traffic can move between them. What matters is how workloads are distributed and how failures are handled.

Ask for: A current dependency map showing which services can continue operating if a provider or region becomes unavailable. AdButler has its investment in its own serving cloud and proprietary CDN, with a 90+ global points of presence.

Our data processing agreement also identifies multiple infrastructure providers, including OVHCloud, HighVelocity, and AWS. These disclosures provide a starting point for discussing current workloads and dependencies.

Architectural independence means having meaningful control over infrastructure choices and failure handling. It should be visible in the design and supported by operational evidence.

2. Are your ad-serving engines distributed across regions, or only your creative files?

“Global infrastructure” can describe different capabilities.

A content delivery network distributes assets such as images and videos. The system deciding which advertisement to serve may operate elsewhere. If that decisioning service becomes unavailable, cached creative files alone cannot complete every ad request.

For a publisher with international readers or a retailer operating across markets, both performance and regional resilience matter. Ask where requests are processed and whether healthy regions have enough capacity to absorb redirected traffic.

Ask for: Separate explanations of ad decisioning and creative delivery, including what happens when a serving location is degraded.

AdButler’s ad delivery specifications include sub-100ms response times and a CDN with 90+ global points of presence for distributing creative assets. Our public status page separately monitors regional ad-serving components, including East, West, Central, EU, and APAC.

For publishers and retailers, speed matters alongside availability. An ad response that arrives after a page, app, or auction timeout can miss the opportunity to deliver.

Ask vendors how they measure latency, including performance by region and during peak traffic.

3. What happens to live campaigns if the management dashboard goes down?

Your team’s ability to log in and your audience’s ability to receive an advertisement are different functions.

A resilient design should limit how a management or reporting problem affects live delivery. Ask whether the serving systems retain the campaign configuration they need to continue operating when another component becomes unavailable.

The answer needs operational detail. Existing campaigns might continue while new changes are delayed. Reporting might arrive later. Your team needs to understand what remains available and how normal operations resume.

Ask for: A walkthrough covering existing campaigns, scheduled changes, pacing, event collection, and reporting recovery during a dashboard outage.

AdButler’s public monitoring distinguishes dashboard availability from regional ad-serving availability. That gives customers component-level visibility rather than a single platform-wide indicator.

It also provides a useful basis for asking how those components depend on one another.

Apply this question to every vendor, including AdButler. Separate monitoring is useful evidence of visibility; the dependency explanation establishes what can keep working.

4. How does automatic failover work when a location becomes unhealthy?

Additional servers only help if the platform can use them effectively.

Failover needs to detect a problem, move traffic toward healthy infrastructure, and preserve enough capacity to handle the additional load. It also needs to recognize degraded performance: a service can respond to a basic health check while taking too long to return a usable ad decision.

Ask when the vendor last tested these scenarios. A recovery plan is more convincing when accompanied by results from an exercise.

Ask for: A recent failover test showing detection time, recovery time, error rates, and performance under redirected traffic.

AdButler’s display ad-serving infrastructure includes automatic failover, load balancing, redundancy, and real-time health monitoring. These capabilities support its approach to fault-tolerant delivery.

Recovery should include a return to acceptable response times. Ask how latency changes when traffic shifts to another region and whether the remaining infrastructure can handle peak demand.

Average response times can hide slow requests, so request percentile measurements—such as p95 and p99—which show the response times within which 95% and 99% of requests complete.

5. What does your uptime figure measure, and how can we verify it?

An uptime percentage needs a service definition and a measurement period.

Does it cover successful ad requests, dashboard access, or both? Is it measured globally or regionally? How are slow responses, partial failures, and planned maintenance treated?

A contractual service level, a published availability figure, and historical performance answer different questions. Ask to see each clearly identified.

Ask for: Component-level availability over an agreed period, applicable service terms, incident history, and an example post-incident review.

AdButler guarantees a 99.99% platform uptime and provides public component-level status monitoring. Evaluate that figure alongside the service definition and historical reporting relevant to your requirements.

Communication matters during recovery, too. Your team should know what is affected, what continues working, and when the next update will arrive. Afterward, a useful incident review explains the cause, customer impact, and corrective action.

Bring a real failure scenario to the conversation

A breaking-news surge, major product launch, or peak shopping weekend gives infrastructure decisions a commercial consequence.

Ask your vendor to walk through a realistic scenario: a serving region becomes unavailable during peak traffic. Follow an ad request through the response. Establish what continues operating, what changes, and what your team would see.

AdButler’s investment in its serving infrastructure, regional delivery, automatic failover, and public monitoring reflects our view that reliability belongs at the foundation of an enterprise ad tech platform.

Your availability requirements deserve a concrete discussion.

Talk to AdButler’s team about the infrastructure supporting your advertising business, the dependencies that matter, and the evidence you need to evaluate resilience.

FAQs

Does using multiple infrastructure providers guarantee uptime?

No. Multiple providers can reduce concentration risk, but shared dependencies, application faults, and configuration errors can still cause disruption. Resilience requires effective failure isolation, recovery capacity, and regular testing.

Can an ad server remain available during a dashboard outage?

Yes, when delivery operates independently enough to retain the campaign information it needs. Ask how existing campaigns, configuration changes, pacing, and reporting behave while the dashboard is unavailable.

Is a global CDN the same as multi-region ad serving?

No. A CDN distributes creative assets such as images and videos. Multi-region ad serving distributes the systems that select and return advertisements. Ask vendors to explain where each function operates and how it handles failures.

Why does low latency matter in ad serving?

An ad response that arrives after a page, app, or auction timeout can miss the opportunity to deliver. Evaluate response times by region, during peak traffic, and during failover. Request p95 and p99 measurements to understand slower requests that an average can hide.

What is the difference between an uptime SLA and historical uptime?

An SLA defines a contractual service commitment and any applicable remedies. Historical uptime records availability over a specified period. Compare them using consistent definitions, including how partial failures and planned maintenance are treated.

Does independent ownership make an ad tech platform more resilient?

Independent ownership can support flexibility over infrastructure decisions. Resilience depends on how the platform is built, operated, monitored, and tested. Evaluate the vendor’s infrastructure control alongside evidence of its performance.

What uptime does AdButler provide?

AdButler provides 99.99% platform uptime, supported by distributed infrastructure, redundancy, and automatic failover. Its public status page provides visibility into regional ad-serving components and dashboard availability.

How does AdButler support fast, reliable ad delivery?

AdButler provides sub-100ms response times and a CDN with 90+ global points of presence for distributing creative assets. Its delivery infrastructure combines automatic failover, load balancing, redundancy, and real-time health monitoring to support availability and performance.