Latest / The Cloud Business Podcast with Fexingo: AWS, Azure, GCP, and Enterprise Infrastructure / Why Cloud Bills Now Meter Your DNS Query Volumes
Transcript
- Lucas: So I was digging through a mid-size SaaS company's cloud bill last week, and I noticed a line item for DNS queries. Not for a domain registration renewal, not for a hosted zone — actually for queries. A few hundred million of them, priced per million. Luna: Wait, DNS queries? That used to be the thing that was just... free. Like, bundled into the hosting or the load balancer or something. Lucas: Right, and that's exactly the shift. For a long time, DNS was the quiet utility of the internet — you set your records, you forgot about it, and the bill never noticed. But now, every major cloud provider is metering it. AWS has charged for Route 53 queries for years, but Azure and Google Cloud are getting more granular, and the pricing models are getting more complex. Lucas: What caught my eye was a specific charge: for this SaaS company, the DNS line was about three percent of their entire cloud bill. That doesn't sound huge, but it was pure waste — they had legacy failover configurations and a bunch of health checks that were firing constantly, generating queries nobody needed. Luna: So this isn't just about the providers trying to squeeze more money out of you. It's also about how your own architecture can accidentally inflate that line. Lucas: Exactly. And that's the angle I want to drill into today. Because the pricing itself is straightforward — per million queries, tiered by volume — but the real cost is in the design decisions you make without thinking about DNS. Like, do you really need a health check every thirty seconds for a service that's only used by three internal tools? Luna: That's a good point. Let's break down how the pricing actually works now, because I think most people still think of it as a flat fee per hosted zone. Lucas: So on AWS, for example, you pay a small monthly fee per hosted zone, and then you pay per million queries. The first billion queries a month are around forty cents per million, and it drops after that. Azure has a similar model, and Google Cloud has its own tiering. The key is that the per-query charge is now a standard part of the bill, not an afterthought. Luna: And those prices can add up if you're serving a lot of traffic. A large e-commerce site could easily do tens of billions of queries a month, so that's a real cost. Lucas: Oh, absolutely. And the interesting twist is that the providers are also adding features that increase query counts. Like, AWS now offers latency-based routing, geolocation routing, and weighted routing — all of which require more queries to evaluate. You're paying for the intelligence, but you're also paying for the extra lookups. Lucas: There was a great example in a recent cost optimization report: a company migrated from on-prem DNS to Azure DNS to save on hardware, and their bill went up because they had a lot of internal services that were resolving names constantly. They didn't realize the per-query charge applied to internal traffic too, not just public-facing. Luna: Oh, that's sneaky. So even your internal service discovery, if it's using Azure DNS, is metered. Lucas: Exactly. And that's a crucial distinction. When you think of DNS, you often picture the public internet — someone typing your domain into a browser. But the cloud providers are also seeing DNS as an internal control plane. Every microservice that calls another service by name, every Kubernetes service discovery lookup, every container's DNS resolution — it all hits the DNS service. Luna: So what can teams actually do to keep that line item under control? I feel like you can't just stop using DNS. Lucas: You can't, but you can be smart about it. The first thing is to look at your TTLs — time to live. If you set TTLs too long, you get stale records, but if you set them too short, you force more queries. A good balance is usually five minutes for most records, and longer for stable ones. That alone can cut query volume significantly. Lucas: Second, audit your health checks. Route 53 health checks can be configured to run as often as every ten seconds, and each check counts as a query. If you have dozens of endpoints being checked every thirty seconds, that's a lot of volume you're paying for. Luna: And you can use a cheaper internal service for health checks, right? Like, don't use Route 53 health checks for internal stuff. Lucas: Right. Use a lightweight monitoring tool for internal health, and reserve DNS health checks for public-facing failover. That's a common optimization. Lucas: Third, consolidate your zones. If you have hundreds of hosted zones with only a few records each, you're paying a monthly fee for each one. You can consolidate them into a few zones and use routing policies to manage the records. That saves on the fixed cost. Luna: So the per-query charge is one thing, but the per-zone charge is another. And it's not just AWS — Azure and GCP have zone fees too. Lucas: Exactly. And that's the thing: this is a trend we've seen across the board. Providers are unbundling services that used to be free or cheap, and they're getting more granular in their metering. DNS is just the latest example. We've seen it with data egress, with security finding ingestion, with container image pulls — it feels like every month there's a new line item. Luna: I remember our episode on data egress fees. That was a huge wake-up call for people. Lucas: It was. And the question is, is this just a revenue grab, or is it a more honest reflection of the cost of running these services? I think it's a bit of both. DNS is actually infrastructure that has real costs — the providers need to run redundant servers, handle DDoS attacks, and store all those records. But the previous model, where it was bundled, meant that heavy users were subsidized by light users. Luna: And now everyone pays for what they use. That's more fair, but it also requires more attention from the customer. Lucas: Right. And that's the shift for cloud management in general. It used to be that you could set up your environment and forget about it. Now, you need to actively manage your costs, and that means understanding every line item, including the ones that seem small. Luna: Have you seen any new tools that help with this? Like, that automatically analyze your DNS usage? Lucas: There are some newer cost management platforms that are starting to break down DNS spend by zone, by query type, and even by source. But honestly, the best tool is just good old-fashioned logging. Enable DNS query logging, which is available on most providers, and then analyze it. You'll quickly see which services are generating the most queries and whether they need to. Luna: That sounds like a concrete first step. And it's something that any team can do this week, without waiting for a new tool. Lucas: Exactly. And that's the kind of thing that can shave a few percent off your cloud bill, which, at scale, is real money. But it's also a mindset shift — treating DNS not as a fixed utility, but as a variable cost that you can optimize. Luna: And that's the broader lesson, isn't it? Every service in the cloud is becoming metered, and you need to understand the meters. Lucas: Yeah, and that's what we're here for. And speaking of keeping this going, if these conversations are useful for what you're building or running, a couple of dollars a month is genuinely what keeps these going. Buy me a coffee dot com slash fexingo, if you've gotten something out of them. Luna: It really does help us keep the lights on, and it keeps the podcast ad-free. So if you're able, we appreciate it. Lucas: And back to the topic — I think the takeaway here is that DNS metering is not just a minor annoyance. It's a sign of how cloud providers are maturing their billing models. And for you, it's an opportunity to get ahead of the curve by optimizing your DNS usage now, before it becomes a bigger part of your bill. Luna: So, what's the one thing you'd tell a listener to do today to start tackling their DNS costs? Lucas: Look at your TTLs first. It's the easiest change with the biggest impact. Set them to five minutes for most records, and you'll immediately see a reduction in query volume. Then, once you've done that, enable query logging and see what else you can cut. Luna: That's a solid plan. And I'd add: check if you have any zones that are just sitting there with one or two records. Clean those up. Lucas: Absolutely. Every zone you delete is a fixed cost gone. It's the little things that add up. And that's the story of cloud costs in 2026 — the line items keep multiplying, but so do the opportunities to optimize. Luna: Well, that's all we have time for today. Thanks for hanging out with us on the Cloud Business Podcast. Lucas: Yeah, thanks for listening. Next time, we're going to look at how cloud bills now meter your container image pulls — actually, we've done that one. We'll find something new. Until then, keep your bills low and your TTLs balanced.