Latest / The Cloud Business Podcast with Fexingo: AWS, Azure, GCP, and Enterprise Infrastructure / How Cloud Contracts Are Rewriting Disaster Recovery
Transcript
- Lucas: If you manage enterprise cloud infrastructure, you know the disaster recovery section of your contract is usually a few paragraphs of boilerplate. But that's changing fast — and if you haven't looked at yours since 2023, you may be exposed. Luna: Exposed how? I thought most providers already guarantee 99.99% uptime or something similar. Lucas: Right, but uptime percentage only tells you if the service was available — not whether your data survives an actual regional outage. A major European cloud region went down for 14 hours in late 2024 and a bunch of enterprises discovered their 'replicated' data was in the same physical zone. Luna: Ouch. So the SLA said replicated but the fine print didn't specify geo-diverse? Lucas: Exactly. That outage became a forcing function. Now the big three — AWS, Azure, GCP — are all rewriting their disaster recovery language. What we're seeing is a move toward what I'd call 'geo-diverse replication mandates' in contracts. Luna: And enterprises are starting to demand specific recovery time objectives, not just uptime. Lucas: Right. A Flexera survey from Q1 this year found 42% of enterprises now require explicit RTO guarantees across multiple regions. That's up from 28% two years ago. And providers are reacting. Luna: But doesn't cross-region replication cost a fortune? Data egress, storage, compute — it adds up. Lucas: It does. And that's the tension. Some providers are bundling disaster recovery with reserved compute commitments. You lock in a three-year contract on compute, they throw in geo-replicated storage at a discount. But that means you're committed to one provider for the long haul. Luna: So the disaster recovery clause becomes a lock-in mechanism. Lucas: Exactly. And this is where the smart infrastructure teams are paying attention. If your renewal is coming up, you need to check three things: first, does your SLA specify the maximum distance between primary and replica regions? Luna: Distance? Not just different zones — like, actual kilometers? Lucas: Yes. Some providers have started including minimum geographic separation — say, 500 kilometers. That's a meaningful change from 'different availability zone within the same metro.' Luna: Second thing? Lucas: Who pays for data transfer during failover? In a lot of old contracts, the customer gets charged standard egress rates when they trigger a failover. That can be a nasty surprise — moving terabytes across regions at peak times. Luna: And third? Lucas: Audit rights. Can you actually test the failover? Some providers now allow quarterly recovery drills with no penalty. Others still treat it as a 'best effort' service. If you can't test it, you don't really have a recovery plan. Luna: That's a good checklist. And it's not just about avoiding downtime — it's about avoiding surprise bills. Lucas: Exactly. And speaking of avoiding surprises — if these deep-dive conversations on cloud contracts and infrastructure are useful for what you're building or running, listener support is what keeps this show ad-free and independent. We've set up a page at buy me a coffee dot com slash fexingo if you'd like to chip in. Luna: Yeah, it's a small way to help us keep digging into the details that matter for enterprise teams. No pressure at all. Lucas: So back to the contract shift — one more thing I'm watching is how cloud providers are starting to differentiate their disaster recovery offerings by region. For example, Azure now has a 'paired region' model that is fixed — you don't get to choose your failover region. Luna: That's a concern if the paired region is in a different jurisdiction with conflicting data residency laws. Lucas: Exactly. And some enterprises are pushing back. They want the ability to designate their own failover region — even if it costs more. AWS allows that with its 'custom' replication, but you pay a premium. Luna: So the trade-off is between cost and control. And that's a conversation every infrastructure team needs to have with their finance department before signing. Lucas: Right. Because once you sign a three-year deal with geo-replication baked in, switching providers becomes much harder. The data gravity pulls you in. Luna: That's a good note to end on. Check your contract, know your RTO, and test your failover. Lucas: And don't let disaster recovery be an afterthought — because when you need it, it's already too late to negotiate.