Is multi cloud the future, or just a buzzword with hidden pain?

[#1]

🚨 Atlassian Status Page Lights Up!
Couple of weeks ago I woke up and saw a lot of messages on teams, I noticed Atlassian reporting an active incident affecting Jira, Confluence, Bitbucket, and more.
A quick look revealed the root cause an outage in AWS US-EAST.
It’s fascinating (and a bit scary) how one region in AWS catches a “cold,” and suddenly half the tech world starts sneezing 🌎🤧
Behind the scenes, so many critical platforms productivity tools, CI/CD systems, even enterprise backends are running on the same infrastructure fabric.
It made me think:
👉 Should companies diversify their cloud strategy to reduce dependency on a single provider?
Or
👉 Would that just introduce architectural complexity, operational overhead, and higher costs that outweigh the resilience benefit?

In a world where “cloud” often means “AWS,” maybe resilience isn’t just about redundancy within a region but across clouds.
Curious to hear what others think:
Is multi cloud the future, or just a buzzword with hidden pain?

[Mathew Lederman]

Why not just have DR and fallback options that support multi-region tenancy? Sure, you still run into an issue if all of AWS goes down, but then again, if 3/4 of the world stops working what does your ticketing system really matter?

To answer your question more directly. Should Flo and Bob's Rug Emporium build out a multi-cloud solution, probably not. Should SaaS tools be able to stay up if a single region in a single Cloud provider goes down, probably (unless I've pinned my data to that provider/region).

Slack is on AWS. Didn't hear a single thing about a Slack outage.

[Chris]

Yeah, happy new world with cloud. It's more expensive and the risk of loosing access to one's own data is bigger, then having a lean system on premise.

I would never ever allow my company to run critical infrastructure on cloud.

Atlassian Cloud solutions are a very useful tool set. Have you ever had a look at the open bug list or the open feature requests? They tend to produce 80% solutions and make their customers alpha tester. Happy new cloud world.

If I could, I would go back to Jira Server.

[Jeffery Candiloro]

To reiterate @Mathew Lederman's comment - multi-region tenancy or redundancy is the answer. With AWS, you don't even need to leave the geographic region to have multi-region redundancy. Over to Atlassian to implement it.

[Richard Scholtes]

I think for Atlassian it's already happening. As I have to always keep my eyes peeled if Atlassian is using new data providers, I saw that as an example Google appeared as an additional general cloud provider in the list.

[Matt Smith]

It was a DNS issue that affected the MFA element of AWS.

Services were not affected directly as they are distributed,

However if you needed to authenticate to get access then this did cause some access issues !

To avoid cloud and the many benefits it brings is (in my opinion) naive and narrow minded.
 but hey lets scare people who have not got the knowledge, by people who appear not to have the knowledge or skill to manage cloud effectively. :-)

[Rick Westbrock]

Maybe a better example of an outage was the recent Microsoft Azure CDN outage which didn't just affect logins but took "everything" down due to a misconfiguration which killed their Kubernetes clusters around the world. In that situation even having some sort of DR in another region would not have helped because they basically took down every region.

If I think back to how many times PaaS vendors have had such widespread outages they seem to be fairly uncommon and don't normally last for extended periods so unless I were running a critical application like online sales which make up the majority of my overall sales I don't think the added complexity and expense of multi-vendor or even multi-region redundancy would be justified.