To multicloud, or not: Former PayPal head of engineering weighs in

The push is on to get every enterprise thinking they're missing out on the next big thing if they don't adopt a multicloud strategy. That shove in the multicloud direction appears to be working. More than 75 percent of businesses are now using multiple cloud providers, according to Gartner. That includes some big companies, …

[deadlockvictim]

Size

How big do you have to be to benefit from a multi-cloud setup?

You need developers, DBAs and sysadmins who are not only familiar with each environment but also with how the environments interact with each other.

I imagine that the various providers, and especially Microsoft, will make you pay handsomely when it comes to inter-cloud communication.

And at what point do on-premises (or locally co-located servers) servers become more expensive or less worthwhile than a cloud setup let alone a multi-cloud setup.

[Anonymous Coward]

Re: Size

They're likely not doing any of it in-house, but relying on MSP/ISVs to do the heavy lifting for them.

(Anon, because I work for an ISV).

[Claptrap314]

Lazy terminology -> bad results

Calling an AWS shop that uses GCP for analytics "multi-cloud" is nonsense. I mean, if you want to make the board happy, I guess. To be meaningful, multi-cloud needs to be talking about having an application running production loads in multiple cloud providers. If you're even trying to be serious, that means having DNS resolve to load balancers in multiple clouds, with the LBs in each cloud capable of routing all traffic to any cloud.

Actually doing THAT is not likely to make sense for any but a very narrow slice of businesses. As mentioned, the complexity of getting it right is a substantial cost in its own right. Getting any actual improvement in resiliency means understanding a LOT about the physical locations of the servers (information which, for some reason, cloud providers are reluctant to provide) and ensuring that barn X for CP A is sufficiently isolated from barn Y for CP B.

FAR better to get a deep understanding of reliability capabilities of your provider, and set up resilience using their tools. Your engineers are certainly NOT any better than theirs, and unlike yours, they DO have both the data needed to do it right, and the job description of only getting it right.

Unless you are on Azure. Then you can look for the occasional multi-hour all-systems outages.

But for AWS & GCP, if your business actually needs more than four nines, then you can read up on what exactly it takes to get there. I would think that 99.997% uptime is quite doable with single cloud in either. But if you actually need five or so, it's probably worth looking into building your own over going multi-cloud.

[Anonymous Coward]

Novel concept

The idea that you should actually do research into your own needs as well as into cloud companies' offerings instead of just going with Gartner Trends? Nah, it'll never fly.