Stop adopting multicloud to achieve application resilience, says Honeycomb's Charity Majors

3 years ago 287

Commentary: Multicloud whitethorn beryllium bully strategy successful immoderate instances but not for delivering exertion resilience. Here's why.

For enterprises inactive searching for the beatified grail of multicloud, Honeycomb co-founder and CTO Charity Majors has a suggestion: support looking. Or, rather, don't.

To beryllium fair, Majors doesn't suggest that multicloud isn't utile for exertion portability (though Gartner has), oregon that it can't beryllium a large instrumentality successful declaration negotiations (Duckbill expert Corey Quinn volition guarantee you that it's not). No, Majors' constituent is simply that multicloud won't output "magical pixie dust" of resilience. Put much bluntly, she declares, "it is NOT the mode to hole your reliability problems."

Why?

SEE: Research: Managing multicloud successful the enterprise; benefits, barriers, and astir fashionable unreality platforms (TechRepublic Premium)

When multicloud works

But first, it's worthy remembering that contempt each the hype astir multicloud, it's mostly conscionable however endeavor IT works and ever has. If 1 large committedness of unreality computing was to unshackle developers from cumbersome hardware procurement processes ("Yes, I request a server for my application. Can I get that by adjacent year? Please??"), that state has led developers to bargain into a scope of clouds to entree cloud-specific services to physique their applications.

This is what I telephone "incidental multicloud." It conscionable happens.

There is besides "intentional multicloud," but wherever I chiefly spot this is from vendors (including wherever I enactment now, MongoDB) who request to conscionable their customers wherever they hap to be. If a lawsuit comes to Confluent and wants to tally the Confluent Cloud connected Microsoft Azure, it's a mediocre income strategy to archer them, "No, we lone enactment Alibaba" (or immoderate cloud). So vendors with unreality services whitethorn enactment multicloud to guarantee customers tin usage information stored successful antithetic clouds to tally a azygous exertion (perhaps storing information successful Cloud X for cost/other reasons but analyzing that information successful Cloud Y owed to superior analytics services, without pushing the lawsuit to manually determination the data).

But that's the vendor taking attraction of that dense lifting for the lawsuit to guarantee a seamless infrastructure/database/application/whatever acquisition for the customer. That's not what Majors calls crazy. For customers who presume the load of multicloud absorption successful the hopes of greater resilience ("I'll drawstring my app crossed aggregate clouds successful lawsuit 1 goes down"), Majors wants to speech you retired of it.

It turns retired that the way to information isn't accrued complexity.

Making hard things adjacent harder

"I recognize not wanting a azygous constituent of failure. But erstwhile you adhd a unreality you don't get much reliability; you astir surely get less," Majors noted. Why? Because adding complexity doesn't simplify things. Majors added: "Instead of worrying astir AWS being down a fewer min a year, present you person to interest astir AWS, GCP, and the unholy plumbing betwixt them."

This seems obvious, yet seemingly it's not. Not to everyone, anyway.

In fact, she went on, it's the plumbing that creates the astir problems, "because you are not amended than AWS oregon GCP astatine gathering and operating systems. Promise." As mentioned above, determination are companies who bash person the resources to guarantee seamless cognition of peculiar infrastructure/services betwixt clouds (continuing the illustration above, Confluent promises it tin "automate[] gathering and monitoring information pipelines and streaming applications, portion offloading the operational load of your developers"), but they don't purport to beryllium generalists delivering resilience and different benefits nary substance the workload. They're specialized truthful that IT needn't be.

Multicloud whitethorn good assistance with availability issues, but it's not truly a strategy for getting to improved resilience–not without assistance anyway. There's simply excessively overmuch that can't easy beryllium predicted successful the connections betwixt the clouds (Majors: "[I]f you're treating [multicloud] similar a blistery failover, you volition not trial it often capable to forestall nuts & bolts flying disconnected each time"). Again, if you person a vendor with the expertise to grip these connections for you, fine. But trying to tally generalized workloads crossed clouds connected your ain isn't a look for resilience–it's a look for tons of resilience whac-a-mole.

Disclosure: I enactment for MongoDB, but the views expressed herein are excavation alone.

Cloud and Everything arsenic a Service Newsletter

This is your go-to assets for XaaS, AWS, Microsoft Azure, Google Cloud Platform, unreality engineering jobs, and unreality information quality and tips. Delivered Mondays