Some people are calling 2023 The Year of Cloud Repatriation. I think this is a bit inflammatory.
I call it The Year of Right Clouding. But the sentiment is the same.
It all starts with the nature of “hybridness” and the intelligent placement of workloads.
Hybrid, meaning some on premises and some off premises, is already the default architecture for IT and has been for a while now. This year, however, is subtly different. Until this point it is probably true to say hybrid has been seen as that interim state on a journey to what many might condescendingly call “cloud maturity” enroute to full public cloud. Now it seems IT leaders realize hybrid may not be a transitory state. It may actually be the preferred one.
While we’ve been reading similar predictions for years, the stars are finally in alignment. More customers are looking at hybrid as a permanent state because of the welcome flexibility it brings to IT, allowing organizations to use existing architecture in their data center while still taking advantage of cloud’s benefits to optimize costs and extend on-premises IT capabilities.
Hybrid’s popularity is also getting a boost from companies that are predominantly on premises, but eager to exploit the latest technology found in cloud models. Organizations need greater processing power as they turn to advances such as AI to help them process and analyze data faster and more efficiently. The cloud route offers enterprises a flexible and scalable way to meet these increasing processing needs without needing to invest in costly hardware, while also addressing cybersecurity concerns. This is an excellent use of public cloud. But not all use cases are, so the “hybridness” remains.
But this isn’t a transition you can navigate without careful planning. In fact, if you’re too hasty, you might move things into the cloud that will leave you saddled with a negative ROI.
Some companies still get tripped up by failing to do proper planning and wind up moving the wrong workloads to the cloud, forcing them to backtrack and start again from scratch.
In addition to the expense and the embarrassment of an ill-advised migration, they’re then left on the sidelines as data continues to grow at exponential rates. Instead of becoming more efficient and productive, they’re unable to achieve the desired business outcomes that would be the reward of a successful cloud migration and risk falling behind the competition.
Why do some migrations fail? What often happens is infrastructure teams are forced to scramble to keep up with the demands and requirements as applications get moved to the public cloud. The net result is a kind of an inefficient hodgepodge solution that hurts performance.
That’s because cloud projects are usually driven by software architects who often map out these projects without involvement or feedback from the organization’s infrastructure architects – with predictable results. Cloud masterfully abstracts the performance-impacting nature of infrastructure realities, and software architects are all too keen to believe this apparent absence is actual absence of anything to worry about. That is often not the case.
The goal should always be to put the right workload with the right data into the right place. But some types of workloads are not going to work well in the cloud because they don’t translate well to a microservices architecture with higher latency from network, storage and inter-services messaging. In those instances, perhaps the best decision is to leave things as they are for some applications. As one customer once said to me “sometimes monolithic is optimal.”
Straightening this out requires better communication between the infrastructure and software architecture teams. This will take some doing considering how developers have usually been the ones spearheading cloud initiatives inside their organizations. This effort ought to be part of a push to better optimize hybrid cloud deployments and find the right balance.
There’s an urgency to fix this sooner than later because, as I noted earlier, hybrid cloud demand is kicking into high gear.
More than three-quarters of enterprises now use two or more cloud providers, and one-third have more than 50% of their workloads in the cloud. Both on-premises and public cloud investment are only expected to increase in the coming years, with Gartner predicting that end-user spending on public cloud services will reach nearly $600 billion this year.
Hybrid is a way for organizations to leverage the benefits of both public and private clouds as they move their workloads. Further, it offers more options to host workloads – whether that be on premises, in a hosted private cloud or in a public cloud. That flexibility also contributes to a more optimized IT infrastructure and lower costs as companies optimize resource allocation and improve the performance of their cloud infrastructure.
Companies also stand to reap a security bonus by being able to leverage the security benefits of public and private clouds. With private clouds, they gain more control over security; at the same time, they benefit from the more advanced security approaches necessary in public clouds.
That trend has continued in the aftermath of the accelerated embrace of the public cloud during the pandemic when so many organizations sped up existing plans to meet dramatic shifts in customer demand – as well as the move to hybrid work. Now cooler heads are starting to reevaluate that public cloud mania.
Many organizations are also looking at hybrid cloud to cap their skyrocketing monthly public cloud bills, a challenge that’s been exacerbated by cloud sprawl, where organizations wind up operating several different clouds for the same or similar workloads instead of matching the right workload for the right cloud. Further, the cloud storage costs associated with managing all that data have added strain on corporate budgets at a time when CFOs are in no mood for financial surprises.
Pay as you go is great – provided you understand exactly how you “go.” Otherwise, you may be paying for something that’s different than what you intended. This is particularly true of storage, which tends to be long lived and steadily growing, the worst of combinations for pay as you go.
All of this causes me to wonder if “cloud first” is dead? I think for the most part it is.
“Cloud first” was the idea that a decision maker was so bought into the public cloud sales pitch that they would insist on everything moving to public cloud unless it was impossible to move. The problem with this thinking is sometimes things that are not impossible are still ill advised.
Cloud smart means making the best use of cloud concepts whether they are on premises or off but fundamentally making the most rational choice of locality as part of the thinking.
A cloud smart architectural approach is essential for enterprises because it enables them to optimize their on-premises IT infrastructure and leverage the benefits of the cloud as well. With cloud smart architecture, enterprises can design and deploy highly available, scalable and resilient solutions that have cloud operating characteristics to adapt to their changing business needs.
After the initial rush to public cloud, this belated dose of reality is a positive. It reflects the recognition that there needs to be a smarter balance right between what’s on premises vs. what’s in the public cloud. Knowing how to strike the right balance – with the understanding that not every application is meant for the cloud – can ensure that you optimize performance, reliability and cost, driving better long-term outcomes for your organization.
By Premkumar Balasubramanian
