What should you consider when deciding whether to use a centralized, decentralized or distributed network architecture?

940 views7 Comments

CIO, 5,001 - 10,000 employees
What cloud gave us was agility. And if you were willing to pay the price and you needed that agility, it made a ton of sense. It allows you the freedom to experiment and walk away from it without millions of dollars worth of sunk cost. And if it's architected to turn things on and off, it allows you to get access to compute and storage instantaneously rather than going through a complex purchasing process. For certain companies who don't have enough scale for those workloads, Amazon might still be a good answer, because Amazon's purchasing power may make that gap very tiny. But it doesn't make sense at a certain scale. For example, if your company has a certain capacity of always-on, always-running compute, it’s not practical. And folks act like you're an idiot if you say that.

Whenever anybody asks me about cloud, I ask, "What problem are you trying to solve? And what do you need to solve that problem? Does cloud fit that model?” If it does, then go as fast as you can. But if it doesn't, and you're being intellectually honest, why should you do it? It's hard to solve all of the problems with one approach, unless that one approach can encompass and make all the different models possible.
Director of Technology Strategy in Services (non-Government), 2 - 10 employees
If the problem you're trying to solve is to be the best in the city you're in and only ever in that city, then centralized makes perfect sense. As you scale and think about taking your business further afield, decentralized will make more sense. But if you start from day one thinking, "We want to take on the world," then using a distributed model might make sense at that point, and that's what you should be planning for. But the key is knowing what you're trying to do and where you're headed. That becomes the question to ask, rather than, "How do we want to get there?"
CEO in Services (non-Government), Self-employed
If you think two years ahead, how do you want your business to operate then? In terms of cloud, I often hear people say, "Why did I go down the GCP/AWS path only to find out six months later that it really doesn't fit the need that I have? I need the compute, the power, and the scale. I don't need all the other things that come with it. I don't need the ingress fees. I don't need the egress fees."

If you ask somebody what problem they’re trying to solve and they tell you, "I need cloud for X," they typically can't tell you what they need which cloud for, other than one specific workload. That's the whole thing. Why does it have to be based on workload? Where's the bigger vision to go forward with? Otherwise, you're just creating another silo.
4 Replies
Director of Technology Strategy in Services (non-Government), 2 - 10 employees

I'd advance that further because oftentimes, it's not just that they can't tell you what workload. Another issue is that, if they're non-technical, they're working on the assumption that cloud will save them money. And that's why they want to get there. They may not even be able to signal what they want to move to the cloud, but there's still a perception that going there will save them money. That's their driver.

Head of Security and Compliance in Software, 51 - 200 employees

Choosing cloud to save money is absolutely the wrong assumption. You do get agility and perhaps flexibility from cloud, so that even if you don't want to move your production workload, you can have a scalable throwaway infrastructure that you can adapt for development and testing. And if you want to deploy to your production environment, in your own dedicated environment, you have the flexibility to do that. But cost is going to kill you at some point. I run cost management for all our cloud native surveys and we grapple with cloud costs every week. Every Monday, there is a meeting about where the costs are, what the burn rate is and which team or customer to attribute it to. The goal is to figure out how to contain this thing but it's tough.

CEO in Services (non-Government), Self-employed

It's a lot cheaper to use a co-located data center or set up a couple of edge servers than it is to keep dealing with that cost.

Content you might like

Too many active projects at once42%

Poor communication49%

Too many customizations46%

Misalignment with business priorities35%

Skills gaps26%

Lack of resources20%

Other (please list in the comments)1%


564 views1 Upvote

CTO in Software, 201 - 500 employees
Without a doubt - Technical Debt! It's a ball and chain that creates an ever increasing drag on any organization, stifles innovation, and prevents transformation.
Read More Comments
46.6k views133 Upvotes324 Comments


Yes, but we’re moving away from container-based architectures.43%


Other (please specify)0%


1.4k views1 Comment