DevOps and cloud lend massive agility and scale to most organisations, but some have an easier time than others.
While large, mature firms can see massive benefits, they often struggle to adopt a DevOps culture due to their significant legacy: monolithic applications running in datacentres full of servers.
"We're almost at a stage where very small businesses...typically are almost DevOps native," said Mark Ridley, CTO of Ridley Industries, speaking during a panel on scaling up at Computing's DevOps Live this week.
"They come from a position where they are building infrastructure as code for automation, and they often have a much better understanding of how to build quality through the stack."
Wealth management firm Saunderson House is an SME by most definitions. Head of information technology Nick Rosser agreed with Ridley and stressed the importance of starting early.
"One of the lessons I've learned over the years is, it can be worthwhile putting processes, procedures and regulatory regimes in place early on.
"You're obviously always meeting those regulatory requirements, but the processes and policies are maybe pushed down the road and introduced later, as the firm starts to mature and is maybe not at such an agile phase itself."
Rosser argued that putting rules around regulation in place early pays dividends later, "even if you are trying to be an agile startup." Changing culture and processes after a firm has grown can be difficult, time-consuming and expensive.
Highly regulated industries like finance face a trade-off between how quickly you can optimise to scale versus how quickly you can expand to scale, said Marcus Corner, interim CTO at Modulr Finance. He warned against losing sight of regulatory compliance in the rush to grow.
"What you might end up doing is scaling your infrastructure so you meet those demands in the short term, whilst you optimise to fix your code in the long term.
"You can't take any shortcuts."
Regulatory frameworks often vary on a country-by-country basis. Jas Sagoo - Okta's senior director of solution engineering - said you must consider this when planning development.
"Think about where your largest clients are and where the problems you are trying to solve are. That's where you probably want to try and make sure that you are addressing the regulatory requirements around those regions, in those countries.
"This is where scale comes into it: don't lose sight of [regulation]. You have to cross geographies, continents, and your platform solution has to be able to deal with the compliance and regulations in those countries and regions from a data perspective."
Automation can make things worse
So, DevOps and the cloud can dramatically accelerate scale-up efforts, but handling them correctly is key. They're not a silver bullet.
Ridley explained: "The ‘why' of DevOps is quality. If your team don't understand quality, it won't help you. What it might do is actually speed up the disaster."
How does that happen? Automation is a large part of it.
Automating tasks is meant to make everything faster and more efficient, but if you automate a bad process - say test cycles, or deployment - it gets locked into code and becomes hard to change.
"If we don't start from a position of high quality, we can automate disaster," said Ridley.
He gave an example:
"I've worked with a business that wanted to rapidly modernise their entire stack. They wanted to replace all of their servers, containerise everything, use micro services, and they had started a transformation programme to deliver. But what they didn't do is actually produce quality code.
"They had no understanding of quality inside the organisation culturally. There was no testing, there were no gates to prevent bad code being put into production; and effectively all DevOps does speed up that route into production."