How to Cloud in a Time of Covid-19

Enterprise IT management should make the assumption that the Covid-19 pandemic is not going to end soon and that things are never going to get back to our pre-virus normal.

It’s time to place cloud computing in the front and center of the company’s IT strategy, its business strategy, and its financial execution strategy. We’ve been talking for decades about getting IT and the business “in alignment.” If there’s never been an urgent need to give this idea more than lip service, well, the need is more than urgent now. 

The business is the computer.

From serving as Conference Chair of Cloud Expo for a decade, and conducting research within my Tau Institute that’s been focused on national-scale IT ecosystems and performance throughout the world, I’ve come to a hold of number of beliefs about cloud computing and what enterprise IT should think about it.

So here are a few fundamentals to get your enterprise ship turned in the water and headed full-steam in the right direction.

I. Cloud First. This was the name of a strategy from the US government’s first CIO, Vivek Kundra, appointed in the early days of the Obama Administration. The program was watered down to “Cloud Smart” in 2017, but the original name was the correct one. Cloud First in an enterprise means what it says.

New applications should be deployed on a cloud, no exceptions

Existing applications -- that pesky 80% of your IT that runs on existing equipment -- should be virtualized at minimum and considered for cloud deployment.

II. Cloud Second. Archived data, data mines and lakes (if you still have them or even want to build a new one), and data repositories of stuff that’s not accessed often or in anything resembling HA should also be migrated to the cloud. Don’t keep filling those old silos; you need to get everything off of your property.

III. Beware of Your Premises. It is not realistic for anyone to say getting everything off-prem. There are regulated environments, there are old cobwebs in systems that are seldom used but critical when they are needed (eg when legal disputes occur), and there are a lot of sunk costs in private data centers and the real estate they occupy that can’t just be unsunk with the snap of a finger and a snappy new cloud strategy.

But push to move your data off of your premises and into a public cloud as much as possible. Make the challenge of staffing mission-critical data centers during a time of plague and natural disasters someone else’s problem. 

IV. Take on the Big Three. This one may seem to be along the lines of “tug on Superman’s cape,” but let’s think about it for a bit. If the growth of public cloud so far was contingent on performance alone, then about 95% of enterprise workloads would be on public cloud. But there’s an intuitive resistance to given up control of one’s data and processes to a third-party vendor, there are regulatory complexities and data sovereignty concerns, and the threat of simply trading in decades of vendor lock-in for Vendor Lock-in 2.0 is real. 

Yet Amazon in particular has been aggressive and innovative in developing services and features, Microsoft has found the path along which to trundle its PC-era customers to the cloud, and Google remains the eternal bridesmaid. A few dozen other vendors (not to denigrate IBM and Oracle with the “other” tag) have carved or are carving their niches. Will it always be such? The fear of Amazon becoming “the Amazon of Cloud” should keep it from utterly dominating for awhile.

What an uncomfortable status quo! Wouldn’t it be nice if you could build your own cloud in association with your peers (say in the financial services industry), or throughout a state, province, or region? What would it take? What is stopping, say, an FSI Cloud or a Midwest Cloud or an Africa Cloud from emerging?

V. Open Up to Open Source. I see you rolling your eyes, you’re preparing to skip over this section or to endure a religious sermon. Please bear with me. The world of open-source software is large and diverse. I think of it as a vast toolbox of languages, frameworks, and (to shorten this sentence considerably) “tools” that perform tasks, sub-tasks, and the tasks leading to those. 

Red Hat discovered fire when they realized most companies would never master the complexities of Linux on their own, so added layers on top of it to make it much simpler and highly functional, and therefore practical, for those companies. So it goes. With open-source software, you can get the free versions and have at it, an approach that will work if you dedicate enough training, internal guruship, and patience to the task. 

Or, you can enlist one of innumerable startups (oftentimes the creator of the free version) that’s put on the usability stuff. This approach can be viewed as a sort of vendor lock-in, but it’s lock-in only for specific tasks. It is painful to switch particular technologies, to be sure, but does not approach the trauma (or impossibility) of changing major vendors as in the pre-open-source days. Set a vision and realize that you can, for example, build your own cloud through a stack of open-source software. I’ve worked for 

VI. Hire, Train, and Retain, Dammit! I think I avoided being dialectical in my previous point, but here is where I’ll sound like Karl Marx to some people because I advocate hiring people, training them continuously, and paying them well and keeping their jobs interesting so you won’t lose them. 

Focusing on people may seem contrary to a post-Covid world in which gatherings of large numbers of people are a point of failure (to think in engineering terms). Plus, we’re moving toward ever more automation, self-healing, self-balancing, and AI-driven process management and improvement in enterprise IT. The post-Covid era also carries with it a strong presumption that scrums and other closely knit teams, even modest in size, are not going to be the ideal model.

Working from home is not the answer for every organization or team, but hybrid approaches will surely emerge. Within this (and any other context), there must always be room for people who are highly paid as possible, especially in an IT shop that is (hypothetically) planning and executing the business plan. 

Hiring people and paying them is the right thing to do if you’re a fan of capitalism, too, a concept that’s been lost over the past 40 years. More people with more money means more things are bought and more more-expensive things are bought.

Suppress wages as much as possible and you have economies like I’ve seen (and lived in) throughout the developing world; factory wages equivalent to a dollar-an-hour (and sometimes less), college graduates knocking down less than $1,000 a month, a centralization of wealth that generates armor-plated vehicles with security guards, razor wire guarding gated communities with more security guards, a simmering tension that’s occasionally (and suddenly) brought to a boil, and an unrelenting immigration pressure on developed nations by the tens of milliions who try to get out of the unlucky places in which they were born, among billions who would also love to try to get out some day. Late-stage capitalism is looking more untenable by the day. 

We can do our part by building business plans wrought of hard work applied toward developing profitable businesses that are the consequence of well-educated, well-applied grey matter put to the task, rather than a modern version of building the Pyramids by hauling stones over a carpet of rolling logs.

VII. Atlas Up and Shoulder the Burden. My bullet points have gotten longer as I became more impassioned. We must all become impassioned, very much so, if we are to bail ourselves out of our current predicament. Covid-19 disease is just one of Earth’s warning shots, indicating to us the mortal danger in which humans live, and in which we will always live. 

Exacerbated natural disasters in recent years seem to be other signals that it’s time to be as smart about using our tools as we are about creating them. It may indeed be too much to ask a humble enterprise IT manager or executive to take on the big problems of the world when the job in front of you is simply to improve the workflow of an inventory management process or shorten the time required to consolidate the books. 

But as the old commercial about buying an oil filter said, “you can pay me now or pay me later.” 

I’m in favor of now.