Despite multi-cloud’s popularity, many organisations are still struggling with delivery. Which led us to the question: what’s going wrong with multi-cloud adoption? More specifically: how can we help organisations who are currently getting it wrong, get it right?
Our Founding Partner Kris Saxton recently presented at Computing’s Cloud and Infrastructure Live on the ‘7 Sins of Multi-Cloud – Pitfalls and how to avoid them’. If you missed the presentation, you can find the recording here. This article addresses some of the questions that followed from the presentation and offers our best insight from over 60 projects worth of experience on how to successfully implement a multi-cloud strategy.
First and foremost, what is multi-cloud?
Multi-cloud is pursuing a strategy of adopting more than one public cloud at the same time.
The multi-cloud strategy emerged because most definitions of hybrid cloud are not yet workable, yet people still want to keep their options open. Even though most people’s fear of lock-in is misplaced, it is still the main driver behind multi-cloud.
It is important to also cover what multi-cloud is not.
Multi-cloud is not adopting more than one cloud provider at the same time and trying to abstract those cloud providers away and move workloads between them dynamically. There’s a great story that’s pedelled to clients about checking the spot price for public cloud resources and automagically migrating workloads based on a lowest cost model. To be clear, you are aren’t going to get that. Those that have tried end up locking themselves into esoteric, proprietary abstractions layers (the very problem they trying to avoid) or suffer a chronic paucity of features as they play catch-up with public cloud providers whose annual spend on their own platforms exceeds the entire revenue of most organisations.
What multi-cloud does give you is a way of matching different workloads with the best cloud provider, commercial leverage with each of those cloud providers and an ability to monitor and manage the risks and costs of moving between cloud providers, should you ever need to.
Is multi-cloud flawed strategically? As by its definition, it avoids economies of scale from providers.
Quite a difficult question to answer as it can be interpreted differently. To answer it directly and literally, multi-cloud doesn’t avoid economies of scale. The whole point of any provider is that they gain from economies of scale through their size, and they give you the benefit through a pay-per-use model. So someone ordering a single compute instance gets access to huge economies of scale. A different interpretation could be that, by splitting your efforts across multiple providers, you’ll be doubling-up on tools, processes, and management overhead and there’s definitely truth in that claim. For a large organisation, these costs are outweighed by the strategic options that adopting more than one provider offers, but for smaller organisations, it would be a perfectly sensible approach to go ‘all-in’ with a single provider.
How do you respond to resistant, traditional IT operations people when implementing new tech/cloud solutions?
There are so many reasons to resist change, describing them all would take a week.
It’s actually a common myth that there is much resistance any more, what we’re really seeing is bad adoption. Most people are now over the idea that public cloud has no place in their hosting strategy, with only a handful of organisations with very niche requirements still holding out.
The retail and media industry has led the way in changing these attitudes, as they were quick to feel the impact of digital transformation. We’ve seen enterprises disappear overnight, ones that had over 50 years of history are now gone. The ones that survived are likely to be hosted on public cloud so they can move fast.
You spoke about the importance of the Product Owner, what if you don’t have product owners or aren’t a scrum house?
Not having product owners suggests you haven’t made the transition to product or service orientated teams, which means it’s likely all the people required to make public cloud work are all operating in their own functional silo. This implies the strategy is more lift and shift than transformation. In which case, you’re still going to get some benefits, possibly a reduction in the cost of changing and maintaining these systems, but your hosting costs may actually go up.
My advice would be to organise yourself around products or services and hire dedicated product owners whose sole purpose is owning the vision for the team and keeping them aligned to that vision through clear prioritisation of work. Without focusing on organising yourself around products and services first, the owners have nothing to own but a bunch of technology.
How do you manage to find people with the right skills for a multi-cloud project?
One of the nice things about multi-cloud is that you can search for those people with specific skills for a specific provider independently. You don’t necessarily have to have all the skills for everything in one person, or even in one team.
It’s not easy though, that’s why we grow our own with the DevOps Academy. If you look at CW jobs, there’s something like 450,000 people registered. The types of people with the skills we need are in the 100’s. Anyone that’s serious about public cloud strategy needs to be looking to build their own capability, it’s unlikely you’ll be able to do all the hiring externally.
There are so many opinions on whether various technologies will replace or create jobs, what is your opinion when it comes to cloud?
Specifically on cloud, because cloud goes hand in hand with software designed infrastructure and automation, cloud, done well, means you can support more services with fewer people. Those people need to have slightly different skills that are more akin to a software developer than a systems administrator, though the good ones generally have both. I don’t think it will lead to a reduction in jobs, because I’ve never had a client whose operations team weren’t already completely overstretched and overworked, and usually with mundane tasks. With Software-defined Infrastructure, you can eliminate a lot of that drudgery.
What is the ‘Golden Rule’ of being Agile/Lean?
There aren’t rules, but there are principles and values. These are all well documented, and I won’t repeat them here, but even more fundamental (and easy to remember) is the mindset behind agile, which goes as follows:
Respect people. Continuously improve.
That’s it. If you can hold those in your mind’s eye as you go about your work, you’ll be more likely to succeed in almost anything and everything
Is there an ideal route to take to maximise cost efficiency?
Your initial exploration into multi-cloud should not be about optimizing cost. It should be speed and quality. You should be perfectly happy for your cost to increase in the short term but know you will recoup those cost in other areas, such as fewer service problems, better products and services and happier people.
What advice would you give someone in the early stages of considering a multi-cloud strategy?
Spend the time upfront setting a clear vision for what business goal you want to achieve through the work and how public cloud is going to help you get there. Ideally, public cloud should be the only thing that will get you there. And hire a dedicated product owner.
If this hasn’t answered your questions on something you would like to know more about, you can get in touch with us any time at firstname.lastname@example.org address.