It’s time to expect more from managed infrastructure services
When it comes to managed infrastructure services, automation is the buzzword of the day. The problem is that automation only really works if you have high-quality data to support it. If you need to manage a large quantity of servers, how can you automate anything if you haven’t even documented which operating system is running on each of them?
The quality of your configuration data is not just a limiting factor. It’s THE limiting factor. In a previous series of articles, I wrote about how IT managed services was becoming a software-driven business that requires software engineering skills. Unfortunately, it is also exposed to the same risks as software engineering — specifically, ensuring quality at enterprise scale.
Even working with hyperscalers is complicated enough. Although they are fully automated with their own control plane, API and configuration database, most large enterprises have a mix of IT environments that include on-premises or private data centers which have spilled over into cloud. This hybrid cloud or mixed cloud setup adds another layer of complexity.
So, if you face a mix of cloud and on-prem, modern and legacy, and automated and manual infrastructure management processes, how can you possibly manage it all effectively?
The trouble with manual infrastructure management
The right way to address this complexity and need for quality is to invert the equation. Start thinking about delivering infrastructure services as if they are automated – even if the processes are manual. It takes discipline and it requires behavior change, but it’s worth it.
Let’s look at an example. If your IT environment includes legacy infrastructure, it will still require manual work. Provisioning a server, connecting it to a network, or plugging cables into a switch may all still be done manually.
However, when you have the process maturity to thoroughly document these manual processes, it makes the legacy process transparent. But that's still not enough.
In managed infrastructure services, data quality is not just a limiting factor. It’s THE limiting factor. If you are managing infrastructure data after it’s deployed, you can’t implement automation at scale.
Overcoming challenges with data
Documenting your infrastructure and processes is a step in the right direction, but ensuring that you are capturing the correct data is an entirely different issue. You need to take human errors (and manual data entry) out of the equation.
Deploying automation requires very accurate and precise information about the actual state of the environment. Hence, it cannot exist without an underlying data strategy that is exhaustive, precise, and of very high quality. This requires a mature, large-scale and high-quality approach to data — something along the lines of real-time data updates or a data warehousing or data governance strategy.
The truth is that if you are managing data about your infrastructure after it is deployed, you can't fully trust it to implement automation at scale.
Before calling a data center engineer to plug a new blade into a server rack, document it. You must accurately capture all data about the new piece of hardware at the time of request. This data is stored in a management database which not only documents the current state of your infrastructure, but is the first step towards automating future processes.
It also enables traceability, rollback or recovery, and better analysis of infrastructure performance to improve the overall quality and efficiency of managed services. In addition, it may also provide the incentive needed to start working towards eliminating manual processes.
Is it possible to automate legacy infrastructure management?
Let’s be clear – true transformation takes the will to change, and no company will embrace a major change without justification. However, if you can demonstrate the advantages of automation, it will push the enterprise towards automated infrastructure services on demand.
Deriving value from infrastructure relies on economies of scale, and the three contributing factors are cost, quality and speed.
When it comes to managed infrastructure services, automation is the only way to achieve success in all three aspects simultaneously. For large-scale multi-cloud setups, infrastructure automation is the only way to meet changes in demand without a linear increase in costs and extreme cognitive load for application teams.
What if you could consolidate all your managed infrastructure tools into a single platform accessible via a control plane? If the goal is to offer the ease of a hyperscaler experience even for legacy infrastructure services, what would that look like?
Designing the future of managed infrastructure
Based on my experience helping global enterprises solve their infrastructure challenges, I have been thinking a great deal about this topic. If I had to design such a product from the ground up, my requirements would include:
- A single interface to create and manage infrastructure services across legacy and cloud
- Ensuring that services and infrastructure metadata is accurate and
- The ability to access and manage services through a set of APIs
- A catalog of infrastructure building blocks that enforce your policies
- Support for multi-cloud environments
- Automation by design
- A SaaS-like consumption model
Such an approach would constitute a true “build your own cloud” service. It would make all infrastructure services transparent, with a single software control plane leveraging a set of APIs or GitOps methods. It could employ an app-store-like model to deploy infrastructure automation components — which are either available off-the-shelf or developed in-house by citizen developers.
The product would manage infrastructure deployed with hyperscalers like Azure, Google, AWS and VMware, as well as legacy data centers. It would also take the human element out of capturing services and infrastructure metadata by making state data an integral part of the automated process to reduce the risk of errors.
Above all, automation would be at the heart of the solution, bringing hyperscaler-like convenience to companies that must manage a mix of modern and legacy infrastructure.
What we're doing about it
A managed infrastructure solution like this is exactly what I and many of the hybrid cloud and infrastructure experts at Atos have been working towards for some time.
We’re calling it the Enhanced Programmable Infrastructure Service, or EPIS. It’s a standardized, fully automated, software-driven foundation that controls what service and workload is being delivered, how it is configured and covers the complete lifecycle management of services and workload assets.
We believe it will be a game changer that will enable enterprises to overcome complexity, streamline infrastructure management and deliver increased agility at lower cost.
Watch this space to learn more!
Posted on: December 12, 2024
A version of this article was originally published in Intelligent CIO Europe.
Alexis Mermet-Grandfille
Group CTO Strategic Technology AdvisorMember, Atos Research Community
View detailsof Alexis Mermet-Grandfille>
