AWS Outpost, Azure Stack, Google Anthos, and IBM Satellite: The Race to Edge-to-Cloud Architecture Utopia | Blog
Since the days of mainframes and Java virtual machines, enterprise technology leaders have been yearning for one layer of architecture to build once, deploy anywhere, and bind everything. However, Everest Group research indicates that 85 percent of enterprises believe their technology environment has become more complex and challenging to maintain and modernize.
As the cloud ecosystem becomes increasingly central to enterprise technology, providers like Amazon Web Services, Google Cloud, IBM, and Microsoft Azure are building what they call “Edge-to-Cloud” architectural platforms. The aspiration of these providers’ offerings is to have a consistent architectural layer underpinning different workloads. And the aim is to run these workloads anywhere enterprises desire, without meaningful changes.
Will this approach satisfy enterprises’ needs? The hopes are definitely high as there are some key enablers that weren’t there in earlier attempts.
The architecture is sentient
This is a topic we discussed a few years back in an application services research report. Although evolutionary architecture has been around for some time, architectural practices continue to be rigid. However, the rapid shortening of application delivery cycles does not provide architects with the traditional luxury of months to arrive at the right architecture. Rather, as incremental delivery happens, they are building intelligent and flexible architectures that can sense changing business requirements and respond accordingly.
Open source is the default
As containers and Kubernetes orchestration become default ways to modernize and build applications, the portability challenge is taken care of, at least to some extent. Applications can be built and ported, as long as the host supports the OS kernel version.
Multi-cloud is important
We discussed this in earlier research. Regardless of what individual cloud providers believe or push their clients toward, enterprises require multi-cloud solutions for their critical workloads. This strategy requires them to build workloads that are portable across architectures.
The workload is abstracted
The stack components for workloads are being decoupled. This decoupling is not only about containerizing the workload, but building services components that can run on their own stack. This capability helps to change parts of workloads when needed, and different components can be deployed across scale-distributed systems.
With all this, the probability of achieving architectural portability may indeed be different this time. However, enterprise technology leaders need to be aware of and plan for several things:
- Evaluate the need for one architecture: In the pursuit of operational consistency, organizations should not push their developers and enterprise architects toward a single common underlying architecture. Different workloads need different architectures. The focus should be on seamless orchestration of different architectural choices.
- Focus on true architectural consistency versus “provider stack” consistency: This consistency issue has been the bane of enterprise technology. Workloads work well as long as the underlying platforms belong to one provider. That is the reason most of the large technology providers are building their own hybrid offerings for Edge-to-Cloud. Although many are truly built on open source technologies, experienced architects know very well that workloads porting across platforms always require some changes.
- Manage overwhelming choices: Enterprise architects are struggling with the number of choices they now have across different clouds, terminologies, designs, and infrastructures, which makes their job of building a unified architecture extremely difficult. They need automated solutions that can suggest architectural patterns and choices to drive consistency and reliability.
So, what should enterprise architects now do?
Enterprise architects have always been envied as the guardians of systems, as they set the broad-based strategic underpinning for technology systems. Going forward, they will be bombarded with more choices and provider-influenced material to choose their underlying platforms to build workloads. All of the platforms will be built on a strong open source foundation and will claim interoperability and portability. The architects will need to be discerning enough to understand the nuances and the trade-offs they have to make. At the same time, they should also be open and unsuspicious of technology choices. They must have transparent discussions with technology providers, evaluate the offerings against their business needs, and assess the drivers for a unified architecture.
What is your take on unified architecture from Edge-to-Cloud? Please share your thoughts with me at [email protected].