Cloud has disrupted traditional business models, offering unprecedented flexibility and efficiency. It also has altered customer expectations, creating demand for unwavering reliability. Users expect timely, always-on access to the online services and platforms they depend on.
However, when it comes to moving mission-critical workloads from on-premises data centers to the cloud, IT leaders continue to encounter internal organizational resistance and face technical challenges that thwart these initiatives. This is mainly due to the lack of clarity surrounding the process of migrating these workloads to the cloud while meeting the demands for continuously available and reliable services.
Another area of uncertainty manifests when considering architectural and operating models, because the traditional approach to addressing and ensuring resiliency needs to evolve.
Given pervasive cloud platforms and cloud-native patterns, organizations must shift from resiliency approaches that facilitate infrastructure failover and recovery mechanisms to a model that maintains end-to-end service reliability by transparently obscuring failures and disruptions.
We call this Always On.
The Always On model embraces a holistic, all-encompassing user-centric perspective of reliability. It is achieved by going beyond the boundaries of infrastructure and reimagining the end-to-end service stack, including applications, platforms, data, deployment patterns, networks, and even organizational culture.
My ebook, "Cloud Adoption for Mission-Critical Workloads: Principles for Designing Always On Applications"
presents a detailed look at the Always On model. The takeaways are strategies for organizations like yours to successfully adopt cloud platforms and cloud-native patterns.
The following excerpt sets the stage for that conversation by introducing the core principles organizations need to embrace in order to move to the Always On model.
Four core principles of the Always On model
Engineering Always On services requires a set of core principles, which are delineated across four domains:
- Multi-active deployments: An Always On architecture establishes a multi-active deployment model, dispersed in several cloud location scopes, to ensure that applications remain active.
The architecture is designed to transparently steer traffic to other active cloud location scopes in the event of application bugs, component failures, regional disasters, and even human operational errors.
- Application patterns: Reliability is not solely reliant on the resiliency of the infrastructure. The application itself and its dependencies must be dependable to increase the overall service reliability. This is why an Always On architecture explores modern development patterns and techniques to contribute to the reliability of the end-to-end service.
- Continuous operations: Mission-critical applications require a genuine culture of collaboration where SREs shoulder the responsibility for the overall service reliability.
An Always On approach entails balancing velocity and rapid feature releases while upholding Service Level Objectives (SLOs) and delves into strategies that harmonize SREs with services rather than confining them to technology-driven silos.
- Evidence and confidence: Merely establishing well-defined SLOs and SLAs falls short unless the organization possesses the substantiating evidence and confidence to sustain them.
The Always On approach revamps testing methods by adopting chaos engineering as a daring necessity to enhance longer term reliability. This is done by performing controlled experimentation on the system to understand end-to-end service behavior, uncover potential weaknesses and as a result, improve the overall architecture.
Always On is a strategic journey
For businesses to succeed in their cloud adoption, they must embrace a new target operating model and adopt a next-generation architecture.
The comprehensive Always On approach, which incorporates deployment strategies, application design, and operational practices, gives businesses a foundation to thrive in a business and IT landscape that continues to evolve.
Haytham Elkhoja is a Principal Architect and Director in Kyndryl’s Office of the CTO.
