IT Brief Australia - Technology news for CIOs & IT decision-makers
Story image
Moving applications between cloud and data centre
Tue, 19th Mar 2019
FYI, this story is more than a year old

Enterprise transformation is rife with stories of moving legacy systems and applications to the cloud. These transitions are ostensibly planned to leverage the security, scalability, efficiency or availability of cloud resources.

Yet sometimes it makes more sense to repatriate applications from the cloud back to legacy data centers or on-premise infrastructure.

This article will review what applications can be repatriated, and when it makes sense to do so,

Hybrid infrastructure has opened up options.

The value of hybrid infrastructure is that it provides both the flexibility and ease of on-demand scalability, along with governance and control of on-premise infrastructure.

Sometimes, applications need to repatriate to take advantage of the latter. With hybrid infrastructure, IT teams can pick and choose depending on requirements.

Modern, hybrid enterprises are ready to adopt and deploy infrastructure and application utilities across environments to ensure application continuity and evolution to maintain competitive advantages.

The many considerations of application infrastructure.

Application architecture, data security, and compute and network resource availability are key considerations when choosing hosted infrastructure options.

It may be worthwhile for certain applications to run on-premises while disaster recovery runs in the cloud to leverage flexibility and agility and avoid business disruption and customer impact.

It makes sense to pull a custom application out of cloud and back to an on-premises enterprise private cloud environment if it meets any of these use cases:

  1. It's a business critical application that is a heavy, continuous consumer of compute resources (memory, CPU, etc.)
  2. It demands continuous high connectivity and higher network bandwidth from cloud environments.
  3. It requires greater accessibility and data security.
  4. Data retention or regular access to old data is required by the customer.
  5. An application is first built on the cloud while a data center facility is being procured, and with a migration plan to move that application onto a private host.

Since cloud provides pay-as-you-go model based on resource usage, an application that is consuming compute resources and network bandwidth continuously could be too expensive for public cloud compared to private cloud hosting with a proper visualised environment with modern network and security systems monitored and managed by skilled resources.

The value of the data center remains compelling.

Depending on product strategy, roadmap, and service delivery demands, an enterprise may find itself a good candidate for on-premise application hosting. The benefits include:

  1. Cost controls on management/monitoring/maintenance. Sometimes the total-cost-of-ownership of certain applications is cheaper on-premises versus the cloud.
  2. Accessibility and control of data can protect intellectual property and improve security parameters for the application
  3. Improved performance and availability of the application by ensuring better visibility and usage of the system.
  4. Increased availability of evolving technologies to modernise and improve without depending on the cloud-specific service availability.
  5. Avoiding vendor lock-in for a particular cloud provider.

If any of these criteria apply, it's time to examine a plan for repatriation.

Making the move with minimal disruption.

First, the best candidates for cloud repatriation are applications that are built to be modular. These are cloud-agnostic or cloud-independent and don't rely on customisations from the underlying infrastructure. Microservices architecture is designed to handle use cases like these.

Custom applications ideally should be built leveraging modern microservices architecture for maximum flexibility without using cloud-specific services like serverless technology or RDS to avoid cloud lock-in.

A note about traditional application design: Migrating traditional legacy applications to cloud without modern architecture typically does not provide much cost savings or agility. This is why it's critical that enterprises simplify their application architecture using cloud-agnostic microservices architecture to maintain necessary flexibility and agility to repatriate back to an on-premises data center with minimal customer impact.

Watch out for pitfalls

Of course, as with any digital transformation initiative, there are liable to be challenges. Applications leveraging cloud-specific services and APIs could have repatriation concerns due to customisations or provider-specific features.

Enterprises with a lack of skilled resources or proper infrastructure (like modern data center technology)  could have challenges in maintaining performance and availability.

It's also important to note that the transition does not make sense for companies born in the cloud. These companies never had a data center or the requisite skilled resources to handle these transitions or to provide global, secure access with high availability and performance.

For these enterprises, it may make sense to continue leveraging public cloud but consider modern microservices architecture to avoid cloud vendor lock-in and migrate between infrastructure options. Only then can an organisation be truly modern and hybrid together.