adv

HKCSView: Cloud adoption is a progressive journey

In Hong Kong, most startups have already adopted public clouds by default, but many large enterprises still tend to be conservative towards cloud computing.

For business executives, cloud adoption is no longer a go-or-no-go decision; they have been using clouds for certain applications, such as, public-facing websites, mobile apps, and office productivity tools. However, many of them still hesitate to further deepen cloud adoption.

Hesitation and challenges

The hesitation often comes from data security concerns, skills and organization mismatch of the current IT, and the incompatibility of the capex (capital expenditure) oriented IT financing model with the cloud pay-as-you-go opex (operational expenditure) model.

To enable the latest innovations, such as artificial intelligence (AI), for digital transformation success, cloud computing is a prerequisite. Using the cloud is not only for increasing the IT operational efficiency and cost savings, but is also critical for the enterprise to stay competitive in the market.

Instead of a big bang process, cloud adoption should be a progressive journey, especially for large enterprises.

Cloud computing is a paradigm shift for IT in many ways: from operation-orientation to service-orientation, from technical architecture to enterprise architecture, and from capex to opex. It could take years to go through not only the technology transformation but more importantly the business transformation. The key is to formulate a strategy and roadmap to start the cloud journey as soon as possible. The journey can be a three-stage progression: cloud-ready, cloud-first and cloud-native.

Being cloud-ready

Many enterprises have reached the cloud-ready stage to certain extent. Virtualization is the key technology behind cloud computing. It is common for an enterprise to use virtual machines (VMs) to deploy applications. However, in addition to virtualizing computing, the enterprise should also seek to virtualize storage, networking, and other data center resources for full infrastructure virtualization. This is also called software defined data center (SDDC).

Beyond the infrastructure layer, the enterprise should also make use of ‘cloud-friendly’ software platforms. In the virtualized infrastructure, computing resources are mostly commoditized; in other words, they are not designed for extreme performance. For example, a VM is often less powerful than a high-performance physical server; a database system that can only scale up the capacity through adding more CPUs and memory to a single physical server is not cloud friendly. A cloud-friendly platform must be able to scale out with a distributed architecture, i.e., expanding capacity through adding VMs and virtual disks, which can be provisioned across multiple servers.

In addition, some non-cloud-friendly software packages are still licensed with charging the number of CPU cores in a physical server that run the software. In a cloud-ready environment, enterprises should seek to select software packages that are licensed on a pay-per-use basis, e.g., number of virtual CPUs, number of users, and storage capacity. In particular, the open-source licensing model is commonly a better fit for the cloud environment.



Comments