In this expert guide on cloud migration in Asia-Pacific, we highlight the key considerations and strategies that enterprises can take when moving their workloads to public cloud
At Flybuys, Australia’s biggest loyalty platform, migrating its legacy systems to the cloud has improved efficiency and customer experience, enabling it to deliver new offers from partners to its customers much faster than before.
Its journey to the cloud took a year, during which it modernised its user applications, data and analytics platform and rebuilt its technology stack to support real-time transactions and insights, among other tasks.
While companies such as Flybuys have successfully transitioned to cloud, achieving the business outcomes that it had set for itself, a successful cloud migration remains an elusive goal for most companies.
According to a global study by Accenture, 52% of organisations fail to attain their expected cloud outcomes, often because they see cloud migration as a one-time exercise and a static destination, says Duncan Eadie, managing director and cloud, infrastructure and engineering lead for Asia-Pacific, Middle East and Africa at Accenture.
“Our research has shown that enterprises that treat cloud as a new operating model to continuously reinvent their businesses realise greater business value that goes well beyond cost savings,” he says. “They are 3.7 times more likely to innovate and achieve 1.5 times more cost reduction as compared to their peers.”
One cloud operating model developed by Contino, a consulting firm, comprises components such as team structures, technology and business capabilities, as well as building blocks including ways of working, whether it is agile or lean, as well as culture and talent, among others.
“A big part of achieving the goals that you want to achieve with cloud is not only about tools and techniques; it’s also about your operating model and strategy,” says John Knight, transformation principal at Contino.
When it comes to strategy, Paul Migliorini, CEO at Versent, a cloud services consulting firm that helped Flybuys make the move to Amazon Web Services (AWS), says organisations should not try to take a legacy enterprise and transplant it into a cloud context.
“Successful transformation and modernisation means thinking about the organisation and its needs as a whole,” says Migliorini, a former AWS executive. “If we approach cloud migration from that strategic point of view, we can optimise the opportunity for growth because we’re not going to be held back either by the constraints of pre-cloud platforms or the limiting thinking that goes with that.”
Strategic considerations
A cloud migration strategy, says Eadie, cannot be developed in isolation and needs to consider the company’s most critical business objectives while balancing short-term budgets with medium- and long-term rewards.
And once you have determined your desired business objectives, an application-led approach to migration strategy should be used to optimise value in achieving these goals.
An analysis of applications that make up a business service would look at the technical aspects (is your application cloud compatible or is there an opportunity for modernisation?), financial aspects (how much will it cost to migrate/transform, what run-rate savings can I expect?) and organisational aspects (do I have the skills I need and how would my processes need to change to achieve business benefit?)
Eadie says that beyond the basics of whether a workload can be technically migrated (what is your target architecture and what is your relative costs?), there are a number of industry or architecture specific factors that will influence decision-making. These include compliance regulations, software licensing regimes and required data flows, all of which can have a dramatic effect on cost if not properly solutioned for.
Workload assessment
Every cloud migration effort should begin with a cloud readiness assessment of an organisation’s current state. A high-level scan can help identify quick wins that can enable you to get started and realise value early through a sprint-based approach, but investment in a more in-depth analysis of critical applications and complex, integrated workloads before accelerating migrations will pay dividends later.
Ultimately, a decision on whether or not to migrate comes down to cost and value. Any business process can be delivered on the cloud, but the question for non-standard circumstances is how much an organisation is willing to spend to achieve this.
According to Eadie, the criteria to consider in assessing which workloads can be moved to cloud include:
If you can invest more time and money into your migration programme, then replatforming to containers is a great strategy that delivers more bang for buck, both in terms of ongoing operating cost reductions and speed to market
Duncan Eadie, Accenture
Application architecture: Most x86 workloads can simply be rehosted (although there may be benefits from doing more); other workloads which need replatforming should take the transformation cost into account when determining the business case.
Application licensing: Some software suppliers have commercial models which are not designed for the cloud (or certain non-proprietary clouds) and at face value would lead to a significant cost increase should their software be migrated. The commercial implications need to be assessed and potentially renegotiated.
Compliance: Industries and geographies with data localisation laws may have limited options for target state architectures and solutioning workarounds in jurisdictions which do not have a public cloud availability zone, or a readily available private cloud option may damage the business case. Additionally, many banking regulators for example are required to sign off on migrations before they occur, and this can be a time-consuming process.
Data flows: Cloud providers can charge high prices for data egress so if an enterprise needs to use a hybrid or multi-cloud architecture, such as for compliance reasons, then careful consideration needs to be given to how data moves between databases or warehouses, applications for processing and output for business use.
Skills: The business case can only be fully realised if the organisation has the necessary skills required to operate in the cloud and can evolve to adopt new processes and ways of working, such as DevSecOps to drive agility.
There is also a timing consideration. If a refresh or decommissioning initiative is coming up, an enterprise may find that it pays to wait and align its move to the cloud with those projects.
Lift-and-shift, refactor or replatform?
Migliorini says that many organisations tend to do “lift-and-shift” cloud migrations, which are appealing only because it’s the quickest way. But organisations will still need a refactoring process, where business processes are broken down into smaller chunks and supported by cloud-native applications. Those two activities can happen in parallel so they support each other rather than interfering and creating additional problems to solve.
Eadie that says while refactoring monolithic systems can deliver substantial value, it can also be time consuming and expensive. That’s where replatforming can help to achieve a good balance between modernisation, speed and transformation cost.
With replatforming, applications are moved into the cloud with a few “upgrades” that remediate any underlying infrastructural inefficiencies. These may include operating system upgrades, database and middleware remediation or use of containers to simplify operations and reduce cloud consumption spend.
Ultimately, Eadie says that choosing between lifting-and-shifting, replatforming, rearchitecting applications to cloud-native through refactoring, or even completely reimagining business processes, depends on an enterprise’s business circumstances and rationale for starting their cloud journey.
For example, if your back is against the wall with a pressing deadline such as a forced datacentre exit or a merger and acquisition (M&A), Eadie says that lifting-and-shifting to emphasise speed is likely to be appropriate. This is also the cheapest migration approach, but it will also leave you with the highest ongoing costs.
“If you can invest more time and money into your migration programme, then replatforming to containers is a great strategy that delivers more bang for buck, both in terms of ongoing operating cost reductions and speed to market,” he says.
“For companies that are moving to cloud in order to drive business innovation, modernising applications should be the priority focus. This involves not just rewriting applications for speed, agility and customer-centricity using cloud native technologies, but also fundamentally rewiring and realigning the core of the company’s technology, processes, and people.”
Cloud migration tools
Kalyan Madala, chief technology officer at IBM’s technology group in ASEAN, says that besides the usual cloud-native tools in areas such as DevSecOps and GitOps, there are two sets of tools organisations should consider.
“One of the biggest gaps in most cloud journeys is the lack of automation in decision making and data collection about existing applications, infrastructure and their inter-dependencies,” Madala says, adding that IBM has a cloud advisory tool that analyses an organisation’s application portfolio and identifies opportunities to optimise and automate IT infrastructure.
There are also tools for application modernisation, where much of the work is still done manually. “This is an evolving space, so look out for opportunities for leveraging artificial intelligence and automation for greater agility, velocity and risk management,” says Madala.
With data, you’ve got security, compliance and data governance, which are not really being tackled that much in the cloud
Michael Ewald, Contino
Versent uses a cloud management platform developed by its partner, Stax, which leverages its methodology for planning and executing cloud transformation projects.
“Stax unlocks AWS cloud opportunities faster because its automation tools remove the typical constraints of building or maintaining an environment. That means it’s easier to follow best practices, and you can easily maximise time and cost efficiency, without sacrificing security,” Migliorini says.
Versent also deploys a host of other migration tools, including Flexera which collects real-time data for capacity and integration modelling, as well as Dynatrace to capture before-and-after migration data to ascertain if it is on target to achieve business objectives for its clients.
Ultimately, cloud migration tools and services should address every step of the cloud journey, from discovery, assessment and planning to governance, migration, assurance and operations, says Deepak Gadkar, head of AWS business unit at Tata Consultancy Services (TCS) Asia-Pacific.
He singled out TCS’s proprietary automation tools and accelerators to reduce the time to market and deliver differentiated outcomes under its Machine First Delivery Model, a structured approach to digital transformation that automates all business process activities that can be automated and turns them into intelligent processes.
The data challenge
As with any migration effort, organisations have to be prepared to deal with challenges that stand in their way. Moving data to the cloud, in particular, can often be complex, risky, expensive and time-consuming, says Gadkar.
Furthermore, Gadkar notes that the lack of expertise and understanding of data can lead to errors that generate higher unexpected costs. Incomplete and inaccurate data backups can also pose a serious threat and lead to critical process failures and loss of crucial and confidential organisational data.
Indeed, according to a recent global study by Contino, only 16% of organisations have fully realised the business benefits of moving data to cloud – even though over 60% consider their cloud data programmes mature.
Michael Ewald, director of technology at Contino, says that the data challenge is far more acute for larger organisations that are looking to move on-premise workloads and data to the cloud as they have to ensure that existing operational systems are kept going during the migration process.
One way organisations can address the cloud data challenge is through DataOps, an agile approach to designing, implementing and maintaining a distributed data architecture that will support a wide range of open source tools and frameworks in production.
Similar to DevOps, DataOps breaks down organisational silos between development, operations, data science and data analytics teams to enable organisations to get the most out of their data.
“With data, you’ve got security, compliance and data governance, which are not really being tackled that much in the cloud,” Ewald says, noting that this is despite the fact cloud suppliers offer automation tools to help organisations eradicate manual compliance processes.
“But to do that, you’ve got to think about your operating model and organisational structure so that even if your architect may not be fantastic in the data realm, you still have a data steward and data owner who are part of the team from a controls perspective,” he says.
Switching cloud providers
In 2019, Singapore-based logistics startup Ninja Van switched to Google Cloud from AWS and found that its applications were running faster than before. The move had also enabled it to reduce its cloud bill.
Such moves are uncommon, however, as the major cloud providers have, at their core, fairly similar features and pricing, and the others quickly innovate to catch up where competitive differences in capability or cost emerge for one platform.
Besides this, there are also a number of economic factors reducing the desirability to do so, says Accenture’s Eadie.
“For architectures using cloud-native microservices, platform lock-in is strong and you would most likely find yourself simply rewriting your applications directly on your new platform which can be a time consuming and expensive undertaking for a large enterprise,” he says.
By contrast, Eadie notes that migrating a virtual machine or container-based architecture from one cloud to another is “theoretically quite simple, but nevertheless is still unusual in the real world”.
“Data is one major reason for this. Data gravity – pooling your critical data for shared usage – is important for deriving business value, thus large data volumes would likely be involved in any platform migration and the cloud providers charge a high cost for data egress.”
Versent’s Migliorini notes that organisations will want to get their cloud migration right the first time and choosing the right cloud platform is going to be the first important decision they will have to make.
“We don’t see companies changing cloud providers because we’ll counsel our clients to choose carefully and start with a platform that’s not just going to do what they need now but be able to deliver ongoing benefits for decades to come.”
