The Lifecycle of a Software Being
The life cycle of a digital data platform is surprisingly humanlike. It is “born” (designed and built), then it goes through “infanthood” while it is incrementally improved as per the desired business scenarios. The platform enters its golden age, or adolescence, when business users extract value from it. After a period of time, it gradually becomes obsolete: the system can no longer handle the workload and/or accommodate new data sources and use cases; it is cumbersome to add new functionality and difficult to hire and onboard staff to support it. Refactoring and band-aid patches could keep it up-and-running for some time, but eventually, it becomes too resource- and time-consuming to maintain it in a sound technical condition. Maintenance measures could hardly prevent the platform from deprecating (you’ve got the metaphor).
The good news is that some data platforms reincarnate in cloud-native applications.
Why Would Your Data Platform Benefit from Cloud Migration
Migration to cloud is a great way to save a data platform from becoming obsolete, exit an on-prem data center, or change a software licensing strategy. Very few companies these days are ready to invest a fortune in large and complex on-premise data platforms. These companies are, as a rule, bound by strict regulations and data security standards. The rest of the companies can enjoy the benefits the cloud offers: flexibility and manageability, resilience, scalability, cost optimization… The list goes on.
Though migration of a data platform to a cloud offers numerous benefits, do not blindly rest all your hopes on it. No cloud by itself would magically re-architect your workloads or optimize costs. In fact, spending on cloud hosting may exceed the price you are used to paying for on-prem infrastructure. To save yourself from a sticker-shock, choose an optimal approach for migrating a data platform to the cloud.
Pick Your Approach for Cloud Migration
An optimal migration strategy should be chosen based on the specifics of the data platform, such as its technology stack, age, the data sources used, as well as the business requirements and future roadmap for data management and analytics in an organization.
In this post, we focus on 3 out of 6 Rs, six common application migration strategies identified by Gartner and Amazon: rehosting (or lift-and-shift), replatforming, and refactoring (or re-architecting). We’ll help you choose the right migration approach depending on a combination of attributes of your data platform.
Cloud migrations are often complicated by the underlying architecture or restrictive legacy technologies of the data platform, requirements of business users, and peculiarities of tools offered by different cloud providers.
- Rehosting, or Lift-and-Shift Approach to Migration
Rehosting or lift-and-shift is the least complicated, relatively fast, cost-efficient, and not resource-intensive approach to migrating data platforms to the cloud. The system is simply re-hosted in the cloud as is or with minimal changes. Rehosting can be automated with dedicated tools offered by the cloud provider.
Which companies is this approach suitable for?
This approach best suits an organization that wants to quickly evacuate a data center, rapidly increase server storing capacity, or quit a license agreement with a hosting provider. For some companies, the cloud serves as a secondary data center. By a lift-and-shift migration, a company can buy some time for planning a pervasive cloud adoption strategy and prepare the ground for transforming the system into cloud-native. The other approaches — replatfroming and rearchitecting — will be easier to accomplish when the platform already resides in a cloud.
The lift-and-shift approach by itself has limited applicability. It is feasible for data platforms driven by certain business requirements, like being cloud-agnostic. Some organizations mentioned earlier, like those in banking, finance, insurance, or healthcare domains, which comply with strict legal regulations for storing sensitive data can lift-and-shift their data platform to store the actual data in a private cloud and be able to quickly transfer it back to on-prem or to a different cloud, if need be.
- Replatforming, or Lift-Tinker-and-Shift Approach to Migration
Replatforming, sometimes called “lift-tinker-and-shift”, refers to a partial upgrade of the data platform and all associated data with the purpose of cloud optimization and transforming an application to be partially cloud-native. This approach is relatively cost-effective as it does not involve major changes in business logic or architecture of the data platform. Many companies, as they start their cloud journey, prefer to start small and have the cloud environment grow as their data platform scales up, so they opt for this approach.
Which companies is this approach suitable for?
Replatforming would best suit companies that migrate their relatively modern data platforms and are looking to optimize a certain component in it, e.g., data storage, ETL/ELT, analytics. Before this approach to migration is chosen, the company should conduct its cloud readiness assessment: make an inventory of all existing tools and infrastructure components in use and check if each is cloud-compatible or requires modification, then evaluate available cloud-native and/or cloud-based technologies-as-a-service. This way, the strategy, scope, budget, and timeframe of the cloud migration become less elusive. Plus, the company gains a clear idea of whether re-architecting would be required in the foreseeable future.
- Re-architecting, or Refactoring After Migration
Re-architecting is a more advanced cloud migration approach. It presupposes a complete transformation of the data platform into a cloud-native one. This approach involves a re-engineering of the entire data landscape, sources, and all system components. It is quite time- and resource-intensive. But this approach is the most rewarding as far as return on investment, and long-term business outcomes are concerned.
Which companies is this approach suitable for?
Re-architecting best suits companies with data platforms that do not keep up with the business needs, do not scale up or lack performance capacity. To plan this migration, stakeholders and migration team must start with an in-depth discovery of business goals, processes and needs in terms of data consumption and analytics. Based on the findings, they design a new architecture, while still keeping the legacy platform up and running. It is not deprecated until the new, cloud-native platform is built, and all data is migrated from the legacy system.
Conclusion
There is no one-size-fits-all approach to cloud migration, and hybrid approaches are common. Lift-and-shift is often just a curtain-raiser for a more advanced cloud transformation. Some companies take a super cautious approach and migrate their dev/test environment first before the live one is transitioned.
Learn how Cloud helps businesses turn threats into opportunities
At DataArt, we know a journey to the cloud requires a comprehensive discovery phase, sometimes even architecture re-design, security analysis, and integration work before cloud technologies allow your data platform to work like a charm. You may contact our experts if you plan a cloud migration.
The article was written by Oleksii Kovalchuk and originally published at blog.dataart.com.
