Legacy modernization is being accomplished in several ways. Applications can be subdivided into a number of smaller service components and each service is being rewritten using a modern programming language while guaranteeing the legacy application's functionality. Even the original architecture of the legacy application can be changed using the latest architectural styles, such as SOA, EDA, or a combination of both. The legacy applications currently running on centralized and mainframe servers can be modernized to run on completely new platforms and infrastructures, such as cloud environments. Thus, the modernization strategy development and planning are done by taking several principles, goals, technologies, and tools into account. At different layers and levels, the required modernization gets done in a risk-free and strategically sound manner: 

• Refactoring typically refers to reorganizing an application's source code in order to bring some clarity and changes. Suppose the code written isn't modular, or the code has a lot of duplicated code as it's been written by different teams over the years of its development and maintenance. The integration with newly-acquired company software may also bring in some complexities. There are several other reasons to modernize. So, factoring out the code in order to fulfill some new requirements (business, technology, and users) is being touted as code refactoring. 
• Rearchitecting and redesigning both mean the same thing. Legacy applications can be modernized by rearchitecting the application using newer technologies, such as middleware solutions, database systems, execution platforms, visualization toolkits, and software infrastructure solutions. Also, with newer patterns for designing application components, redesigning is gaining prominence. With these architectural and design changes, new applications can have higher performance, scalability, security, availability, resilience, and sustainability. With powerful message-oriented middleware platforms and newer databases, such as NoSQL and NewSQL, modernization is a continuous affair to satisfy evolving business needs and to ensure the delight of customers. 
• Replatforming involves taking the legacy application to be deployed and running it on newer platforms and infrastructures in order to accrue the platform's and infrastructure's benefits. 
• Rewriting the application source code using modern programming languages is done in order to experience the distinct advantages of the new programming language. For example, Ballerina is a new programming language and is considered by some as the best-in-class language for coding microservices. This language simplifies service, data, and application integration. 

Here's a list of a few best practices while performing modernization:

• We must be sure that the refactoring technique works. After taking an easy-to-do application component out of the legacy application under transformation, then we must embark on refactoring or rewriting; that is, we can refactor the source code of the application component to convert it into a microservice. Otherwise, we can rewrite the code to craft a microservice without changing the functionality of the original application component. Then, the idea is to take the version that's very close to the final implementation.
• There's no need to refactor or rewrite the other components of the legacy application. Use the stub mechanism to check whether the refurbished or rewritten component mimics the old behavior; that is, it's mandatory to establish a seamless and spontaneous connectivity for the existing application components (not yet refactored or rewritten) with the newly-formulated service to understand whether there's any deviation or deficiency. This way, little wastage of time is guaranteed for legacy modernization, which happens to be a time-consuming job.
• Once a microservice is readied, it has to be taken to one or more software repositories (public and/or private) to enable access and usage by software developers and testers. This kind of centralized storage also contributes for source-code version control. There are continuous integration (CI) solutions to integrate, build, and test software services. With the overwhelming adoption of the enterprise DevOps concept, business applications get continuous and consistent integration, delivery, and deployment in order to take applications and services to their end users quickly, without any risk.
• With the unprecedented popularity of the containerization movement, containerizing software applications has been getting more attention. Containerizing microservices brings a number of technical advantages to the table. There are write-ups and articles detailing how to quickly deliver and deploy microservices individually and collectively. With container clustering, orchestration, and management platform solutions emerging, the containerization paradigm is becoming penetrative, participative, and pervasive. There are a litany of automated tools to augment the various activities associated with cloud orchestration, realizing multi-cloud container applications, and infrastructure as code. The convergence of containers and microservices opens up a growing array of innovations, disruptions, and transformations. There are integrated platforms that enable the dream service era.
• Application components have to be chosen for refactoring or rewriting using some priority factors. Not all components are suitable for modernization. There are a few important parameters required to do the prioritization properly. The key parameters to be considered include the ease of extraction, the roadmap, and possible risks. Some of the components may become obsolete. Thus, prioritization plays an important role in accomplishing the modernization job in an affordable, risk-free, and strategically-sound manner.
• Data virtualization is the way forward for application modules to correspond realistically with the data sources of the legacy application. It's therefore recommended that instead of focusing on data-structure transformation, focus on creating and using appropriate proxies that can transform data from different databases into a standardized format. This standardization helps the newly-created microservices (whether through refactoring or rewriting) to connect and interact with the data without any major hitch.

Microservices are small and accommodative. Fixing defects, security holes, and vulnerabilities is quite easy with microservices. Performance increments, root cause analysis, and unified threat management form composite applications that are business-critical, process-aware, and cloud-enabled. The faster deployment of microservices in production environments is another positive factor. Specific microservices in microservices-centric applications can be replaced with newer ones, whereas this sort of facility isn't available in monolithic applications. Adding features is therefore a smooth affair.