Workload Migration is the act of moving a workload – typically a program or service – from one infrastructure environment to another, for example from on-premises datacenter to the public cloud, from one cloud provider to another, or from the cloud back to on-premises infrastructure.
When organizations undertake a cloud migration, databases, backup and restore procedures, and workloads are migrated from on-premises servers to one or more cloud providers. This can be done for many reasons, chief among them to utilize highly scalable infrastructure managed by a cloud provider, to locate workloads in specific global regions, to reduce to cost of fixed capacity infrastructure, to use a pay-per-use cost model, or to take advantage cloud-native services that may not be available on current infrastructure.
Although not every workload is appropriate for a cloud migration, the best candidates include:
Businesses across all industries reply on applications to drive digital transformation efforts. They want to modernize their fleet of applications. Most have a desired future state in mind, which is typically composed of cloud-based, containerized, microservice-based applications created to run with DevOps processes, often across multiple clouds. While the vision is often well-defined, the path to the vision is not. There are huge technical, organizational, and operational hurdles to overcome and businesses find these challenges are often more difficult and time-consuming than originally expected.
Workload portability and migration is at the heart of application modernization and enterprise cloud adoption. The ability to migrate applications enables organizations to take advantages of the features and price-performance models of multiple cloud providers and enables migration back to on-premises infrastructure if that makes the most sense for business or technical reasons.
Without workload migration, organizations could be locked-in to a single cloud provider and at the mercy of their pricing, policies, and performance characteristics.
Organizations must ensure they have the right skill sets before starting workload migration. Consideration should be given to training and education on the new target environment to ensure services can be properly managed and applications will run smoothly, since cloud providers operate very differently from on-premises data centers and local virtual machines (VMs). New security protocols must be adopted to ensure both cloud-based and on-premises components are secure end to end.
The compatibility of source and destination platforms, and selection of best fit-for-purpose migration tools both have a big impact on the speed and cost of migration efforts.
Although migration steps will vary by organization and by workload, there are some commonalities for most migrations, including:
- Cloud Provider selection
- Cost evaluation for ROI and TCO
- Performance and security requirement evaluation
This application rationalization process typically has five possible outcomes:
- Refactor: The application is rewritten, typically to a microservices architecture
- Replatform: Usually from VMs to containers and often to public cloud infrastructure
- Rehost: Migrate “as-is” to the Cloud
- Replace: Usually with a SaaS version of the needed capabilities
- Retain: Keep an application on its existing on-premises infrastructure.
- Retire: Eliminate the application from the portfolio
As businesses rationalize their applications, various considerations for that application may necessitate a certain cloud as its destination. Some applications, such as those from Microsoft, may run best on Azure, while others may want to take advantage of Google Cloud Platform’s AI capabilities. Still, others may be replaced with SaaS applications that run in only one cloud. Typically, these decisions are made by each application team based on their application’s individual needs. This naturally leads to multi-cloud proliferation. Because of this, multi-cloud is a reality for every business as they rationalize and modernize applications.
Once the method of migration is selected, bandwidth calculations should be made to determine if the initial transfer of data and VMs is done over the network or offline. Very large data transfers that could stifle other network traffic might be better performed by shipping a disk instead.
Next, the workloads should be stress-tested to ensure that performance is satisfactory at projected loads and to prevent unpleasant surprises.
After physical migration, the major effort shifts to management, including tracking performance and usage. These cloud workload migration tools are often overlooked.
Substantial savings can be realized by migration of workloads to a cloud provider. Organizations only pay for what they use, and do not have to make significant infrastructure purchases or upgrades. Instead, cloud providers perform upgrades and updates to their infrastructure as part of their business model, and organizations gain the benefit of those advances simply by maintaining their cloud workloads.
Many organizations find they can reduce real estate expenditures and operating costs associated with power and cooling my migrating workloads to the cloud. Recently, Deloitte found organizations spend almost two-thirds of their IT budget on maintenance; these expenses are absorbed by cloud providers and instead organizations pay a predictable monthly fee for their workloads.
2. Scalability and Workload Balancing
Cloud providers simplify the means to scale up and down based on changing demand and business factors. Additionally some organizations use workload balancing strategies as part of their migration plan, enabling load balancing between on-premises and cloud, between clouds, or all of the above.
Cloud providers can be more secure than on-premises infrastructure, if organizations understand that there is a shared responsibility and that both provider and user must do their part. As organizations increasingly adopt zero-trust security strategies, cloud workloads benefit from the most stringent physical security policies in practice today. Since cloud providers are inherently multi-tenant, offering services to financial, medical, and government clients worldwide, they must demonstrate the most stringent security practices and meet the broadest range of government regulatory practices.
Most cloud providers also offer many built-in security features like security analytics, periodic updates, cross-enterprise visibility, and keeping unwanted traffic from accessing the machines on which your workloads reside.
The very nature of cloud workloads means they can be accessed from anywhere there is a secure network connection to the cloud. Many cloud migrations are undertaken just for this benefit.
Anywhere, anytime, any device access is a key feature of digital transformation. Additionally, cloud-based backup or archives help speed restores, and can drive the recovery point objective (RPO) and recovery time objective (RTO) after a data loss or failure to near zero levels.
Organizations that wish to modernize their applications with microservices and APIs are by definition taking a cloud-native approach to development and deployment. Modern, containerized applications are born, deployed, and enhanced in the cloud, and businesses that take advantage of application modernization are more likely to retain both employees and customers through more interactive, richer applications.
Not every application performs as expected in the cloud, which amplifies the need for stress-testing each and every application when migrated.
Latency can deep-six some applications. Others may be more costly to operate than originally anticipated, whether due to CPU utilization or data egress charges associated with API usage or report generation.
User error can also thwart workload migration. For example, choosing the wrong AWS instance type is a common error - each instance should specify the correct amount of CPU, memory, network connectivity, and storage for the workload.
Other common challenges to address during workload migration include:
- Interoperability issues with other cloud-based and on-premises workloads
- Backup and business continuity issues to reduce or eliminate downtime
- Security, especially for cloud-native workloads that are coupled to other app workloads
- Performance impacts due to latency
- Choosing the right cloud provider for desired functionality
- Which migration strategy to apply to each workload
Workload migration is complicated, requires detailed plans and either in-house or outside expertise to achieve a successful migration.
First determine whether the workload migration will be performed by in-house staff or via the use of third-party workload migration services offered by cloud providers and/or their partners.
Ensure that workloads are suitable for migration and that there are clear, measurable goals for undertaking the migration, such as improved scalability, lower cost, or better performance.
Once costs and cloud providers have been determined, consider how it will perform after migration, including whether bandwidth is adequate or whether application dependencies could complicate matters.
Consider whether re-architecting the workload could extend its useful life and consider migration tools that can greatly accelerate workload migration including VMware Cloud on AWS.
Also remember that some workloads may not be appropriate for migrating to the cloud. Consider all aspects of the execution environment and ensure that given the service parameters promised by the provider that the same levels of capacity, performance, utilization, security, and availability can be achieved. If not, the workload might best be left on-premises.
Finally, consider how the cloud infrastructure addresses compliance, including regulatory factors including HIPAA, PCI, and GDPR. Ensure you understand your current workloads and determine how closely their requirements, both for present and future evolution, can be satisfied.