Cloud modernization strategies - Lift-and-shift migration versus temporary burst to cloud

Learn about cloud modernization strategies.

Organizations employ two complementary strategies for hybrid cloud adoption: workload migration and cloud bursting. While traditional migration involves the permanent relocation of applications and datasets to the cloud for modernization, cloud bursting dynamically extends a private data center into a public cloud. This provides temporary, on-demand compute to handle demand spikes, scaling down as capacity needs subside.

These two strategies coexist. Migration is a long-term approach for modernizing to cloud-native workloads, whereas bursting provides immediate compute elasticity for workloads that are retained on-premises, bypassing physical hardware procurement cycles.

To enable native cloud bursting capability, Cloudera is introducing Cloudera Hybrid Environments. This new architecture is designed to provide cloud agility while leveraging existing infrastructure. It facilitates dynamic workload movement—as distinct from planned migration—by natively discovering and accessing datasets on-premises, all while strictly adhering to established on-premises security and governance protocols.

Feature	Cloud Burst / Remote Access strategy	Migration/Data Replication Strategy
Primary Goal	Elastic scaling and temporary peak load handling. Extends existing on-premises resources temporarily.	Resilience, modernization, and geo-availability. Moves workloads closer to the cloud environment.
Data Residency	Data stays on-premises. Only processing or temporary application components move to the cloud.	Data exists in two or more locations (on premises and cloud) simultaneously.
Workload Type	Stateless or loosely coupled workloads. Compute-intensive tasks, batch processing, web tier scaling (for example, VDI or render farms).	Stateful workloads. Databases, file systems, applications requiring fast, local data access (for example, ERP or CRM).
Latency Tolerance	High tolerance for latency. Performance depends heavily on the network connection to the on-premises data center.	Low tolerance for latency. Data must be accessible quickly by cloud applications (since the data is local to the cloud environment).
Implementation	VPNs, dedicated network links (for example, AWS Direct Connect or Azure ExpressRoute), and firewalls configurations to allow remote access to on-premises databases.	Data synchronization tools (for example, Cloudera Replication Manager), SAN replication, or specialized software-defined storage.
Cost Driver	Network egress fees. Moving large amounts of data from the cloud back to on premises is expensive.	Cloud storage costs. Paying for duplicate storage space and for the replication service itself.
Management Complexity	Network management. Ensuring a stable, low-latency, high-bandwidth connection between sites.	Data consistency management. Ensuring data integrity, conflict resolution, and synchronization lag between the two copies.
Disaster Recovery (DR)	Poor or non-existent. The cloud is only used for compute, if the on-premises data center fails, the application fails.	Excellent. The replicated data can be used to instantly spin up services in the cloud (Active-Passive or Active-Active DR).