Cloud Migration Considerations
Cloud Migration Considerations
Beyond choosing a migration type and strategy, a real migration plan has to account for platform fit, cost, availability, vendor risk, and the compute, storage, network, and management realities on the other side of the move.
Platform Compatibility Considerations
Most cloud service providers offer plenty of platform options to host virtual machines, websites, applications, databases, and other resources — but it's still worth confirming the platform requirements are actually met.
Confirm CPU, GPU, and memory allocations match the application's needs.
Confirm capacity and throughput are sufficient for the workload.
Confirm the networking features the application depends on are supported.
Place services near consumers to reduce latency and improve experience.
Confirm the provider's controls meet your organization's requirements.
Cost Considerations
Cost always plays a part in selecting a cloud host. While planning the migration, work with the service provider to understand and optimize cost — and don't be afraid to compare offerings from various providers. Many organizations today host resources with different providers just to optimize cost.
Weigh guaranteed availability against the steep discounts of interruptible spot capacity.
Plan for surging demand so you pay for capacity when you need it, not year-round.
Match instance size to actual usage instead of over-provisioning "just in case."
Service Availability Considerations
Cloud service providers don't always make the same solutions available in all regions, due to political, economic, or infrastructure factors — so compare service availability between providers. That said, migrating to the cloud generally increases service availability over what's practical with on-premises physical systems.
Redundant systems are often more economical to implement in the cloud, and failover solutions may be more cost-effective. If the data centers in Europe become unavailable, customers could still reach redundant resources in Asia or North America — though not all services are available in all regions or from all vendors.
Vendor Lock-In Considerations
Lock-in occurs when the cost — whether time, money, or resources — is too high to switch from the current CSP to a different, desired provider. Avoiding it can be difficult given the dominance of GCP, AWS, and Azure. Click to expand each example.
You may hear the term "data gravity" — the size of stored data causing decisions around applications, services, and CSPs to be made with the data itself as the critical, immovable factor.
As a cloud expert, make sure decision-makers and architects ask:
Industry standards help avoid lock-in:
Ways to guard against lock-in:
Compute Considerations
Compute resources — a combined allocation of CPU, GPU, and memory — let a VM process data. Performance needs vary by application, so check the program's documentation for the vendor's suggested configuration; many software vendors now provide cloud-specific deployment guidance directly.
Migration considerations include initial specifications, but remember that a key benefit of cloud VMs is on-demand scalability.
Set the boundaries within which an instance can grow or shrink automatically.
Dedicated, spot, on-demand — pick the model that matches the workload's pattern.
Tag resources to monitor usage and attribute cost accurately.
Make sure instance resizing is available if requirements change later.
AWS, Azure, and GCP all offer preconfigured instance types to get started — even for migrations.
Storage Considerations
Storage migrations can be challenging simply due to the quantity of data involved — transfers can take a very long time, which matters when the data needs to stay available. CSPs offer a variety of services to help.
Admin time, downtime, bandwidth
Admin time, downtime, transfer time
How much downtime results
Right tool for the transfer type
Protected throughout the process
Data integrity means ensuring data hasn't changed unexpectedly, such as during a network transfer — use reliable transfer methods that confirm everything sent was received, and consider hashing to verify integrity. For very large transfers where the internet connection can't keep up, consider a storage appliance: Google, for example, ships a Transfer Appliance you load locally and mail back to be transferred into your storage buckets.
Networking Considerations
Network considerations begin with employee access to the internet — expect this traffic to grow dramatically as employees stop reaching on-premises servers and start relying on cloud-hosted resources.
Ensure plenty of low-latency bandwidth is available.
Confirm reliable 24×7 access; consider a second ISP to remove a single point of failure.
Use strong network encryption — lots of confidential traffic will now cross this link.
Restrict communication between virtual networks — e.g. isolate dev from production.
Design access control for network administration — a distinct skill set from system or app admin.
Design an IP scheme for cloud network resources to manage and control communications.
Ensure customers can reach your cloud resources to use and purchase your products.
Carefully control exactly which cloud-based services and data customers can view.
Management Considerations
Beyond the traditional technical and system administration concerns, questions around management tools, data sovereignty, availability, and cost all impact the migration process.