Cloud migration to AWS, Azure or Google Cloud

The 6 R's, one decision per workload

Each application gets one strategy. Forcing every system through a lift-and-shift, or every system through a rewrite, is how migration programs stall.

• ✓Rehost. Move with as little change as possible. This is the fastest route out of a data center when the application can tolerate it.
• ✓Replatform. Keep the business logic and change the platform underneath: managed databases, containers or a platform service.
• ✓Repurchase. Replace the product with a SaaS service that already performs the function, after a data-export plan is agreed.
• ✓Refactor. Re-architect for cloud-native scale, only where the business case pays for the engineering.
• ✓Retire. Switch off what nobody uses. Discovery often finds applications that still cost money and licences.
• ✓Retain. Leave it where it is, on purpose, and record the reason and a review date.

• The 6 Rs of cloud migration

Discovery, TCO and wave planning

Nothing moves until discovery has a written output: an inventory, a dependency map, a total-cost comparison and a wave list. Discovery counts servers, licences, databases, file shares, identity, scheduled jobs and the people who know how each one runs. Dependencies set the wave order.

We do not pick a cloud first and then fit the workload to it. Discovery scores each application against the same criteria, and the result is written down per application:

• ✓Data location rules and contracts you already hold
• ✓Managed-service fit, such as a database engine your team already runs
• ✓Skills your team can operate after handover
• ✓Estimated run cost on each platform, with the migration cost counted
• ✓Exit path: how the workload would leave the platform later

If the best answer for a workload is to stay where it is, the plan says so.

Data location is a design input, settled before any copy is made. Data-protection rules apply as applicable in your jurisdiction, including where personal data may live and how transfers are documented. No hyperscaler region is in Bangladesh, so if a workload must stay in a given country, discovery answers that question first. A cloud partner does not make you compliant with any regulation; your obligations stay with you.

Our partner status is listed so you can check it yourself:

Regions in our current working set are AWS Mumbai (ap-south-1) and Singapore (ap-southeast-1), Azure Central India, South India, West India and Southeast Asia, and Google Cloud asia-south1, asia-south2 and asia-southeast1. Other regions are confirmed per engagement.

The sequence, and what you still own

Each cloud has its own migration tooling, such as AWS Application Migration Service, Azure Migrate and Google Cloud Migrate to Virtual Machines. We use the native tool where it fits, and tool choice is made per wave.

Downtime approach and rollback

We do not promise zero downtime for the whole estate. Each wave has a cutover window, a rollback trigger and a named owner who signs it off. Some waves are near-zero because replication keeps the copy current. Others need a declared freeze, and we write its length into the plan.

Rollback has three stages, and each one costs more than the one before:

• ✓Before cutover. Traffic switches back. This is the cheapest stage, and little data is at risk.
• ✓After cutover, before the new system takes writes. The switch is reversed inside the window agreed for the wave.
• ✓After the new system accepts writes. Rolling back means a reverse sync or a restore from backup. It takes longer, and writes made after cutover may need reconciling. The wave plan states that exposure in hours and in data before cutover begins.

Some changes do not reverse cleanly: schema changes, third-party callbacks, DNS caches at clients and licences tied to a host. We list those in discovery so they are decided before the window opens, not during it.

Timelines by workload size

These are planning ranges for a funded team, measured in calendar time rather than effort-days. The quote follows discovery and replaces these ranges with the timeline for your estate.

Funding and what the move costs

AWS Migration Acceleration Program (MAP) funding may apply to qualifying migrations. MAP is an AWS program, and eligibility and any funding are set by AWS for each engagement. We do not quote a co-funding percentage or a credit amount.

The professional-services price is a fixed quote after discovery. The run cost after cutover follows your cloud agreement, or partner billing where a contract includes it. Where partner billing applies, compute and storage pricing has typically come in 15 to 35 percent below list, and CloudFront bandwidth up to 40 percent lower. These are estimates, not a quote guarantee. Use the AWS pricing calculator for an indicative monthly run rate; it is not an invoice.

Costs that belong in the total-cost model: dual running during waves, licence mobility, extra storage for replicas, and the people who still test each wave. Leaving them out is how a cheap-looking lift becomes an overrun.

Quotes are in USD and quote-based. Other currencies are available by contract only; ask. Taxes apply as applicable in your jurisdiction and the contract.