Single service comparison

AWS Device Farm is fundamentally an execution-hosting service: it owns the device/browser fleet and returns run artifacts, with thin test-case metadata layered on top. CodeArts TestPlan is the inverse: a test management backbone (requirements traceability, test case libraries, test plans, iterations, mind-map and spreadsheet case authoring, review workflows) that orchestrates execution through CodeArts Deploy and external runners, not a hosted real-device farm. So Device Farm substitutes for the fleet, while TestPlan substitutes for the planning layer; mapping AWS Device Farm to TestPlan alone drops the device execution capability, which Huawei does not provide as an equivalent managed farm.

API and data models diverge sharply. Device Farm exposes devicefarm:* APIs (scheduleRun, GetRun, ListArtifacts) and integrates with Appium/Espresso binaries as the unit of work, with artifacts pushed to S3. CodeArts TestPlan exposes REST endpoints scoped to projects/iterations for cases, suites, and execution records rather than device scheduling, and stores results inside CodeArts object storage with no public artifact bucket model. Integration points differ too: Device Farm hooks into CodePipeline/CodeBuild and CI runners; TestPlan hooks into CodeArts Pipeline/Deploy, requiring re-platforming automation rather than API translation.

Scaling, region, and operational responsibility differ. Device Farm offers unmetered device slots and metered device-minutes across multiple AWS regions with AWS-managed capacity, HA, and fleet patching. CodeArts TestPlan is seat-based and project-scoped in Huawei regions, with Huawei managing the control plane; capacity for parallel execution depends on the customer's runner fleet (ECS/CCI or self-managed) rather than a Huawei-provided device pool. The customer therefore retains responsibility for execution infrastructure, runner health, and concurrency licensing that Device Farm abstracts away. Note also that AWS has deprecated remote-access and limited new feature investment in Device Farm, narrowing long-term parity assumptions.

Migration is not a like-for-like port. Treat TestPlan as the test-management successor for planning, case libraries, and traceability, and accept that real-device execution must be re-homed on a commercial device cloud, Huawei CCI/ECS runner fleet for simulators/emulators, or partner MaaS providers through custom automation. There is no Huawei service that ingests Device Farm schedules or artifacts, and no documented source-to-target migration tooling, so the move is rebuild rather than lift-and-shift.

Assess scope boundaries first. Inventory Device Farm usages split into two categories: (a) test management metadata—test suites, runs, pass/fail history, pipeline gates—and (b) device/browser execution and artifact storage. Confirm with Huawei regional sales whether CodeArts TestPlan is available in your target region and which seat tier covers projected users, because TestPlan is project-and-seat metered with no device-minute analog. Validate that Huawei CCI or ECS compute can host the Appium/XCUITest/Espresso runner images you need, since Huawei provides no managed device farm equivalent.

Re-home execution before migrating metadata. Stand up runner infrastructure (ECS/CCI images with Appium, device emulators, or partner device clouds) and wire it to CodeArts Deploy and CodeArts Pipeline. Re-platform CI hooks: replace AWS CodePipeline/CodeBuild Device Farm actions with CodeArts Pipeline stages invoking the new runner, and redirect artifact storage from S3 to OBS. There is no automated Device Farm-to-TestPlan conversion; test scripts and capabilities must be re-pointed to the new runner endpoints and target device lists.

Migrate test management metadata via CodeArts TestPlan import. TestPlan supports importing test cases from files, so export Device Farm suite metadata and run history, normalize supported fields to TestPlan's case and iteration model, and import them into matching TestPlan projects. Re-establish traceability links to CodeArts Req and CodeArts Pipeline so pass/fail gates and release reports regenerate. Plan for historical run-data loss: Device Farm artifacts (videos, logs, screenshots) do not migrate and must be archived in OBS if retained for audit.

Validate parity and adjust the cost model. Run the Device Farm suite and the TestPlan-driven runner in parallel across representative device/browser matrices; accept that pass rates may shift due to different device OS levels and runner versions. Recalculate TCO on Huawei terms: TestPlan seat fees plus CCI/ECS runner compute, OBS storage for artifacts, and CodeArts Pipeline execution minutes, compared with AWS device-minutes or unmetered slots. Document residual gaps—no hosted real-device farm, no remote interactive access in TestPlan—as accepted risks before production cutover.