| name | databricks-to-huawei-cloud-skill |
| description | Use this skill when migrating Databricks tables, notebooks, SQL warehouse flows, or Spark pipelines to Huawei Cloud. It helps analyze the source workload, map Databricks patterns to OBS plus MRS Spark and Hive, generate sanitized migration scripts, validate result parity, and compare execution behavior without relying on environment-specific details. |
Databricks to Huawei Cloud Skill
Overview
Use this skill for Databricks-to-Huawei Cloud migration work where MRS is the default compute target for Spark workloads and OBS is the preferred object storage target. It is optimized for notebook migration, open-format export, functional parity testing, and repeatable migration execution.
Quick Start
Follow this sequence by default:
- Inspect the Databricks source workflow.
- Separate business logic from Databricks-specific platform behavior.
- Prefer
Parquet as the interchange format and CSV only for raw/simple inputs.
- Map the source into
raw/bronze -> standardized/silver -> analytical/gold.
- Rebuild the execution path on
MRS Spark + Hive.
- Compare row counts, aggregates, business metrics, and run time.
- Document operational gaps such as object-store authorization separately from logic migration.
Workflow Decision Tree
Start with the migration shape:
- Simple file-to-table flow:
- Read source files from Databricks
Volumes or tables.
- Export to
Parquet.
- Recreate the managed or external table pattern on MRS.
- Read references/migration-patterns.md.
- Multi-step notebook with joins, features, or rules:
- Treat notebook cells as staged Spark transformations.
- Preserve analytical intent, not notebook UX.
- Rebuild as one Spark job or a small job chain on MRS.
- Read references/migration-patterns.md.
- Result validation or performance comparison:
- Migration failure or mismatch:
Core Rules
- Preserve business semantics first. UI parity is not the goal.
- Default target pattern:
Databricks source -> open files -> OBS -> MRS Spark -> Hive tables -> curated Parquet
- If
OBS access is blocked during migration or validation, continue with an HDFS or local-node fallback to validate logic.
- Treat
OBS agency or temporary-credential issues as operational blockers, not logic blockers.
- Keep all scripts and examples sanitized:
- use placeholders such as
<bucket>, <catalog>, <schema>, <warehouse_http_path>, <mrs-master>
- never copy hostnames, usernames, passwords, tokens, access keys, project IDs, or customer names
Default Deliverables
When using this skill, prefer producing:
- a migration mapping from Databricks patterns to MRS patterns
- a minimal Databricks-side export or reconstruction script
- a minimal MRS-side Spark job template
- a parity-check metric table
- a short gap list:
- functional gaps
- operational gaps
- performance caveats
Script Use
Use the bundled templates when you need a starting point:
scripts/databricks_sql_warehouse_pattern.py
- generic SQL warehouse execution and metric collection pattern
scripts/mrs_spark_submit_pattern.py
- generic MRS Spark job pattern with Hive output and metric emission
scripts/compare_metrics_template.py
- generic exact-match and delta comparison for Databricks vs MRS metrics
Reference Use