Alibabacloud Migration Mas Product Mapping

Cloud Product Mapping Skill (Multi-Vendor + Instance-Level Spec Mapping + Real-Time Pricing)

Map source cloud vendor (AWS / Tencent Cloud / Huawei Cloud / Azure) resources to Alibaba Cloud with precision, outputting a complete comparison table of product names, instance specs, usage, and reference prices.

When to Use

• User has resource inventory/bills/survey reports from other cloud vendors and needs mapping to Alibaba Cloud products and specs
• Before migration planning, need to estimate costs on the Alibaba Cloud side
• User specifies source cloud vendor and needs product mapping relationships and migration order

Supported Vendors

Cross-vendor common rules (Database / Redis / MongoDB / FC / Disk): references/common-spec.md Alibaba Cloud Pricing API Index (33 products): references/pricing-api.md

Prerequisites

Runtime Environment

• Python: >= 3.9 (scripts use f-strings, type annotations, dict | None syntax)
• OS: macOS / Linux / Windows (verified on macOS 15 / Ubuntu 22)
• Disk: >= 200 MB for Python dependencies and intermediate results

Dependency Installation

cd .qoder/skills/product-mapping/scripts
pip install -r requirements.txt

Main dependencies:

Alibaba Cloud Credentials

This Skill relies on the Alibaba Cloud default credential chain for authentication. NEVER explicitly handle AK/SK credentials.

The pricing service uses the default credential provider chain which resolves credentials in the following order:

1. Environment variables (ALIBABA_CLOUD_ACCESS_KEY_ID / ALIBABA_CLOUD_ACCESS_KEY_SECRET) — set by the runtime environment automatically
2. OIDC Role ARN
3. ECS RAM Role (Instance metadata)
4. Credential file (~/.alibabacloud/credentials)

No manual credential configuration is needed. The runtime environment provides credentials automatically.

Mandatory Rules:

• When user request includes price estimation, cost comparison, or real-time pricing, start the pricing service normally
• The pricing service resolves credentials automatically via the default credential chain
• If credentials are not available in the environment, the service will return appropriate error responses
• MUST still send standard requests to local pricing service (e.g., POST /api/pricing/batch) to record call traces in api.json
• Only when user explicitly says "mapping only, no pricing" can pricing be fully skipped, with report annotated "Pricing not enabled"

Prohibited: Hardcoding AccessKey in scripts; explicitly reading/exporting AK/SK in Skill instructions; using root account AccessKey.

Required Permissions

For detailed RAM permission policies and API permission reference, see references/ram-policies.md.

Input Format

Supports any one or more of the following formats:

Format 1 - Plain Text:

  "3 台 EC2 m5.xlarge，2 个 RDS MySQL db.r5.large（500GB），1 个 S3 桶约 2TB"
  "4 台腾讯云 CVM S5.LARGE8，1 个 CDB MySQL 4C8G"

Format 2 - JSON:

  [
    {"vendor": "aws",     "service": "EC2", "type": "m5.xlarge",  "count": 3},
    {"vendor": "tencent", "service": "CVM", "type": "S5.LARGE8",  "count": 4}
  ]

Format 3 - Excel Survey / Cloud Vendor Bill CSV

When vendor is not specified, auto-detect from product names in context (EC2/S3 → AWS, CVM/COS → Tencent Cloud, ECS/OBS → Huawei Cloud, VM/Blob → Azure).

Execution Steps

1. Confirm Source Cloud Vendor

Confirm which source cloud vendor (AWS / Tencent Cloud / Huawei Cloud / Azure) the user wants to map. If unclear, prompt user to choose directly; do not assume defaults.

2. Confirm Alibaba Cloud Target Region

Must confirm with user before pricing; different regions have significant price differences.

• Extract region info from source data (e.g., AWS ap-southeast-1 = Singapore)
• Show source region to user and ask for Alibaba Cloud target region
• Common Alibaba Cloud regions: cn-beijing, cn-shanghai, cn-hangzhou, cn-shenzhen, ap-southeast-1 (Singapore), ap-southeast-5 (Jakarta)
• Do NOT default to cn-beijing; must be explicitly specified by user
• Pass confirmed region as --region parameter to excel_mapping.py and pricing service
• Post-execution validation: After script completes, must verify the "target region" field in output file matches the --region value exactly; if mismatch (e.g., falls back to cn-beijing), prompt user to check script parameter parsing or force override ALIBABA_CLOUD_REGION_ID

3. Product Name Mapping

Load corresponding product mapping table from references/mappings/ based on source vendor. Table structure: Source Product | Alibaba Cloud Product | Recommended Migration Tool | Layer | Complexity.

For source products not in the mapping library, match by category to similar Alibaba Cloud products and annotate with "AI Recommended, please verify manually".

4. Instance Spec Mapping

Load spec mapping detail table by source vendor:

• AWS → references/aws-spec.md (EC2 instance families, ARM Graviton, Intel/AMD, spec size conversion)
• Tencent Cloud → references/tencent-spec.md (CVM all generations, CDB MySQL, TDSQL-C, Redis, MongoDB)
• Huawei Cloud → references/huawei-spec.md (ECS including Kunpeng ARM, RDS, GaussDB, DCS Redis, DDS)
• Azure → references/azure-spec.md (VM all series, Database for MySQL/PG, Cache for Redis, Cosmos DB)
• Cross-vendor common rules → references/common-spec.md (RDS / Redis / MongoDB / FC / Disk / Others)

4.1 Spec Family Normalization Rules

• ARM instances: Map uniformly to Yitian ARM series (c8y / g8y / r8y)
• GPU instances: T4 inference → ecs.gn7i, V100 training → ecs.gn6v, A100 → ecs.gn7e, H100 → ecs.gn8
• Bare metal: ebmg7 / ebmc7 / ebmr7 / ebmgn7e

4.2 Spec Family Auto-Fallback

When mapped spec family is unavailable in target region, fall back in this priority:

1. Same-generation substitute (e.g., g8i → g7)
2. Same-ratio substitute (e.g., g7 → c7 when 1:2 fits)
3. General-purpose fallback (ecs.g7)
4. Special: ecs.t6-c1m2 only available at large size; xlarge+ must fall back to ecs.c8y.xlarge

4.3 Disk Configuration (Required Reading)

• System disk: ESSD PL0 uniformly, minimum 20GB
• Data disk >= 20GB → ESSD PL1; < 20GB → must downgrade to ESSD PL0 (PL1 minimum is 20GB, otherwise API error)
• ECS pricing must include both system disk and data disk

5. Alibaba Cloud Real-Time Pricing

5.0 Pre-Pricing Service Start (Mandatory)

Before calling any pricing API, must perform:

1. Start pricing_service.py — it resolves credentials automatically via the default credential chain
2. MUST call /api/pricing/batch or corresponding product pricing endpoints
3. MUST NOT skip API calls under any circumstances
4. If credential chain resolution fails:
   • The pricing service will return 401/403 or preset reference prices
   • Caller handles per Section 5.3 error table
   • Inform user: "Credential chain resolution failed, pricing API returned error; ensure the runtime environment provides valid credentials"
5. Only when user explicitly replies "no pricing" or "mapping only" can pricing be skipped, with report annotated "Pricing not enabled"

5.1 Start Pricing Service

cd .qoder/skills/product-mapping/scripts
pip install -r requirements.txt
# macOS note: AirPlay occupies port 5000, recommend port 5001
PRICING_PORT=5001 python pricing_service.py

Service exposes pricing capabilities for 33 products (see references/pricing-api.md).

5.2 Batch Pricing (Recommended)

POST multiple products at once to /api/pricing/batch to significantly reduce latency. Full examples in references/pricing-api.md.

5.3 Response Handling and Error Processing

All pricing service responses are unified JSON. Caller must implement the following; otherwise cost estimates will be inaccurate:

Mandatory Requirements:

• Timeout: HTTP client timeout 30 seconds (PolarDB DescribeClassList is slow in large regions)
• Retry: 429 / 5xx / timeout use exponential backoff, max 3 retries
• Fallback: All failed pricing items must be annotated "Pricing failed / Reference price missing, please verify manually" in report; silent discard prohibited
• Credential missing is NOT a fallback scenario: When AccessKey not configured, must interrupt and prompt user

5.4 Pricing Coverage

Detailed list with product_code/parameters for each product: references/pricing-api.md

6. Output Complete Mapping Report

6.1 Resource Mapping Summary Table

| # | Source Product | Alibaba Cloud Product | Recommended Migration Tool | Layer | Migration Complexity | Notes |

6.2 Cost Comparison Estimate

| Resource Type | Source Spec x Qty | Alibaba Cloud Spec x Qty | Source Monthly Cost (ref) | Alibaba Cloud Monthly Cost (ref) | Savings % |

6.3 Cost Summary

┌──────────────────────────────────────┐
│ Alibaba Cloud Monthly Cost Estimate  │
├──────────────────────────────────────┤
│ Compute: ECS             ¥ xxx       │
│ Database: RDS/PolarDB    ¥ xxx       │
│ Cache: Redis/Tair        ¥ xxx       │
│ Storage: OSS+Disk+NAS    ¥ xxx       │
│ Network: SLB+CDN+EIP     ¥ xxx       │
│ Other: Kafka/FC/SLS etc  ¥ xxx       │
├──────────────────────────────────────┤
│ Monthly Total:           ¥ xxxxx     │
│ Annual Est.(85%):        ¥ xxxxx     │
│ 3-Year Est.(50%):        ¥ xxxxx     │
└──────────────────────────────────────┘

7. Output Migration Priority Recommendations

Ordered by risk from low to high:

1. Batch 1 (Low Risk): Network infrastructure (VPC, Subnet, Security Group, NAT, Load Balancer, CDN, DNS)
2. Batch 2 (Low Risk): Object storage (S3/COS/OBS/Blob → OSS)
3. Batch 3 (Low Risk): Security & Identity (IAM → RAM, KMS, WAF, DDoS, Certificates)
4. Batch 4 (Medium Risk): Compute instances (EC2/CVM/ECS/VM → ECS), using SMC whole-machine migration
5. Batch 5 (Medium Risk): Databases (RDS/Aurora/TDSQL-C/GaussDB/Azure SQL → RDS/PolarDB), via DTS sync
6. Batch 6 (Medium Risk): Cache & Search (Redis / Elasticsearch)
7. Batch 7 (High Risk): Message queues (Kafka / RocketMQ / RabbitMQ / Service Bus)
8. Batch 8 (High Risk): Containers & Serverless (EKS/TKE/CCE/AKS → ACK, Lambda/SCF/FunctionGraph/Functions → FC)
9. Batch 9 (High Risk): Big Data & AI (EMR / Redshift / Synapse → MaxCompute / EMR, SageMaker → PAI)

Excel Batch Processing Mode

Read resource survey Excel via script, auto-complete product mapping, spec mapping, real-time pricing, and output results to Excel.

cd .qoder/skills/product-mapping/scripts
pip install -r requirements.txt
PRICING_PORT=5001 python pricing_service.py &
python excel_mapping.py <input.xls> --output result.xlsx --pricing-url http://localhost:5001 --region <user-specified-region>

Input: Excel file with each Sheet representing one AWS product (EC2 / S3 / ElasticCache / DocumentDB / AuroraDB / NAT / Load Balancer / EIP / Lambda / Athena / EventBridge / SNS / CloudWatchLog / Other).

Output: *_mapping_result.xlsx with 10 Sheets (Product Mapping Summary, EC2 Spec Details, Database Details, Cache Details, Storage Details, Network Details, Other Services, Cost Summary, Migration Priority, Coverage Report).

Practical Tips: ECS deduplicates pricing by (spec, system disk, data disk, PL) tuple; system disk minimum 20GB; data disk < 20GB downgrade to PL0; t6 burst xlarge+ falls back to ecs.c8y; recommended port 5001.

Credential Requirement: The pricing service resolves credentials automatically via the default credential chain. Always start pricing_service.py and call /api/pricing/batch; do not skip API calls. If 401/403 returned, annotate per Section 5.3 error table.

Mapping Rules

• Prefer Alibaba Cloud official migration tools (SMC / DTS / Online Migration Service)
• When no official tool available, recommend mature open-source solutions (redis-shake, MirrorMaker, etc.)
• Specs matched at equal or slightly higher performance to avoid degradation
• Aurora / TDSQL-C / GaussDB users strongly recommended PolarDB (architecture compatible)
• ARM instances uniformly map to Yitian ARM series (c8y / g8y / r8y)
• Complexity assessment: Low=config migration, Medium=data+config migration, High=requires app refactoring
• Uncertain mappings must be annotated "AI Recommended, please verify manually"

Observability

User-Agent Template

All HTTP requests made by pricing_service.py to Alibaba Cloud OpenAPI MUST include a User-Agent header following this template:

AlibabaCloud-Agent-Skills/alibabacloud-migration-mas-product-mapping/{session-id}

Session-ID Generation Rules

• Generate one UUID v4 per Skill invocation as the session-id
• The same session-id MUST be used across all API calls within a single user interaction
• Format: lowercase UUID without hyphens (32 hex characters), e.g., a1b2c3d4e5f6a7b8c9d0e1f2a3b4c5d6
• The session-id MUST be passed as an environment variable SKILL_SESSION_ID when starting pricing_service.py

export SKILL_SESSION_ID=$(python3 -c "import uuid; print(uuid.uuid4().hex)")
PRICING_PORT=5001 SKILL_SESSION_ID=$SKILL_SESSION_ID python pricing_service.py

Disclaimers

Pricing Disclaimer

• All Alibaba Cloud prices output by this Skill are public reference prices, not contract prices.
• Actual prices are affected by: business discounts, enterprise negotiation, tiered billing, long-term commitment discounts, exchange rate fluctuations, vouchers, and promotional offers.
• Real-time API prices from BSS / ECS / RDS / Redis / PolarDB are based on Alibaba Cloud official pricing at the time of call; results are not commercial commitments.
• Some products (PolarDB-X, ClickHouse, Lindorm, GDB, Tablestore, etc.) use estimation logic rather than real-time API pricing, with lower accuracy.
• Mandatory: All reports must append at the end: "Prices in this report are for reference only; final pricing is subject to the contract signed with Alibaba Cloud sales."

Competitor Spec Mapping Disclaimer

• Source vendor (AWS / Tencent Cloud / Huawei Cloud / Azure) spec mappings are based on public technical docs and market research, not official recommendations from source vendors.
• Different CPU generations, memory ratios, and network performance may differ between sides; performance equivalence does not guarantee 100% consistency.
• Deprecated, newly released, or region-restricted instance families may not be covered by the mapping library; results will be annotated "AI Recommended, please verify manually".
• ARM instances (Graviton / Kunpeng / Ampere) mapped to Alibaba Cloud Yitian ARM require user testing for app compatibility (some natively compiled apps may need rebuilding).
• GPU instance models (T4 / V100 / A100 / H100) are strictly matched; cross-model mapping is prohibited due to large performance differences.

Data Source Declaration

• Vendor spec info: Referenced from each vendor's official documentation (as of 2025 Q1)
• Alibaba Cloud pricing: Obtained via Alibaba Cloud OpenAPI in real-time or referenced from Alibaba Cloud official public pricing
• Mapping rules: Based on practical experience from Alibaba Cloud migration team

Directory Structure

.qoder/skills/product-mapping/
├── SKILL.md                      # This file (entry point and workflow)
├── references/                   # Spec mapping details and pricing API index
│   ├── mappings/                 # Product-level mapping data
│   │   ├── aws.md
│   │   ├── tencent.md
│   │   ├── huawei.md
│   │   └── azure.md
│   ├── aws-spec.md
│   ├── tencent-spec.md
│   ├── huawei-spec.md
│   ├── azure-spec.md
│   ├── common-spec.md            # Cross-vendor common rules (RDS/Redis/MongoDB/FC/Disk/Others)
│   └── pricing-api.md            # Pricing API index + error handling
├── scripts/                      # Python pricing service and Excel batch processing
│   ├── requirements.txt
│   ├── pricing_service.py        # Flask pricing service (port 5001)
│   ├── excel_mapping.py          # Excel batch processing entry point
│   ├── azure_mapping.py
│   ├── huawei_mapping.py
│   ├── price_comparison.py
│   ├── tc_price_comparison.py
│   └── hw_price_comparison.py
└── evals/                        # Evaluation test cases