Tencent Cloud Migration Guide

A systematic, end-to-end playbook for migrating your workloads to Tencent Cloud — from discovery to go-live, built on 200+ real-world migrations across retail, fintech, gaming, and Web3.

Migration Phases

40+

Migration Tools

262

Cloud Service Mappings

200+

Projects Delivered

What This Guide Covers

Your journey from assessment to optimization

Tencent Cloud provides a proven, four-phase migration methodology backed by a mature 40+ toolchain, purpose-built for enterprise-scale migrations from IDC, AWS, GCP, Aliyun, and Huawei Cloud. This guide walks you through every step — from building your Landing Zone, selecting the right tools, executing cutover, to post-migration optimization.

📋 Assess & Plan

Discover your workloads, assess cloud readiness, design your target architecture, and build a Landing Zone.

🛠️ Migrate

Use purpose-built tools for VMs, databases, storage, containers, and big data. Validate with POC and dry-runs.

🔀 Cutover & Optimize

Execute cold or hot cutover with rollback plans. Validate, then continuously optimize with TSA.

Why Migrate to Tencent Cloud

Strengths in SE Asia, gaming, video, and fintech

Regional Coverage

Region	City	Code
Singapore	Singapore	ap-singapore
Indonesia	Jakarta	ap-jakarta
Thailand	Bangkok	ap-bangkok
Japan	Tokyo	ap-tokyo
Korea	Seoul	ap-seoul
India	Mumbai	ap-mumbai

Industry Strengths

Gaming & Social: Powering Honor, Riot, Supercell — millions of DAU
Video Streaming: Tencent's own LVB/CDN backbone
E-commerce: CP Group, Lotus's, Heytea migrations
Fintech: APEX exchange hybrid cloud, MAS compliance
Cost: SA5 AMD servers, in-line/offline co-location — 43%+ cost reduction

Cloud-Native Candidate Check

Good candidates for cloud have three traits: (1) no special hardware dependency, (2) horizontal scalability, (3) compute/storage separation with non-latency-sensitive inter-module communication.

Migration Scenarios

Scenario	Description
Cloud Migration	IDC / competitor cloud → Tencent Cloud
DR Migration	Disaster recovery setup
Cross-Region	Multi-region deployment for proximity
Hybrid Cloud	Rapid expansion with unified management

CAP Migration Theory

Type	Trade-off	Characteristic
CA	Low cost + No downtime	Longer migration timeline
CP	Low cost + Short timeline	May cause service interruption
AP	No downtime + Short timeline	Requires higher investment

Team Training & Enablement

Build Tencent Cloud capability at project kickoff — not after migration

⏰ Start Training at Project Kickoff

Team training is not a post-migration activity. Your IT team should build Tencent Cloud capability before the migration begins, so they can participate in design reviews, execute migration tasks, and independently operate the cloud environment post-handover. Tencent Cloud provides 14 courses across three certification levels (L1 Practitioner → L2 Associate → L3 Professional).

Migration-Specific Courses (Must-Recommend)

Two courses are directly aligned with the migration methodology in Section 04 (4 Phases Workflow). Enroll your team at project kickoff.

🎓 Tencent Cloud Solutions Architect Professional Engineer

Chapter V: Cloud Migration

For solution architects designing target architecture. Covers high-reliability, high-availability, high-scalability, high-performance, and loosely-coupled architecture design — with Chapter V dedicated entirely to cloud migration methodology.

Level	L3 Professional
Duration	Chapter V: 90 min (all chapters ~2 days)
Audience	Solution Architects, Cloud Architects
Prerequisite	Cloud Computing Associate (CCJ)

🔗 View Course Details →

🎓 Tencent Public Cloud Delivery Associate — Cloud Migration

Certification: TMA

Migration best practices end-to-end: target environment design, workload/storage/database migration alignment, design verification, post-migration optimization. The primary certification for migration engineers and delivery personnel.

Exam	60 MCQ · 90 min · 70% pass
Fee	$150
Online videos	Free (self-paced)
Language	English
High-weight domains	Storage migration design (10%) · Migration problems (10%)

🔗 View Certification Details →

Recommended Certification Path (Aligned to Migration Phases)

Phase 1: Discover & Assess (Kickoff)

Cloud Computing Associate (CCJ) + TMA

Foundation for all roles. Complete TMA online videos before assessment begins. Encourage team to take the $150 exam before cutover.

Phase 2: Design

Solutions Architect Professional (CAS) — Chapter V

Architects complete CAS (including Chapter V: Cloud Migration) to design target architecture with Tencent Cloud best practices.

Phase 3: Implement

Role-specific Professional certs

DBA → Database Professional · DevOps → DevOps Professional · SRE → SysOps Professional. Each role builds deep expertise for their migration tasks.

Phase 4: Validate & Handover

Independent operation capability

Client team certified and capable of independent cloud operations — reducing vendor dependency post-acceptance.

TMA Exam Content (16 Domains — High-Weight First)

Domain	Weight
Cloud component migration design - Storage	10%
Problems in cloud migration	10%
Cloud product common issues - Compute	8%
Migration design - Network	8%
Account design	7%
Migration strategy	7%
Database migration design	7%
Other component migration design	7%
DB common issues	7%
Cloud migration basics	5%
Requirement survey	5%
Storage common issues	5%
Compute migration design	5%
Other process	3%
Network common issues	3%
Other common issues	3%

Training Delivery Formats

Format	Description	Price
Online self-paced video	Pre-recorded, anytime, free	Free
Online instructor-led (virtual)	Live video + virtual labs	$6,000/day
On-site instructor-led	Client site classroom + dedicated lab	$6,000/day

Training portal: https://www.tencentcloud.com/edu/training

4 Phases Migration Workflow

4 Phases · 10 Steps · 4 Milestones · Templates per Step

Tencent Cloud migration follows a structured four-phase approach. Each phase contains specific steps (10 in total), with a defined R/S responsibility split and downloadable template files to accelerate execution.

How to Read This Table

Phase = one of four stages · Step = numbered workflow step (1–10) · R/S = who leads (see legend) · 📎 Template = click to download the supporting Excel/Word file

R = Responsible (leads) S = Support (assists)

Four-Phase Overview

The entire migration lifecycle from kick-off to acceptance. Each phase has a clear lead, scope, and exit criteria — scroll down for the detailed 10-step workflow with templates.

Phase 1 · Tencent Leads

Discover & Assess

Collect business requirements, map system architecture, scan resources, assess cloud readiness, and identify dependencies & risks.

Steps 1–2 · Exit: Assessment Report signed

Phase 2 · Tencent Leads

Design

Design Landing Zone, migration strategy (cold/hot), network/security/account/DB/storage plan, cutover & rollback plan, POC verification.

Step 3 · Exit: Technical Plan + Migration Plan confirmed

Phase 3 · Client Leads

Implement

Provision environment via Terraform, execute data migration (DTS/MSP/go2tencentcloud), testing, dry-runs, and performance validation.

Steps 4–7 · Exit: Test & Performance Report approved

Phase 4 · Client Leads

Validate & Handover

Execute cutover, 5-day stabilization, TSA optimization, DR drill, and formal acceptance sign-off.

Steps 8–10 · Exit: Signed Acceptance Report

Responsibility Shift

Phase 1–2: Tencent Cloud leads (R) — assessment, architecture design, and solution development.
Phase 3–4: Client leads (R) — environment deployment, migration execution, cutover, and acceptance. Tencent provides support (S) throughout.

Step	Activity	Description	Completion Criteria	Templates
1	Project Preparation	Tencent assembles delivery team; confirm scope, on-site conditions, communication protocol. Client assigns PM and key contacts.	Team ready; conditions met	SOW & RACI Summary
2	Research & Assessment	Collect app architecture, tech stack, dependencies, resource utilization, performance metrics. Use Cloud Migration (CMG) console as the primary platform — launch `mspcli` scanning from CMG to discover resources across source clouds.	Research Report + Assessment Report signed off	Information Collection (P1, 13 sheets) ð View Online BigData Info Collection (P1) ð View Online

Phase 1 Deliverables: System Information Research Report + Cloud Migration Assessment Report

★ PRIMARY RECOMMENDED TOOL

🚀 Tencent Cloud Migration (CMG) — All-in-One Migration Platform

Cloud Migration (CMG) is Tencent Cloud's centralized migration platform that integrates all Tencent-native and certified third-party migration tools into a single console. It provides end-to-end visibility across discovery, sync, and cutover — the recommended starting point for every migration project. Free of charge — you only pay for the resources consumed during migration.

📊 Centralized Monitoring

All migration tasks (hosts, databases, storage) tracked in one dashboard, grouped by project.

🖥 Visualized Operations

Real-time charts and status views from multiple perspectives — holistic progress at a glance.

🔌 Ecological Aggregation

Integrates Tencent tools (mspcli, go2tencentcloud, DTS, MSP) + certified partner solutions.

✅ For Discovery & Assessment (P1)

Launch mspcli resource scanning from the CMG console
Auto-discover hosts, databases, and storage across source clouds (AWS / GCP / Alibaba / IDC)
Aggregate scan results into project-level inventory views
Estimate migration time and resource costs before execution

🚀 For Migration Execution (P3)

Run host migration (go2tencentcloud), DB sync (DTS), storage migration (MSP) from one platform
Track each task status: Pending → Running → Success / Failed
Batch-migrate by project group; schedule across time windows
Identify bottlenecks early via centralized alerting

🔗 Product Page → 📚 Documentation Free platform · Pay only for resources used during migration

💡 Alternative: Billing Export for Resource Discovery (AWS / GCP)

In addition to mspcli scanning, you can export cloud billing data (CSV) to discover resource usage, costs, and distribution across accounts/regions/projects. This is especially useful when API-based scanning is limited by permissions or when you need historical cost analysis.

☁️ AWS: Cost & Usage Report (CUR 2.0)

Console: Billing & Cost Management → Data Exports → Create
Export type: Standard data export → Data table: CUR 2.0
Config: Check Include resource IDs · Time: Monthly
Format: gzip-text/csv · S3 bucket destination
Historical data: Current month only via Data Exports. For prior months, contact AWS TAM or submit a Support Center case with the export name, target month, S3 bucket, and CSV format request. Allow ~24h.

Key columns (15 recommended): product_servicecode, product_usagetype, line_item_usage_account_id, line_item_unblended_cost, product_location, line_item_line_item_type, pricing_unit, product_operation + 7 more (114 total available)

☁️ GCP: Billing Report (SKU-level)

Console: Search "Billing" → Cloud Billing → Report
Group by: SKU (aggregates by SKU Name + ID)
Time range: Select "Last month" or custom range (supports past 13 months)
Filters (optional): By Service, Region, Project, or Cost Type
Download: Click "Download CSV" button (top-right) · Max 10,000 rows per export

CSV fields: SKU ID, SKU Description, Service Name, Region, Project ID, Usage Amount, Usage Unit, Cost (Discounted), Total Cost (Pre-Discount), Discount Amount, Tax

Prerequisite: Billing Viewer (project) or Billing Account Viewer (account level) IAM role · Data delay: 24-48h · For resource-level detail use BigQuery Export

📄 AWS Export Guide (full steps + screenshots) 📄 GCP Export Guide (full steps + screenshots)

Comparison	AWS CUR 2.0	GCP Billing Report
Data scope	Full CUR 2.0 (114 columns)	SKU-level aggregated (11 key fields)
History	Current month (history needs AWS support)	Past 13 months
Resource-level detail	Yes (check "Include resource IDs")	Use BigQuery Export for resource-level
Export limit	First export within 24h	10,000 rows per CSV (use filters to split)

Step	Activity	Description	Completion Criteria	Templates
3	Solution Development	Detailed migration plan: Landing Zone design (8 dimensions), migration strategy (cold/hot), network/account/security/DB/storage design, cutover & rollback plan, POC verification, investment outline.	Migration Technical Plan + Migration Plan confirmed	Migration Plan (P0, 9 sheets) ð View Online Investment Outline (merged into P0) ð View Online Cloud Adoption Solution Landing Zone & IAM

Phase 2 Deliverables: Cloud Migration Technical Plan + Cloud Migration Plan (with WBS)

Step	Activity	Description	Completion Criteria	Templates
4	Environment Preparation	Provision cloud resources (CVM, DB, storage, network) via Terraform. Set up Landing Zone, monitoring, security baseline.	Resources ready	Migration Tools (P0) ð View Online
5	Migration Testing	Dev environment: storage/DB/middleware/app migration test + cutover dry-run. Validate tool compatibility and data integrity.	Test report approved	Migration Plan (P0) ð View Online
6	Migration Execution	Production: execute DB/middleware/app migration using DTS, MSP, go2tencentcloud, etc. Incremental sync until data gap = 0.	Migration complete; data synced	Migration Plan (P0)
7	Performance Testing	Functional and performance tests by app team. Optional: Goreplay traffic replay at 2x QPS.	Metrics meet baseline	No specific template — use POC benchmark tool

Phase 3 Deliverables: Cloud Resource Preparation Checklist + Migration Test Report + Performance Test Report

Step	Activity	Description	Completion Criteria	Templates
8	System Cutover	Redirect traffic to Tencent Cloud. Execute cold (stop-service) or hot (gray release 1%→100%) cutover per plan. 2-3 dry-run rehearsals mandatory.	Migration Acceptance Report	Cutover Plan (P4, 4 sheets) ð View Online Cutover Manual (merged into P4) ð View Online
9	Stabilization	5 days stable operation post-cutover. Monitor system availability, business metrics, HA. TSA inspection for optimization.	No cloud-caused incidents (15 calendar days)	TSA inspection — no template
10	Service Acceptance	Tencent submits acceptance report; client reviews within 5 business days and signs.	Signed acceptance report	Acceptance report — generated from project records

Phase 4 Deliverables: System Cutover Report + Stable Operation Observation Report + System Migration Acceptance Report

📎 Project-Wide Reference Template

The Migration Plan (P0) is the project-wide template covering the full migration lifecycle — 9 sheets including migration plan (Gantt), investment outline, 10-step process definition, RACI matrix, and data warehouse migration stages. Use alongside phase-specific templates: P1 information collection, P4 cutover plan.

Download Migration Plan (P0, 9 sheets) ð View Online

4 Milestones

Project Kickoff

End of Step 1 · Scope aligned

Plan & Solutions

End of Step 3 · Plans confirmed

Migration Report

End of Step 8 · Cutover done

Delivery & Acceptance

End of Step 10 · Signed off

Acceptance Criteria

After cutover (Step 8), the system must run stably for 5 consecutive days (15 calendar days) without any cloud-caused incidents rated at the client's defined severity level. Tencent submits the acceptance report; the client reviews and signs within 5 business days.

Migration Principles & Strategy

Four golden rules + design principles

Four Golden Rules

1:1 Parallel Migration

Keep environment and component versions identical to avoid compatibility issues from environmental differences.

Data Consistency

Ensure data consistency before and after migration. Use DTS full + incremental sync with verification.

Backward Compatibility

If no exact target version exists, follow backward compatibility or evaluate self-hosting.

Full Testing

Same component may have multiple migration options — test all with real business scenarios.

Design Principles

💰 Cost Optimization

Prefer high cost-performance instances (SA5 AMD)
Use Direct Connect to reduce traffic costs
Right-size via load testing
TSA continuous cost scanning

👷 Reduce Manual Effort

Professional PM for overall planning
Automated tools for batch migration
Terraform for IaC provisioning
CMG platform for end-to-end management

📉 Minimal Business Impact

Scientific migration + rollback plan
Pre-cutover testing & dry-runs
Edge-to-core, test-to-production
Component-level unit migration

Migration Strategy: Big-bang vs Phased

Dimension	Big-bang (All-at-once)	Phased (Incremental)
Approach	Single cutover window	Incremental batch cutover
Best for	Large-scale, complex dependencies, tight deadline	Small-scale, clear dependencies
Pros	No decoupling needed; shorter overall cutover; faster savings	Lower risk; minimal downtime; better testing
Cons	High overall risk & pressure; full planning needed	Longer implementation; complex dependency management

Cold vs Hot Migration

Choosing the right cutover model

Dimension	Cold Migration (Stop-Service)	Hot Migration (Smooth)
Downtime	>1 hour	<1 hour
Customer perception	Aware (maintenance notice)	No perception
Technical difficulty	Low	High
Cost	Low	High
Risk	Low	High
Best for	Traffic valleys; simple architecture; short downtime acceptable	24/7 services; high downtime cost; mature microservices
Gray release	None	1% → 100% traffic ramping; DB read-first then write

Downtime Formula

T_total = T1 + T2 + T3 + T4

T1 = Final data consistency check time → Minimize via daily incremental tracking, early DTS start, pre-built validation scripts
T2 = Business switch time → Minimize via clear dependency mapping, thorough dry-runs
T3 = Business test & validation time → Minimize via automated test scripts
T4 = Unexpected issue handling time → Reserve buffer

Cold Cutover Timeline (Example)

Step	Action	Duration
1	Close user access (security group)	5 min
2	Verify message queue consumption complete	5 min
3	Stop source DB writes	5 min
4	Data sync consistency check	10 min
5	Enable Tencent Cloud monitoring & alerts	5 min
6	Publish Tencent Cloud DB/Redis/MQ config, enable writes	10 min
7	Enable reverse proxy to Tencent Cloud	20 min
8	Update DNS resolution (non-blocking)	30 min
9	Validate business via proxy + Tencent domains	60 min
10	Open partial access	5 min
11	Observe business status	10 min
12	Open full access	5 min
13	Monitor traffic	30 min
Total		~5-7 hours

Hot cutover follows a 6-step flow: Deploy environment → Data sync (DTS) → Traffic gray (5%→20%→50%→100%) → DB cutover (read-first, then write) → DNS switch → Observe.

Landing Zone Design

8 dimensions for a solid foundation

Before moving a single workload, set up your Landing Zone — the foundational account, network, and security baseline. This is the foundation of your house.

1. Resource Planning

Account structure, resource IDs, quota management

2. Identity & Access

CAM, SSO, RBAC+ABAC, MFA

3. Cost Management

Unified payment, cost visualization, tag-based billing

4. Compliance Audit

Preventive / Detective / Post-audit

5. Network Planning

VPC, CCN, public ingress, firewall

6. Security Design

Resource / access / data security

7. Monitoring & Ops

Resource / app / business monitoring

8. Automation

Deployment / config / governance automation

Multi-Account Architecture (TCO)

Root Account
├── Core Admin Account (by function team)
│   ├── Log Administrator Account
│   ├── Resource Management Account
│   ├── Financial Management Account
│   ├── Security Management Account
│   └── Shared Service Account
├── Business Accounts (by business team)
│   ├── Game Business Account
│   │   ├── Pro Account (production)
│   │   ├── Dev Account
│   │   └── Test Account
│   ├── Social Business Account
│   └── Sensitive Business Account

4 Design Principles: Plan by Responsibility · Plan by Business (tag-based) · Plan by Environment (Pro/Dev/Test) · Plan by Special Requirements

Account Security Principles

Minimize Exposed Surface

Use SSO instead of Root login · Enforce MFA · Sub-users use temporary credentials (STS Token)

Shorten Exposure Time

Applications use STS Token, not persistent AK · On-demand AccessKey usage · Clean up idle credentials

Cost Tag Management (4 Steps)

Create tags and bind them to cloud resources
Set as cost allocation tags in the billing console
Each tag appears as a column in the bill — tag values show per resource row
Export bills to COS → analyze via big data platform

📖 Deep Dive: Landing Zone & CAM

Explore the full Landing Zone target design — including multi-account OU architecture, 4 baseline permission sets, ABAC authorization, 11 mandatory tags, 6 compliance libraries (190+ rules), 12-layer security product matrix, Terraform CI/CD pipeline, and governance metrics.

→ View Landing Zone & CAM Detailed Guide

CAM & Identity Management

Cloud Access Management best practices

Identity Types

Type	Use Case	Auth Method
Personnel Identity	Employee console access	SSO / CAM sub-user / MFA
Program Identity	App API access	STS Token / AccessKey

CAM Policy Structure (5 Fields)

Effect:    Allow / Deny
Service:   e.g., TencentDB for MySQL
Action:    e.g., Create / Delete / Describe
Resource:  * (all) or specific ARN
Condition: e.g., Source IP, Time range

Three Authorization Modes

Mode	Description
Operations Auth	RBAC or ABAC, least privilege
Project Auth	User Group batch authorization
Target Auth	Tag-based authorization per team/role

SSO (Single Sign-On)

Protocols: SAML 2.0 and OIDC
Two modes: User SSO and Role SSO
IdP integration: Microsoft AD, Azure AD, GCP IdP
CIC (Cloud Identity Center): Connect external IdP or manage users/groups directly

MFA (5 Types)

STS Token Flow

Avoid Hardcoded AK/SK

Create and authorize a CAM Role for CVM
Attach CAM role to CVM instance
CVM gets STS Token via Metadata
App uses STS Token to call cloud APIs

CloudAudit

Create CloudAudit trail → continuous operation log audit
Enable Config for credential compliance monitoring
Multi-account log tracking → COS + CLS (180-day retention)
Centralized analysis via CLS analytics & alerting

Network Planning

VPC, Direct Connect, CCN, security

VPC Planning Rules

Requirement	Recommendation
Single region	1 VPC
Multi-region DR	≥1 VPC per region
Cross-AZ DR	≥2 subnets in different AZs
CIDR ranges	10.x / 172.16-31.x / 192.168.x (mask 12-28)
Max subnets per VPC	100 (default)

Subnet Capacity

Mask	Usable IPs
/24	253
/22	1,021
/20	4,093
/16	65,534
/12	1,048,574

Cloud Connect Network (CCN)

20+

Regions

100Gbps

Max Bandwidth

99.99%

Availability

5Gbps

Free Tier

Direct Connect Speed (100TB migration)

Bandwidth	Time
1 Gbps	~9.5 days
2 Gbps	~4.7 days
5 Gbps	~1.9 days
10 Gbps	~1 day

VPN Comparison

Dimension	SSL VPN	IPSec VPN
Cost	Lower	Higher (complex client)
Security	Higher (port 443 only)	Multiple ports needed
Granularity	App-level	Network-level

Security Product Matrix

Category	Product	Function
Network Security	Dayu BGP Anti-DDoS	DDoS/CC attack protection
Host Security	YunJing (CWP)	Asset mgmt, vulnerability mgmt, intrusion detection
Web Security	WAF	SQL injection, tamper protection
Data Security	ShuDun	Data audit, data masking
Security Ops	Bastion Host	Privilege mgmt, ops audit
Threat Intelligence	AnTu	Threat intelligence analysis
Traffic Security	YuJie	Traffic protocol analysis, sandbox

📖 Deep Dive: Enterprise Network Architecture

Explore the full network Landing Zone design — including hub-and-spoke topology, CCN route table design patterns, 4 traffic flow paths (inbound/outbound/inter-VPC/hybrid), 3-layer defense in depth, GWLB transparent protection, network observability, and multi-account partitioning strategy.

→ View Network Planning Detailed Guide

Migration Tool Catalog

40+ tools organized by phase

Tencent Cloud provides purpose-built tools — always prefer tools over manual moves to reduce human error.

1. Scan & Survey

2. Deploy

3. Test

4. Migrate & Sync

5. Cutover

6. Optimize

Resource Scanning & Information Collection

Tool	Type	Source	Use
`mspcli` (Online)	Tencent Cloud	Aliyun, AWS	Auto-scan specs, quantity, relationships
`mspcli` (Offline)	Tencent Cloud	Aliyun, AWS	Local CLI resource inventory
Product Spec Comparison	Tencent Cloud	IDC/competitor	Spec mapping reference
CMG Platform	Tencent Cloud	Aliyun, Huawei, IDC	Full lifecycle: discovery → mapping → provisioning → migration → dashboard

Download: https://tools-release-1256125716.cos.ap-shanghai.myqcloud.com/package/urp/mspcli.tar.gz

Alternative: Cloud Billing Export (AWS / GCP)

When API-based scanning (mspcli) is limited by permissions or when historical cost analysis is needed, export billing data from the source cloud console. The exported CSV can be used to discover resource usage, costs, and distribution across accounts/regions/projects.

☁️ AWS: Cost & Usage Report (CUR 2.0)

Console: Billing & Cost Management → Data Exports → Create
Config: CUR 2.0 · Check Include resource IDs · Monthly
Format: gzip-text/csv → S3 bucket
History: Current month only. Prior months → contact AWS TAM or submit Support Center case (export name, month, S3 bucket, CSV format). Allow ~24h.
Key columns (15): product_servicecode, line_item_usage_account_id, line_item_unblended_cost, product_location, product_operation + 10 more

📄 AWS Export Guide (full steps)

☁️ GCP: Billing Report (SKU-level)

Console: Search "Billing" → Cloud Billing → Report
Group by: SKU · Time: last month or custom (supports past 13 months)
Filters: By Service, Region, Project, or Cost Type
Download: "Download CSV" button · Max 10,000 rows per export
CSV fields (11): SKU ID, Service Name, Region, Project ID, Usage Amount, Cost (Discounted), Discount Amount + 4 more
Prerequisite: Billing Viewer or Billing Account Viewer IAM role · For resource-level detail use BigQuery Export

📄 GCP Export Guide (full steps)

Comparison	AWS CUR 2.0	GCP Billing Report
Data scope	Full CUR 2.0 (114 columns)	SKU-level aggregated (11 fields)
History	Current month (history needs AWS support)	Past 13 months
Resource-level detail	Yes (check "Include resource IDs")	Use BigQuery Export
Export limit	First export within 24h	10,000 rows per CSV

📎 Complete Tools List (41 tools)

The Migration Tools (P0) template is a comprehensive Excel listing all 41 migration tools organized by phase — including scan, deploy, test, migrate, cutover, and optimize tools. Each tool entry includes its type, applicable source, and use case.

📥 Download Migration Tools (P0) 👁 View Online

Resource Deployment

Tool	Type	Use
Terraform	IaC	Automated resource provisioning — no manual creation
Domain ICP Filing	Tencent Cloud	ICP filing for domestic sites

POC & Performance Testing

Tool	Scope	Access
POC Benchmark Tool	CVM, Redis, MySQL, CBS	Web UI at `http://ip:8081`
Goreplay	Traffic replay & load test	Record Nginx traffic, replay at 2x QPS

Migration & Sync Tools (Core)

VM / CVM Migration

Tool	Mode	Key Notes
`go2tencentcloud`	Online	Source stays running; copies system params; OS: CentOS/Ubuntu/Debian/Windows
Offline Image Import	Offline	Formats: qcow2/vhd/vmdk/raw; MBR only (no GPT); no EFI; needs Virtio + cloud-init
Windows Migration	Special	2003/2008/2012 → upgrade to 2016 first; use disk2vhd for legacy

Database Migration — DTS

Supported DBs	Migration Types	Access Methods
MySQL, MariaDB, Percona, TDSQL-C, TDSQL MySQL, PostgreSQL, Redis, MongoDB, SQL Server	Structure / Full / Full+Incremental	Public network / CVM self-hosted / Direct Connect / VPN / Cloud DB / CCN / VPC

DTS Constraints

• Full migration has a brief lock (seconds-level write block) → run during low-peak
• Only supports basic tables & views — no functions/triggers/stored procedures (client must add)
• Storage engines: InnoDB, MyISAM, TokuDB only
• Incremental sync: no distributed transactions, no STATEMENT binlog, no binlog clearing, no DDL
• Do not modify/delete source or target DB user info

Object Storage Migration (COS)

Tool	Source → Target	Key Features
MSP Platform	OSS/OBS/S3 → COS	Full-managed (public) / Semi-managed (Direct Connect)
COS Migration	Local → COS	Resumable upload, chunked, parallel, MD5 verify
COSCMD / COSCLI	Local → COS	CLI batch upload/download/delete
COSBrowser	Local → COS	GUI desktop/mobile tool
COSDistCp	HDFS → COS	MapReduce distributed copy
HDFS TO COS	HDFS → COS	Direct file copy
COSFS	Local mount → COS	Mount bucket locally then copy

File Storage Migration

Tool	Use Case
Filetruck	CFS / local FS / COSFS sync; controllable start/stop; MD5 verify
rclone	70+ cloud storage backends; NAS → CFS via Linux mount
rsync	NAS → CFS direct sync
CDM	TB-PB scale offline via dedicated appliance; use when >50TB or >1 week

Container Migration

Tool	Use Case	Key Notes
Velero	K8s cluster backup & restore (ACK→TKE)	v1.5+; enable `--use-restic`; shared COS bucket; K8s API compatibility
image-transfer	Docker image batch migration	CCR/ACR/Harbor/Docker Hub/Quay → TCR; CCR→TCR one-click

Big Data Migration

Tool	Component	Mode
DistCp	Hadoop/HDFS	Full + update parameter for incremental
HBase Snapshot	HBase	Full sync
HBase Replication	HBase	Incremental sync (master-slave)
Oceanus	MySQL → ES	Real-time sync (Flink-based)
mysqldump	Hive metadata	Export → import

Elasticsearch Migration

Tool	Best For
ES Migration Tool	Full process: snapshot + COS
ES Snapshot (COS)	Large data (GB/TB/PB)
elasticsearch-dump	Small data offline (settings → mapping → data)
Logstash	Inter-cluster sync; version gaps

Redis Migration

Tool	Use Case	Note
DTS	Full + incremental	Needs psync permission; Aliyun 2.8 doesn't support
redis-port	Offline RDB export/import	Cross-version supported
redis-shake	Aliyun ElasticCache	AWS psync blocked → backup to S3 then import

Cutover Verification

Tool	Scope	Functions
DB Cutover Verification	MySQL/Redis/ES	Consistency check, read/write settings, connection reclaim, sampling
Big Data Cutover Verification	HDFS/HIVE	ORC/Parquet consistency verification

Monitoring & Optimization

Tool	Dimensions	Use
TSA (Tencent Smart Advisor)	Security / Reliability / Cost / Performance / Service Limits	One-click inspection, risk rating, architecture optimization

Key Tools Deep-Dive

DTS, go2tencentcloud, COS, CMG

🔄 DTS — Data Transmission Service

Principle: Source Dump → Load → Binlog incremental sync & verify

Supported Source → Target

Source	Target
MySQL	MySQL, TDSQL-C, TDSQL MySQL, Percona
Aurora MySQL / PolarDB	TDSQL-C MySQL
PostgreSQL	PostgreSQL
MongoDB	MongoDB
Redis	Redis
SQL Server	SQL Server

Sync Topologies

1:1 unidirectional · Cascading · 1:N · N:1 · Bidirectional · Table-level

Performance Reference

1TB / 4 DBs / 1115 tables via Direct Connect: ~69 hours (full export 5min, import 2h22m, incremental 7h39m, verify 1.5 days). Factors: data size, table complexity, bandwidth (max 60MB/s), source/target specs.

💻 go2tencentcloud

Online VM migration — source machine stays running.

Supported OS

Linux: CentOS 5-8, Ubuntu 10-20, Debian 7-10, SUSE 11-15, openSUSE 42, Amazon Linux, RHEL 5-8, Oracle Linux 5-8

Windows: Server 2008-2022

Usage

# Environment check
sudo ./go2tencentcloud_x64 --check

# Execute migration
sudo ./go2tencentcloud_x64

Docs: https://cloud.tencent.com/document/product/213/65711

Stateful App Caution

Self-hosted caches, long-lived connections, and distributed clusters require special handling during migration.

☁️ COS Smooth Migration (Zero Downtime)

Step 1: Full Sync

Use COS Migration tool. Data modified during sync may cause inconsistency → run multiple times with mtime traversal to minimize gaps.

Step 2: Configure Origin

After switching business to COS, enable 回源 (origin pull) in COS console. When COS lacks a resource, it pulls from source and stores it.

Step 3: Traffic Switch

Switch business read/write to COS. Source remains as fallback. Incremental data continues syncing in background.

🏗️ CMG — Cloud Migration Platform

End-to-end visualization platform for the entire migration lifecycle.

Auto Discovery

Aliyun/Huawei/IDC full scan

Smart Mapping

Auto-recommend TC specs

Batch Provision

Simplify params, reduce manual

Dashboard

Real-time progress monitoring

Efficiency Gain

Overall cycle: 1-2 weeks → 1-2 days. Per 100 resources: 80% efficiency improvement.

AI-Powered Migration Acceleration

XLSmart — 300% speed improvement

Tencent Cloud's XLSmart AI migration toolkit accelerates enterprise migration through automated code analysis, dependency mapping, and infrastructure generation.

300%

Speed Improvement

3→1

Person-Months (IaC)

66%

Engineering Efficiency

100%

Auto IaC Generation

Three Pillars

1. Code & Dependency Inference

Deep application scanning — map complex upstream/downstream topology and hardcoded IPs. Codebuddy plugin parses raw app code, extracts hardcoded IPs and domains for instant risk inference.

2. Predictive Infrastructure Validation

Pre-deployment LLD validation + automated Yehe platform whitelist. AI dry-run prediction eliminates quota/whitelist failures before deployment.

3. Cognitive Core (MCP Server)

Real-time API knowledge access — eliminates AI hallucination, ensures cross-border compliance. Daily incremental updates, extends to international site docs.

AI vs Legacy Workflow

Stage	Legacy Method	AI-Optimized	Impact
App Analysis	Manual code review	Codebuddy + auto topology	Eliminate mapping errors; auto-generate test cases
Resource Audit	Reactive failure	LLD validation + predictive whitelist	Eliminate days of blocking wait
Provisioning	Manual mapping (3 PM)	AI-driven Terraform (1 PM)	66% efficiency gain

Cloud Product Mappings

AWS / GCP / Aliyun → Tencent Cloud

AWS → TC

Multi-Cloud L0

Category	AWS	Tencent Cloud	Migration Method
Compute	EC2	CVM ↗	Rehost / Terraform rebuild / Image import / go2tencentcloud
	EKS	TKE ↗	Replace CA or custom karpenter; Velero backup/restore
	ECS	TKE ↗	CI/CD redeploy
	Lambda	SCF ↗	Re-develop
Storage	S3	COS ↗	MSP platform; replace SDK; S3-compatible API
	EBS	CBS ↗	—
	EFS / FSx	CFS ↗	Filetruck
Network	Transit Gateway	CCN ↗	Route cleanup, Terraform recreate
	ELB	CLB ↗	—
	CloudFront	EdgeOne (EO) ↗	—
	Route 53	DNSPod ↗	Re-configure
Database	Aurora (MySQL/PG)	TDSQL-C ↗	DTS full + incremental
	RDS MySQL	TencentDB for MySQL ↗	DTS
	RDS PostgreSQL	TencentDB for PostgreSQL ↗	DTS
	ElastiCache	TencentDB for Redis ↗	DTS (backup to S3 → redis-shake for AWS)
	DocumentDB	TencentDB for MongoDB ↗	DTS
	Redshift	Cloud Data Warehouse ↗	—
Big Data	EMR	EMR ↗	DistCp / cos-distcp
Big Data	Kinesis	Oceanus ↗	Adapt migration
Messaging	SQS	TDMQ for CMQ ↗	—
	MSK	TDMQ for CKafka ↗	Mirror-maker + reconfigure
	Amazon MQ	TDMQ for Pulsar/RabbitMQ ↗	Refactor, replace SDK
Security	IAM	CAM ↗ / STS ↗	Re-configure
	KMS	KMS ↗	Check CMK support
	Secrets Manager	SSM ↗	Re-configure
	CloudTrail	CloudAudit ↗	Re-configure
Management	CloudWatch	TCOP ↗	—
Management	Trusted Advisor	Cloud Advisor (TSA) ↗	—

↗ = Click to view Tencent Cloud product page

Category	AWS	Tencent Cloud	Aliyun	GCP	Azure
VM	EC2	CVM	ECS	Compute Engine	Azure VM
K8s	EKS	TKE	ACK	GKE	AKS
Container Registry	ECR	TCR	ACR	Artifact Registry	ACR
Serverless	Lambda	SCF	Function Compute	Cloud Functions	Azure Functions
Object Storage	S3	COS	OSS	Cloud Storage	Blob Storage
Block Storage	EBS	CBS	Block Storage	Persistent Disk	Page Blobs
File Storage	EFS	CFS	NAS	File Store	Azure Files
Direct Connect	Direct Connect	Direct Connect	Express Connect	Cloud Interconnect	ExpressRoute
Load Balancer	ELB	CLB	SLB	Cloud LB	Azure LB
CDN	CloudFront	EdgeOne	CDN	Cloud CDN	Azure CDN
DNS	Route 53	DNSPod	Cloud DNS	Cloud DNS	Azure DNS
MySQL	RDS MySQL	TencentDB for MySQL	RDS MySQL	Cloud SQL	Azure DB MySQL
Redis	ElastiCache	TencentDB for Redis	ApsaraDB Redis	Memorystore	Azure Cache Redis
Data Warehouse	Redshift	CDW	MaxCompute	BigQuery	SQL DW
Kafka	MSK	TDMQ for CKafka	MQ Kafka	—	—
IAM	IAM	CAM	RAM	Cloud IAM	Azure AD
Monitoring	CloudWatch	TCOP	CloudMonitor	StackDriver	Azure Monitor
DMS	AWS DMS	DTS	DTS	DB Migration Service	Azure DMS
Big Data Cluster	EMR	EMR	E-MapReduce	Dataproc	HDInsight

Cutover & Rollback

Execution manual and decision criteria

Full Cutover Manual (5 Phases, 27 Steps)

Phase 1: Preconditions

Issue cutover notice (internal + external) — 30 min
Stop daily change releases — 5 min
Security policy check (security groups / whitelists)
DNS domain resolution config check
Verify Tencent Cloud COS write test — 30 min
Incremental data check — DB sync latency — 2 min

Phase 2: Prepare — Close Source Traffic

Close source cloud traffic ingress (LB / public IP) — 5 min
Stop all source cloud tasks (Web layer) — 20 min
Stop source cloud cluster services — 10 min
Stop all scheduled tasks — 10 min
Stop source cloud other services/interfaces — 5 min
Disable source cloud business alerts

Phase 3: Cutover — Data Switch

Set source MySQL to read-only — 2 min
Kill source MySQL tail connections — 3 min
Verify source vs Tencent CDB data consistency — 10 min
Stop DTS sync task — 2 min
Set Tencent Cloud CDB to writable — 2 min
Update Tencent Cloud app DB connection strings — 5 min
Configure COS origin pull — 2 min
Adjust Tencent Cloud Kafka — 5 min
Check Tencent Cloud service component status — 30 min
Quick mock test key business — 30 min
Close maintenance page — 5 min
Configure DNS to Tencent Cloud — 5 min
QA regression test main business path — 30 min

Phase 4: DNS Switch

DNS resolution to new IP
Clear DNS cache
Verify all domain resolutions point to new site
Restart long-connection services in batches
Update office network DNS to new IP

Phase 5: Post-Cutover & Cleanup

Observe service running status
Monitor data metrics
Archive remaining issues
Keep source environment for 2 weeks as rollback support
Clean up source environment

⚠️ Mandatory: 2-3 Dry-Run Rehearsals

Before the real cutover, conduct at least 2-3 full rehearsals to identify 99% of issues and familiarize the team with the process.

Rollback Strategy

Rollback Steps

Set source DB to writable (read_only=OFF)
Restart source Web / scheduled tasks / other services
Verify source service component status
Configure DNS resolution back to source
Close maintenance page
Test main business path
Re-enable source business alerts

Rollback Decision Criteria

Scenario	Action
Data sync interrupted, cannot complete incremental in window	Abort cutover, find root cause
Data inconsistency detected	Abort, investigate sync method & write-stop
Post-cutover instability, short-term recovery unlikely	Re-evaluate migration plan

Decision Standard

If issues unresolved by business peak period, execute rollback. Define the decision standard early to reduce major incident decision time.

SOW & RACI

Service packages, responsibility matrix, deliverables

SOW 6 Service Packages

Service Item	Business Consulting	Business Guidance	Business Implementation	BigData Consulting	BigData Guidance	BigData Implementation
Survey & Assessment	✔			✔
Migration Preparation	✔			✔
Business Migration Design	✔
Big Data Migration Design				✔
Big Data Migration Implementation						✔
Big Data Testing Support					✔	✔
Business Migration Implementation Guidance		✔
Business Migration Implementation			✔
Business Testing Support		✔	✔
Cutover Support		✔	✔		✔	✔
Service Acceptance		✔	✔		✔	✔

R/S Responsibility Matrix

Aligned with Section 04 (4 Phases Workflow)

The phases below use the exact same naming as the 4 Phases Migration Workflow in Section 04. Each task maps to a specific Step (1–10) for traceability.

R = Responsible (leads) S = Support (assists)

Phase	Step	Task	Client	Tencent
P1: Discover & Assess	1	Migration scope definition; team assembly	R	S
P1: Discover & Assess	2	Resource scanning; performance metrics collection	R	S
P2: Design	3	Business architecture; network topology; account/CAM/billing; migration WBS; PoC testing	S	R
P3: Implement	4	Cloud foundation deployment; environment provisioning (Terraform)	R	S
	5	Migration testing (dev env): storage/DB/middleware/app + cutover dry-run	R	S
	6	Migration execution (prod): data sync automation & tools	R	S
	7	Performance testing; gray release runbook & rollback preparation	R	S
P4: Validate & Handover	8	System cutover: Tencent护航 support team; gray release execution / DB switch	R	S
	9	Stabilization: app stability observation; DR drill	R	S
	10	Service acceptance: migration review & summary	S	R

Summary: Phase 1–2: Tencent leads (R) · Phase 3–4: Client leads (R) · Phase 4 Step 10: Tencent leads review (R)

11 Deliverables

Project team confirmation
System Information Research Report
Cloud Migration Assessment Report
Cloud Migration Technical Plan
Cloud Migration Plan (with WBS)
Cloud resource preparation checklist

Migration test report
Performance test report
System cutover report
Stable operation observation report
System migration acceptance report

SOW Exclusions

Not Included in SOW

Data center and physical network design, hardware installation/config/testing
Code optimization (client responsibility)
Third-party software/app system implementation (consulting only)
Cloud resource procurement (client applies)
Customer proprietary app / open-source / non-TC commercial software installation
Stress testing or security scanning (unless agreed — requires additional estimation)
Migration to non-Tencent public cloud

SRE & Disaster Recovery

ISO 22301 BCMS, chaos engineering, DR

Tencent Cloud follows ISO 22301 BCMS (Business Continuity Management System) for cloud product continuity governance.

4 Factors Affecting Continuity

1. Change Quality Control

Strict QA management from change lifecycle start — 14-step pipeline from build to full release

2. Cross-AZ DR Baseline

Cross-AZ as deployment standard; single-AZ products need control-plane cross-AZ + data-plane fast recovery

3. Risk Exposure Governance

BCP use cases systematically identify risks → fix plans → chaos engineering validates

4. Chaos Engineering

Full-stack drills IaaS→PaaS→business; third-party blue team attack/defense

Chaos Engineering Platform

100+

Fault Atomic Actions

10+

Resource Types

5,380+

Practice Drills (2023)

4-step

Drill Process

ITSM Andon Platform

Priority	Definition	OLA Escalation
P1	Service outage (data loss / critical business)	10 min → Leader → Director
P2	Major business impact (slow / partial freeze)	20 min escalation
P3	Business impact	20 min escalation
P4	Standard inquiry	Standard

DR Strategy Design Process

1. Develop DR Strategies

DR tier, RTO, RPO, risk assessment, BIA, drill planning

2. AS-IS Investigation

Current state: business / tech / system architecture / resources

3. DR Priority Ranking

Core systems first → non-core

4. Implementation

New resources, risk mitigation, single-point transformation, data sync, testing, desktop → simulation → live drill

HA Case: Retail Customer

Result: AZ failure recovery 4 hours → 15 minutes

Strategy: Same-city dual-active + remote hot standby. Access layer: WAF/API Gateway/CLB cross-AZ. App layer: multi-AZ cluster. Data layer: DTS sync (MySQL/TDSQL), global replication (Redis), CCR (ES), bucket replication (COS).

RTO: Same-city 15 min · Cross-city 30 min

Cost Optimization

10+ Tencent self-developed cost reduction technologies

Tencent Self-Developed Cloud Migration Results

1 billion cores on cloud · 5 billion CNY cumulative savings · 45% overall CPU utilization (industry-leading)

Technology	Description	Result
SA5 AMD Bergamo	512 vCPU per server (global exclusive)	20%+ per-gen performance gain; 90%+ customers prefer AMD
T-block Self-built Campus	PUE ≤ 1.2, 300K+ servers per campus	50% shorter construction cycle
TencentOS "Ruyi" Co-location	Online + offline mixed deployment (BT scheduling)	CPU utilization 50%+ (cluster 55%)
TencentOS "Wujing" Memory Offload	Multi-level memory unloading	35% memory savings
Zixiao Inference Chip	CV/NLP acceleration	20%+ performance improvement
qGPU	Fine-grained GPU slicing + co-location	30% throughput gain, 15% cost reduction
Xingmai Network	1000-card training network	20% performance gain, 30% cost optimization
TACO-LLM	LLM inference acceleration	2.47x vs vLLM; 3.71x vs TensorRT-LLM
GooseFS + COS	Hot/cold data tiering	30%+ storage cost reduction; 25% training efficiency gain
Compute-Storage Separation	Decoupled compute and storage layers	Independent scaling, cost optimization

Case: Online Education Customer

AMD CPU + online/offline co-location → 43% cost reduction, 60% utilization

Case Studies

200+ migrations across industries

Retail

Internet

Fintech

Web3

Source	AWS (Singapore) + Azure (Big Data) + IDC
Scale	100+ apps in 70+ AWS accounts; 5000+ cores; MongoDB 120+, Redis 47+, PostgreSQL 63+; 400+ microservices
Duration	~1 year
Key Tech	Terraform full automation; KONG/Kafka/ES to replace AWS proprietary PaaS; Palo Alto unified firewall; Landing Zone multi-account; ArgoCD + Terraform CI/CD
Strategy	Phased migration — Lotus's 20+ stages, Makro 10+ stages
PoC	2-week assessment, 1-month PoC, 3-week tech PK vs competitor
Result	IaaS cost reduced 72%; long-term operational cost ~50% reduction

GoTo Group (Indonesia)

Source	GCP (3 regions)
Scale	1000+ microservices, 10000+ VMs, 730+ DBs, 12 Kafka clusters
Duration	10 months
Cutover	Zero-downtime, 4h54m; 6 rounds of rehearsals; 300+ on-site + 100+ remote
Result	Cost savings achieved; UX latency improved

ClientX (Cross-Region, High Coupling)

Source	Aliyun (Hangzhou → Nanjing)
Scale	270+ K8s services, 5000+ VM cores, MySQL 200+ (48T), Redis 200+ (1.5T)
Duration	3 months
Strategy	All-at-once (98% read / 2% write)
Cutover	35 min; Goreplay 2x QPS traffic replay; 4 dry-runs

Motion Social Platform

Scale	3亿 users, 6M DAU; 350+ hosts, 100+ DBs, 600+ big data hosts, 6PB+
Duration	3 months
Cutover	One-time, 6 hours
Key	Kernel + iptables dependency mapping; bandwidth estimation; big data first

APEX Exchange (Singapore Fintech)

Industry	Singapore Tier-1 financial exchange
Architecture	Hybrid cloud: Tencent public cloud + leased racks + APEX SG3 + Office
3-Phase Evolution	Phase 1: Clearing/monitoring to cloud → Phase 2: Trading to cloud, retire IDC → Phase 3: Full cloud-native (containers, DevOps, data lake)
Compliance	MAS Outsourcing Guidelines, TRM Guidelines, Cyber Hygiene, BCM Guidelines, Cloud Advisory
Certifications	ISO 20000/22301/9001/27001/27017/27018/27701/29151, SOC 1/2/3, OSPAR, MTCS
DB Migration	Oracle 11g via Dataguard; MySQL/PG via DTS
DR	Trading: Primary SG3 / Backup Cloud AZ; Clearing: 3-AZ deployment

Footprint (Web3 Data Analytics)

Source	GCP (BigQuery / GCS / GCE / Cloud SQL)
Target	COS (S3-compatible), CVM, TencentDB, Data Lake Compute, EMR (Iceberg + Trino + Kafka + Doris)
Duration	~2 months (PoC → data migration → validation → DNS switch → optimization)
Migration	GCS→COS (7-8 G/s bandwidth); GCE→CVM (online/image); Cloud SQL→TencentDB (DTS)
Result	CPU performance better than GCP; memory/disk comparable

AWS Migration Case Gallery

Customer	Industry	Migration
Haier	Appliance	To Tencent overseas cloud (Moscow)
Lumi	Smart Home	AWS SG → TC; 173 countries, 4.7M+ users
Sanofi	Pharma	Full AWS → TC; Transit Gateway → CCN; Direct Connect hybrid
Education First	Education	AWS/IDC → TC global; CKafka decoupling; EMR for BI
Supercell	Gaming	Multi-cloud (US=AWS, others=TC); Auto Scaling + multi-AZ
Riot Games	Gaming	Multi-cloud; 10M+ DAU; CCN + TKE + COS + CKafka
StarMaker	Social	AWS US → TC Mumbai (SEA/Middle East expansion)
ChuChuJie	E-commerce	AWS → TC full migration; 6-step phased

Strategic Customers

DiDi

Elastic cloud scaling: daily 10-20K cores, peak 90K cores (May Day). Key: concurrent lock queue, cloud disk burst performance, IP allocation efficiency (23K/min), ENI binding (2K/min).

Meituan

Top 5 revenue: compute, network, SMS, CDN, PSTN. 2023 focus: LLM GPU, overseas delivery, live streaming, DR drills. Future: FinOps/Crane, big data cloud, co-location cost reduction.

Honor (Device)

Source	Huawei Cloud
Scale	2000+ ECS / 3W+ cores; MySQL 200+, Redis 200+, MongoDB 30+, Oracle 30+; 10PB+ OBS
Duration	~10 months
Strategy	Cold migration in 4 batches + chaos engineering
Cutover	2.5 hours per batch; 3+ rehearsal rounds
DB Performance	TDSQL-C outperformed GaussDB in benchmark (8 clients: QPS 11835 vs 11694, CPU 87% vs 92%)

Risks & Mitigations

Plan for failure before it happens

Risk	Mitigation Strategy
IP Change Risk	1. Check business config for hardcoded IPs — prefer DNS domains 2. For third-party integration, apply for whitelist 20 days in advance 3. Spamhaus DBL risk prevention — prepare redundant domains/IPs
Business Interruption Risk	1. Choose low-traffic time window 2. Keep source online during migration — only interrupt at final incremental sync + switch 3. Build primary/backup environment (source + target) 4. Ensure every step is reversible
Network Stability Risk	1. Build dedicated migration Direct Connect 2. Use image import for virtualized hosts to reduce network dependency (manually supplement incremental) 3. Stagger DB and data warehouse migration to avoid bandwidth contention
Data Inconsistency Risk	1. Full sync + multiple incremental syncs — stop source writes at final sync 2. Verify via DTS or scripts; full table count check during cutover 3. Client backs up source data before migration
Personnel Stability Risk	1. Prepare backup personnel for both client and Tencent interfaces 2. Regular weekly meetings; client should designate a PM role
Complex Dependency Risk	1. Map all dependencies during assessment phase 2. Identify all stakeholders and third parties

Common Mistakes to Avoid

Preparation

❌ No resource inventory or incomplete scan
❌ No migration schedule or single PM interface
❌ Wrong cutover approach selection
❌ Ignoring audit/compliance (e.g., domain ICP filing)

Testing & Cutover

❌ No compatibility/functional testing of key modules
❌ Fixating on source cloud features, ignoring TC alternatives
❌ Using old-cloud experience configs on new cloud
❌ No dry-run or half-hearted rehearsal
❌ Poor communication — no regular meetings

Real-World Pitfalls & Solutions

Problem	Scenario	Solution
Data consistency	Core DBs need strong consistency	DB cutover helper tool (MySQL quick check)
RDS read-only issue	RDS FTWRL not effective / HA switch	Security group / read-only account fallback; low-peak execution
Redis psync permission	Cloud provider blocks psync	Master-slave: create replication account; Cluster: apply for direct mode; 2.8 unsupported
OSS archive restore	Archive data not real-time accessible	Restore to standard storage before MSP migration
Bandwidth saturation	Two Direct Connect lines maxed out	Set bandwidth limit (DistCP `-bandwidth`)
Environment inconsistency	HBase load spike post-cutover, compaction backlog	Maintain config file consistency; follow parallel migration principle
Egress IP redundancy	Third-party callback IP C-segment blocked	Network planning: reserve backup IPs, add mutual whitelists

Next Steps

Start your migration journey

Engage your Tencent Cloud account team for a discovery workshop
Fill out the Information Collection form — workloads, dependencies, compliance, RTO/RPO
Run a scoped POC on one non-critical workload
Approve the Migration Technical Plan — review target architecture and migration strategy
Provision the Landing Zone via Terraform — accounts, network, security baseline
Execute migration in batches — edge to core, test to production
Cutover, validate, then continuously optimize with TSA (Tencent Smart Advisor)

Ready to Begin?

Migration is a journey, not a switch. With the right methodology, tooling, and team enablement, you will get there. Contact your Tencent Cloud representative to kick off your discovery workshop today.

Tencent Cloud Migration Guide · Based on 200+ real-world migrations

Built from internal migration methodology, tool best practices, and case study documentation