-
Notifications
You must be signed in to change notification settings - Fork 16
Core Services
ravikiranpagidi edited this page May 27, 2026
·
1 revision
This page maps major Alibaba Cloud services by capability and helps you choose the right service for common engineering scenarios.
| Capability | Alibaba Cloud Services | What They Are Used For |
|---|---|---|
| Compute | ECS, Auto Scaling, SLB, Elastic Container Instance, ACK, Function Compute | Run applications, containers, APIs, jobs, and event-driven workloads. |
| Storage | OSS, Cloud Disk, File Storage NAS, Table Store, Hybrid Backup Recovery | Store objects, block volumes, shared files, NoSQL-style data, and backups. |
| Networking | VPC, vSwitch, route tables, NAT Gateway, EIP, SLB, VPN Gateway, Express Connect, PrivateLink, CDN, Global Accelerator | Isolate networks, route traffic, expose services, connect hybrid environments, and accelerate access. |
| Security | RAM, KMS, Cloud Firewall, Security Center, ActionTrail, Cloud Config | Control access, encrypt data, detect threats, audit activity, and enforce governance. |
| Databases | RDS, PolarDB, AnalyticDB, Tair, MongoDB-compatible service, Lindorm, OceanBase, DTS | Store transactional, analytical, cache, document, wide-column, and migrated data. |
| Analytics | OSS, MaxCompute, DataWorks, DLF, Realtime Compute for Apache Flink, Hologres, AnalyticDB, E-MapReduce, SLS | Build data lakes, warehouses, batch pipelines, streaming pipelines, and analytics platforms. |
| AI/ML | PAI, DSW, DLC, model deployment, feature workflows, vector/RAG design | Train, deploy, and govern ML and GenAI workloads. |
| DevOps | Alibaba Cloud CLI, Terraform provider, ROS, Container Registry, ACK, GitHub Actions | Automate infrastructure and application deployment. |
| Monitoring | CloudMonitor, SLS, Managed Service for Prometheus, Managed Service for Grafana, Tracing Analysis | Collect metrics, logs, traces, dashboards, and alerts. |
| Governance | Resource groups, tags, RAM policies, ActionTrail, Cloud Config, budgets | Organize resources, audit changes, enforce standards, and allocate cost. |
| Migration | DTS, database migration tools, server migration patterns, hybrid network services | Move apps, databases, and data into Alibaba Cloud. |
| Cost management | Billing console, tags, budgets, lifecycle policies, reserved capacity, savings plans | Track, allocate, optimize, and forecast spend. |
| Workload | Start with |
|---|---|
| Traditional VM workload | ECS |
| Stateless web app | ECS + SLB + Auto Scaling or ACK |
| Containerized platform | ACK |
| Event-driven API or task | Function Compute |
| Static website or object storage | OSS + CDN |
| Relational database | RDS or PolarDB |
| Large-scale analytics | OSS + MaxCompute + DataWorks |
| Real-time analytics | SLS/EventBridge/Message Queue + Flink + AnalyticDB/Hologres |
| ML platform | OSS + DataWorks + MaxCompute + PAI |
- Need OS-level control? Use ECS.
- Need Kubernetes ecosystem? Use ACK.
- Want less server management? Use Function Compute or managed services.
- Need managed relational database operations? Use RDS or PolarDB.
- Need object durability and data lake storage? Use OSS.
Before choosing a service, verify:
- Region and zone availability
- Quotas
- Pricing model
- Security controls
- Logging and audit support
- Backup and restore requirements
- Integration with existing network and identity
- Team skills and support model
flowchart LR
User[Users] --> CDN[CDN]
CDN --> SLB[SLB]
SLB --> App[ECS / ACK / Function Compute]
App --> RDS[(RDS / PolarDB)]
App --> OSS[(OSS)]
App --> SLS[SLS Logs]
App --> CloudMonitor[CloudMonitor Metrics]
RAM[RAM + KMS + ActionTrail] -.controls.-> App
- Start with business requirements, not service names.
- Prefer managed services when they reduce undifferentiated operations.
- Keep public endpoints intentional and documented.
- Use resource groups and tags for every project.
- Enable logging, metrics, and budget alerts early.
- Design cleanup before running labs or prototypes.
- Revisit service choices as workload scale and team maturity change.
- Choosing Kubernetes for every workload.
- Using ECS when Function Compute or a managed platform is enough.
- Using OSS like a database.
- Exposing databases publicly.
- Forgetting that service mappings across clouds are approximate.
- Ignoring quotas and region availability until deployment day.
"I choose Alibaba Cloud services by workload type and operational responsibility. For VM control I use ECS, for containers ACK, for event-driven code Function Compute, for object storage OSS, for relational data RDS or PolarDB, for analytics MaxCompute or AnalyticDB, and for governance RAM, ActionTrail, Cloud Config, tags, and budgets. I validate region availability, quotas, cost, security, monitoring, and backup requirements before finalizing the design."
- Pick three workloads and choose services for each.
- Write why you would not use Kubernetes for a simple static website.
- Map one existing AWS/Azure/GCP architecture to Alibaba Cloud services.