This article introduces the resource model design of CloudStack: VO (Persistent Object) and DAO (Data Access Object).

1. CloudStack Data Layer Structure

The data layer is divided into:

VO (Value Object → DB Row)
DAO (Data Access → SQL Operation)
Manager (Business Logic)
Orchestration Engine
Agent (Execution)

VO and DAO patterns are extremely critical in CloudStack.

2. VO（Value Object）

A VO is a direct mapping to a database table. For example:

VO Corresponding Table
VirtualMachineVO vm_instance
VolumeVO volumes
NetworkVO networks
HostVO hosts

2.1 Characteristics of a VO (Value Object):

  • Fields correspond one-to-one with the database (DB).
  • Includes resource state fields.
  • Typically immutable objects.
  • Used solely for data storage; does not contain business logic.

    3. DAO（Data Access Object）

DAO’s responsibilities include:

  • CRUD operations
  • Primary key/foreign key based queries
  • State-based conditional queries
  • Optimistic locking operations
  • State machine transitions

Example:

VirtualMachineDao
VolumeDao
NetworkDao
IpAddressDao

3.1 Optimistic Locking

DAO Uses：

UPDATE vm_instance SET state = ? WHERE id = ? AND state = oldState

ensures concurrency safety.

4. State Machine Combined with DAO

VM state updates typically use:

_stateMachine.transitTo(vm, newState, event);

Internally, it will：

  1. Validate valid migrations
  2. Call the DAO to update the database
  3. Publish the event (EventBus)

5. Resource Object Model

CloudStack resource model process:

VO (Persistent Data Layer)
  → DAO (SQL Access)
  → Manager (Business Logic)
  → Orchestration (Cross-Module Orchestration)
  → AgentCommand (Physical Execution)

For Example Start a VM：

vmDao.findById()
vmDao.update()
networkMgr.prepare()
volumeMgr.prepare()
agentMgr.send(StartCommand)