This chapter explains the technical terms used in this document. If you encounter unfamiliar concepts while reading, you can quickly find their definitions and detailed explanations here.
Replication service
The replication service is responsible for managing synchronous replication plans and scheduling jobs such as synchronous replication, failover, failover test, failback, and permanent failover for virtual machines. The replication service ensures data consistency of replication objects.
Site
A site refers to a fault domain managed by a single AOC. A site can include one or more clusters and data centers. Each site is completely isolated in terms of geographical location, hardware equipment, management platform, and software services, and does not share any resources with other sites.
Sites can be divided into the following two types:
Intra-site replication
The source cluster and target cluster are located at the same site. When a source virtual machine is unavailable, you can switch to a replica virtual machine in another cluster within the site, achieving cluster-level disaster recovery. If the AOC that manages the site fails, failover cannot be initiated.
Cross-site replication
The source cluster and target cluster are located at different sites. When a source virtual machine in the source site is unavailable, you can switch to the replica virtual machine at the target site to achieve site-level disaster recovery, regardless of whether the AOC that manages the source site has failed.
Cross-site replication includes the following two types:
Source cluster
Usually a production cluster that hosts production applications and data. It is the cluster to which the source virtual machine belongs and is located at the source site.
Target cluster
Usually a cluster used for disaster recovery. It stores replica objects replicated from the source cluster and is located at the target site.
Protected object
The original object used as the replication data source. For virtual machine replication, the protected object is usually a virtual machine running in the production environment and hosts production applications and data.
Replica object
A complete copy of a protected object created through a synchronous replication plan. It contains the same data and configuration as the protected object. Replica objects are typically used for disaster recovery. When a protected object fails, the replica object can quickly take over the protected object to ensure service continuity and availability. A replica object is usually stored in a target cluster used for disaster recovery.
Synchronous replication plan
A synchronous replication plan is used to plan and schedule data replication operations. Through a synchronous replication plan, you can define information such as replication objects, replication direction, target cluster, replication service, synchronization policy, replica virtual machine settings, and test virtual machine settings.
Replication object
A protected object and its corresponding replica object are called a set of replication objects. For virtual machine replication, a source virtual machine and its replica virtual machine are called a set of replication objects.
Restore point creation interval
The interval at which a synchronous replication job automatically creates restore points. It is used to specify the time interval for creating restore points after a replication object enters the Synchronized state.
Restore point
A restore point is the data state generated each time asynchronous replication, synchronous replication (periodically in the Synchronized state), failover in Generate now mode, failover test in Generate now mode, or failback is performed for a replication object. It is saved as a snapshot on the replication object. A restore point records the data and configuration of the replication object and reflects the state of the replication object when the preceding operation was performed.
Restore points allow replication objects to quickly roll back to a specific data state when needed. For example, in a disaster recovery scenario, you can restore a replica object to a previous point in time to resume normal operations and maintain service continuity and availability.
Fault handling
Recovery operations that can be performed after a fault occurs, including failover, failover test, failback, and permanent failover.
Failover
Failover is an operation used when a protected object is unavailable. When a protected object is unavailable, the replica object can be rolled back to a state consistent with the specified restore point and then switched over for use. Through failover, the system can quickly restore services to the state before the fault, ensuring service continuity and stability.
Failover test
Failover test is an operation that simulates a disaster recovery drill. Without interrupting production workloads on the source side, it generates test virtual machines and verifies the effectiveness of the recovery mechanism on them. Incremental data generated during the drill is not synchronized back to the protected virtual machines.
Failback
Failback means that after the protected object that experienced a fault resumes normal use, the replica object transfers the data changed during failover back, and then services are switched back to the protected object. Failback synchronizes data between replication objects, ensuring that the protected object can continue to provide services after the fault is resolved.
Permanent failover
Permanent failover is a measure taken when a protected object cannot recover from a fault. When a protected object has an unrecoverable failure, the replica object formally takes over the service role of the protected object to ensure continuous service provision and availability.
RPO (Recovery Point Objective)
Recovery point objective. It mainly refers to the acceptable amount of data loss, measured in time in a failover scenario. RPO is measured in time and indicates the time span from the most recent replication point to the time when the fault occurs. A smaller RPO value indicates that the system allows less data loss in the event of a fault. A larger RPO value means that the system can tolerate a longer time range of data loss during a fault.
RTO (Recovery Time Objective)
Recovery time objective. It mainly refers to the amount of time a service system can tolerate from crash to recovery. A shorter RTO value indicates that the service system requires rapid recovery when a fault occurs, while a longer RTO value means that the service system can tolerate a longer recovery time.
VPC (Virtual Private Cloud)
A virtual private cloud is a logically isolated network environment built for cloud resources. You can configure VPC subnets, IP addresses, route tables, floating IPs, gateway services, security services, and other resources in a VPC to achieve network isolation and security control. Replica virtual machines and source virtual machines can be deployed in the same or different VPCs, and flexible network mapping is supported. Test virtual machines must be deployed in a VPC different from the source virtual machine.
VPC subnet
A VPC subnet is a fundamental network module that makes up a VPC. It is used to segment the IP address space of the VPC. Virtual machines in the same VPC subnet can communicate directly by default. Virtual machines across VPC subnets can also communicate with each other if routing is reachable.
Floating IP
A floating IP is a network address translation (NAT) mechanism that uses one-to-one mapping to dynamically bind an IP address from an external subnet to a virtual machine inside a VPC. This allows external networks to actively access virtual machines in the VPC. Floating IPs provide flexible external access for virtual machines in a VPC while ensuring network security.
Network mapping
A configuration rule designed to address network environment differences between a source virtual machine and a replica virtual machine or test virtual machine.
Because replica virtual machines, test virtual machines, and source virtual machines are often in different network environments, completely copying the network configuration of a source virtual machine to a replica virtual machine or test virtual machine may prevent the network from being properly configured or connected in the target cluster environment. By configuring network mapping rules, the VM network to which the source virtual machine belongs can be mapped to the corresponding VM network in the target cluster, ensuring that the replica virtual machine or test virtual machine can connect to the network properly.
IP mapping
A configuration rule designed to address IP address mismatches between a source virtual machine and a replica virtual machine or test virtual machine in different network environments.
Because replica virtual machines, test virtual machines, and source virtual machines are often in different network environments, completely copying the IP address of a source virtual machine to a replica virtual machine or test virtual machine may prevent the network from being properly connected in the target cluster. By configuring IP mapping rules, during failover and failover test, the replication service can automatically set the correct IP address for the replica virtual machine or test virtual machine, ensuring that the replica virtual machine or test virtual machine can provide services according to the expected network configuration without manually updating the virtual machine IP address.
AVE
AVE refers to the ACOS native virtualized computing platform. It is the KVM-based virtualization engine component in ACOS hyper-converged software when ACOS is directly deployed in hyper-converged mode on x86_64 servers. In addition to basic features such as virtual machine lifecycle management, power operations, high availability, and cold and hot migration, AVE integrates with the ABS storage component to provide advanced virtual machine services such as instant snapshots, templates, and cloning.
AVE is easy to use. All operations on ACOS hyper-converged clusters can be performed through AOC.
AOC
Arcfra multi-cluster centralized resource management platform. It is designed to manage data centers within sites as well as multiple Arcfra clusters and system services across data centers, and provides complete and standard RESTful APIs and SDKs in multiple programming languages.
