Updated on 2024-11-05 GMT+08:00

SDRS Usage Restrictions

Before using SDRS, learn about the constraints listed in the following tables.

Table 1 Asynchronous replication constraints

Constraint

Description

Specifications

Constraints on specifications:

  • Minimum specifications recommended for the cloud disaster recovery gateway, production site servers, and disaster recovery site servers are 8 vCPUs and 16 GB memory.
    NOTE:

    For disaster recovery site servers, you are advised to reserve 2 GB memory for reverse reprotection.

  • One disaster recovery gateway can protect a maximum of 20 protected instances and 58 disks.

Server

Constraints on servers:
  • Protected instances can be created for ECSs.
  • Only KVM ECSs are supported.
  • Both x86 and Arm ECSs are supported.
  • Protected instances cannot be created for ECSs of the following series:
    • General Computing-plus C7
    • Memory-optimized M7
    • Disk-intensive
    • Ultra-high I/O
    • GPU-accelerated
    • FPGA-accelerated
  • The OS must meet compatibility requirements.
  • If you delete a server or disks that have been used to create a protected instance, the protected instance will become invalid.

Disk

  • Shared disks and encrypted disks are not supported.
  • Servers' local disks are not supported.
  • General Purpose SSD V2 disks cannot be used as disaster recovery disks.
  • Disks using LVM are not supported.

Network

  • Round-trip time (RTT) ≤ 100 ms
    NOTE:

    You can run the ping command and set the packet size to 64,000 bytes to test the latency between the production site and the OBS domain name at the disaster recovery site.

    ping OBS domain name at the disaster recovery site -s 64000

  • Bandwidth ≥ Changed data volume per minute during peak hours/60 seconds. Minimum bandwidth: 10 Mbit/s
    NOTE:

    Obtain the total amount of changed data during the period with the maximum service load and calculate the peak write bandwidth of the service disk using the changed data volume and time period. To meet the RPO requirements, use the obtained peak bandwidth as the minimum bandwidth. In cross-region replication scenarios, if the network bandwidth is shared, obtain the peak bandwidth of each protected instance, and use the maximum peak bandwidth as the minimum bandwidth.

    Note:

    1. Service load statistics are collected by server.

    2. The baseline bandwidth of the cloud disaster recovery gateway must meet the bandwidth requirements. The recommended minimum baseline bandwidth is 2 Gbit/s.

  • Packet loss rate < 0.1%

Function

Constraints on functions:

  • All disks on an ECS are protected. After protected instances are created, new disks cannot be added to and existing disks cannot be removed from the protected instances, and existing disks cannot have their capacities expanded.
  • Data consistency between protected instances is not supported.
Table 2 Synchronous replication constraints

Constraint

Description

Compute

Constraints on server types:

  • Kunpeng ECSs are not supported.
  • x86-based, GPU-accelerated and FPGA-accelerated ECSs are not supported.

Replication

Constraints on servers:
  • The servers must be from two AZs of the same region.
  • BMSs are not supported.
  • The following types of servers cannot be used to create protected instances:
    • Large Memory (Xen): This type of servers is bound to InfiniBand NICs.
    • Disk Intensive: This type of servers has local disks.
    • Ultra-high I/O: This type of servers has local disks.

Constraints on EVS disks:

  • Disks used to create replication pairs cannot be deleted, and the disk snapshot cannot be used to roll back data.

Storage

Only servers using EVS or DSS disks are supported.

Application

Storage-based synchronous replication ensures disk data consistency but does not guarantee application data consistency. If your applications support crash consistency, you can run and replicate applications.

Deployment model

VPC migration: Servers at the production site and those at the disaster recovery site are in the same VPC. NIC migration and multiple NICs are supported for each server.

Service interconnection

Tag Management Service (TMS) is supported only via API.

Backup and restore

Only servers at the production site can be backed up and restored. Servers at the disaster recovery site can only be backed up.

If the production site AZ becomes faulty, you can run a disaster recovery drill to recovery the services on servers.