AWS DRS individual replication settings
AWS Elastic Disaster Recovery will attempt to help save you costs by consolidating the replication of as many source servers as possible onto the same Replication Server based on the individual Source Server Replication Settings. Source Servers must have identical Replication Settings to be considered for consolidation, and must not have Use dedicated replication instance enabled. For example, DRS will not consolidate Source Servers that have a different Staging area subnet specified in their Replication Settings. To reduce EC2 usage, we recommend having as many as possible Source Servers have identical Replication Settings to one another.
Modifying the Replication Settings of an existing Source Server can impact existing replication, depending on the settings configured. Additionally, most Replication Settings options can be modified in bulk for multiple Source Servers through the AWS Elastic Disaster Recovery Console:
| Replication Setting | Impact | Bulk Editing |
|---|---|---|
|
Staging area subnet |
Small pause while reconnecting Source Server to new Replicator. |
Supported |
|
Replication server instance type |
Small pause while reconnecting Source Server to new Replicator. |
Supported |
|
Dedicated instance for replication server |
Small pause while reconnecting Source Server to new Replicator. |
Supported |
|
EBS encryption |
Full Sync may be required. |
Supported |
|
Data Routing (Private IP) |
No impact. |
Supported |
|
Network Bandwidth Throttling |
No impact. |
Supported |
|
Point in time (PIT) policy |
No impact. |
Supported |
|
MAP program tagging |
No impact. |
Supported |
|
Tags |
No impact. |
Supported |
Replication server configuration
Replication Servers are AWS EC2 Instances automatically launched by AWS Elastic Disaster Recovery to support Continuous Data Replication from Source Servers.
Staging area subnet
The Staging area subnet setting defines which VPC Subnet that the a Replication Server for a Source Server will use. A Source Server must be able to successfully initialize connections to the subnet configured within its Staging area subnet setting. The best practice is to create a single dedicated, separate subnet for recovery in your AWS Account. Learn more about creating subnets in this AWS VPC article. Unless Use private ip is enabled and valid routing within the VPC exists, Replication Servers must be in a Public subnet. By default, a Replication Servers will assign itself a Public IPv4 without any additional configuration needed.
Replication server instance type
The Replication server instance type will determine the EC2 Instance type and size that will be used for the launch of a source server's replication server. DRS Replicators only support EC2 Instances with x86_64 CPU architecture.
By default, AWS Elastic Disaster Recovery utilizes the t3.small instance type, and should work well for most common workloads. We recommend monitoring the Cloudwatch metrics of a replication server, if your Source Server is experiencing frequent Lag or Backlog. Metrics to monitor include DiskWriteBytes or DiskWriteOps, which may indiciate the Replication server instance type is improperly sized to protect your source server.
AWS Elastic Disaster Recovery supports replicating Source Servers with up to 60 volumes, however the Replication server instance type must also support and equal or greater number of EBS Volume attachements. We recommend reviewing the Dedicated Amazon EBS volume limit Documentation to ensure an appropriately sized EC2 Instance Type is selected.
Dedicated instance for replication server
The Dedicated instance for replication server setting specifies whether or not the Source Server's can use to a Replication Server shared with other Source Servers. By default, AWS Elastic Disaster Recovery will attempt to consolidate as many Source Servers as possible onto a single Replication Server, based on a variety of factors. Setting Dedicated instance for replication server to use dedicated replication instance will ensure only this Source Server will replicate data to the Replication Server.
We recommend leaving Dedicated instance for replication server as do not use dedicated replication instance unless the Source Server is experiencing frequent Lag or Backlog due to sharing a Replication Server with other Source Servers. Using a dedicated replication server may increase EC2 costs associated with protecting a Source Server.
