Enterprise Vault has a Safe Copy Check feature that can be enabled by customers at the Vault Store level. This check is to ensure that content archived by Enterprise Vault is not replaced in its original location by an HTM link until it has been backed up to a second archive device through the device’s replication functionality. 

This approach is commonly used by customers who use archive devices with replication functionality that have an eventual consistency model. With Dell Centera, for example, it can be several minutes before replication has created the second copy of the archive content, and the Safe Copy Check is commonly used.

ECS implements a strong consistency model and the latencies that exist in the Centera replication queues do not exist in ECS. Customers can be assured that an object that is written to one Virtual Data Center in a Replication Group will also be available to all other Virtual Data Centers in that Replication Group. However, customers might regard even the smallest delta-t window as being unacceptable and might want to perform the Safe Copy heck.

The concept of a safe copy is a check that an archived Saveset is geo replicated to a secondary VDC. However, within a VDC, archived files are always stored in a resilient manner (such as Erasure Coding). For more information, see the ECS storage documentation.

When Enterprise Vault makes a Safe Copy Check, the ECS Streamer driver responds in one of three ways.

Disabled

The ECS Streamer driver does not check the replication status of the saveset and returns true to Enterprise Vault.

dtQuery

The ECS Streamer uses the ECS dtQuery functionality to detect the replication status of the saveset. If it is 100 percent replicated, the Streamer returns true to Enterprise Vault; otherwise, it returns false.  

This check uses port 9101, and this must be considered when you plan load balancers and firewalls. Port 9101 is fixed and cannot be mapped to another. The dtQuery option is not available for ECS 3.5 and later.

replicationInfo method

This is an extension to the ECS S3 API to check the replication status of an object. It became available with ECS 3.4. It does not require port 9101 because it uses the port used by S3 requests.  

Request:
GET /bucket/key?replicationInfo
Response:
<ObjectReplicationInfo xmlns=http://s3.amazonaws.com/doc/2006-03001/
<IndexReplicated>false</IndexReplicated>
<ReplicatedDataPercentage>64.0</ReplicatedDataPercentage>
</ObjectReplicationInfo>

This call checks the replication status of both the object and the supporting index blocks it uses.

Replication of the object is very quick, but with ECS systems that have a low archive rate, the indexes might take a long time to replicate.

The ECS Streamer can be configured to return true for a Save Copy Check if both the object and indexes have been fully replicated or true when the object has been replicated and ignores the replication state of the indexes.

The recommended setting for the replicationInfo method is to check the replication status of both the object and indexes.

Customers who are experiencing delays in index replication due to low throughput can choose to ignore the state of replications of the indexes. They should discuss this with their Dell Support team. 

The following table shows the three possible values for the Safe Copy Check parameter. 

Table 3.      Safe Copy Check property