Planning a resiliency domain for efficiency and fault tolerance
Before you deploy Veritas Resiliency Platform, you should plan how to scale the deployment for efficiency and fault tolerance.
Although a resiliency domain requires only one Resiliency Manager, you can add multiple Resiliency Managers instances to the domain. For example, you can distribute Resiliency Managers geographically for efficiency of local access. For resiliency, you can even have multiple Resiliency Managers in one data center.
If you have multiple Resiliency Managers in the resiliency domain, the recommended WAN latency between two Resiliency Managers is recommended to be less than 30 milliseconds. The maximum WAN latency allowed between the two Resiliency Managers is 50 milliseconds. This latency requirement applies to any two Resiliency Managers - within a single data center or across two different data centers.
The recommended minimum deployment for disaster recovery to on-premises data center in production environment would be four virtual appliances: a Resiliency Manager and Infrastructure Management Servers (IMS) in the source data center and a Resiliency Manager and IMS in the target data center. In a test environment, you can have only one Resiliency Manager at the target data center.
The recommended minimum deployment for disaster recovery to cloud data center would be three virtual appliances: an IMS in the source data center and a Resiliency Manager and IMS in the target data center.
The source and target data centers do not require a one-on-one mapping of IMSs. For example, you can have two IMSs in the source data center and one IMS in the target data center. You can add multiple Infrastructure Management Servers (IMS) to a resiliency domain. For example, if there are multiple data centers in different geographical locations to be managed, you configure a separate IMS for each geographical data center location. You can also configure more than one IMS in the same data center.
If you plan to use Resiliency Platform data mover for replication, then additionally you need minimum one Replication Gateway in each data center. Resiliency Platform supports asymmetric pairing of Replication Gateways. This feature facilitates deployment of only the required number of Gateways on each side, based on data transfer rate and technology specific limits. One Gateway on source site can be paired with multiple Gateways on target site and vice versa. One Gateway can be paired with up to 16 gateways on the peer site.
Both the source and the target gateway must have external storage equivalent to 6GB for each virtual machine protected by the gateway pair. For example, if a gateway pair supports 10 virtual machines, the source and recovery (target) gateway must each have 60GB of external storage. The minimum size of the external storage must be 50GB.
Veritas Resiliency Platform is dependent on host name resolution to work between the resiliency platform appliances and its configured assets. Trying disaster recovery when there is a partial network infrastructure failure, such as DNS failure, can lead to incomplete disaster recovery activities, if the appliances and assets are not configured using network host names.
In order to avoid encountering such issues, you must use fully qualified network host names in the following scenarios:
Joining a new resiliency manager to a domain
Adding an infrastructure management server to a data center
Adding a resiliency platform datamover to a data center
Preparing hosts for replication using resiliency platform data mover
Adding application hosts
Adding Hyper-v servers
Adding vCenter server
Adding Storage Arrays
Adding Storage Discovery Host
Table: Reference topics
For more information on... | Refer to... |
|---|---|
Network | |
Capacity planning for Resiliency Platform replication appliances | |
Network objects and mapping of network objects required across data centers |