1. Administrator Guide
  2. Disaster Recovery
  3. Enabling Disaster Recovery on the Appliances

Enabling Disaster Recovery on the Appliances

This section explains how to connect the systems that participate in the disaster recovery setup. It requires two Oracle Private Cloud Appliance systems installed at different sites, and a third system running an Oracle Enterprise Manager installation with Oracle Site Guard.

Oracle Private Cloud Appliance racks that have been factory reset to the 302-b892153, 302-b925538, or 302-b946415 versions need to have a common encryption key for the ZFS Storage Appliance storage pools at both the source and destination.

If you supply outside certificates to establish a CA trust chain for the Oracle Private Cloud Appliance, you must add two PTR records to the Data Center DNS when you set up disaster recovery. A PTR (Pointer record) in DNS maps an IP address to a hostname. This behavior is the reverse of the usual IP address lookup for a supplied hostname, which is provided by an A record in DNS.

You must create two ReverseIp lookup zones for the two ReplicationIps used in disaster recovery. The DNS requests are forwarded to the Private Cloud Appliance in the same way as requests for the Private Cloud Appliance Service Zone are forwarded. If only the zfsCapacityPoolReplicationEndpoint is defined, then only a PTR record for that IP address in is needed.

To create a ReverseIp lookup you need to create a DNS zone for the ReverseIP lookup. You create one or more reverse lookup zones depending on how the Replication IPs are configured. How to create these PTR records depends on the interface for the Data Center's DNS servers.

For example, if the rack domain is myprivatecloud.example.com, and the Capacity Pool IP is 10.170.123.98 and the Performance Pool IP is 10.170.123.99, the Private Cloud Appliance requires two zones with the following mappings: 

98.123.170.10.in-addr.arpa rtype PTR rdata sn01-dr1.myprivatecloud.example.com
99.123.170.10.in-addr.arpa rtype PTR rdata sn02-dr1.myprivatecloud.example.com

Collecting System Parameters for Disaster Recovery

To set up disaster recovery for your environment, you need to collect certain information in advance. To be able to fill out the parameters required to run the setup commands, you need the following details:

  • IP addresses in the data center network

    Each of the two ZFS Storage Appliances needs at least one IP address in the data center network. This IP address is assigned to the storage controller interface that is physically connected to the data center network. If your environment also contains optional high-performance storage, then two pairs of data center IP addresses are required.

  • Fully Qualified Domain Names (FQDNs) in the data center network

    If you have upgraded your racks to 302-b892153, you need to use the FQDNs of the hosts and not their IP addresses. This FQDN is assigned to the storage controller interface that is physically connected to the data center network. If your environment also contains optional high-performance storage, then two pairs of data center FQDNs are required.

  • Data center subnet and gateway

    The ZFS Storage Appliances need to be able to exchange data over the network. Their network interfaces connect them to a local subnet. For each interface included in the disaster recovery configuration, the subnet address and gateway address are required.

To complete the Oracle Site Guard configuration, you need the following details:

  • The endpoints of both Private Cloud Appliance systems, where API calls are received. These are URIs, which are formatted as follows: https://<myRegion>.<myDomain>

    For example:

    https://myprivatecloud.example.com

  • An administrative user name and password for authentication with the Private Cloud Appliance services and authorization of the disaster recovery API calls. These credentials are securely stored within Oracle Enterprise Manager.

Connecting the Components in the Disaster Recovery Setup

The ZFS Storage Appliances installed in the two Oracle Private Cloud Appliance racks must be connected to each other, in order to replicate the data that must be protected by the disaster recovery setup. This is a direct connection through the data center network; it does not use the uplinks from the spine switches to the data center.

To create the redundant replication connection, four cable connections are required at each of the two sites. The ZFS Storage Appliance has two controllers; you must connect both 25Gbit SFP28 interfaces of each controller's first dual-port Ethernet expansion card to the next-level data center switches. At the other site, the same four ports must also be cabled this way.

The replication connection must be used exclusively for data under the control of disaster recovery configurations. Any other data replicated over this connection might be automatically destroyed.

In the next phase, the network configuration is created on top of the interfaces you cabled into the data center network. On each storage controller the two interfaces are aggregated into a redundant 25Gbit connection. The aggregation interface is assigned an IP address: one controller owns the replication IP address for the standard performance storage pool; the other controller owns the replication IP for the high-performance storage pool, if one is present.

Note:

Link aggregation needs to be configured on the data center switches as well. The MTU of the ZFS Storage Appliance data links is 9000 bytes; set the data center switch MTU to 9216 bytes.

The administrators at the two sites are not required to configure the replication network manually. The configuration of the ZFS Storage Appliance network interfaces is automated through the drSetupService command in the Service CLI. When executing the command, the administrator provides the IP addresses and other configuration settings as command parameters. Use of the drSetupService command is described in the next section.

Your Oracle Enterprise Manager does not require additional installations specific to Private Cloud Appliance in order to perform disaster recovery tasks. It only needs to be able to reach the two appliances over the network. Oracle Site Guard is available by default in the software library of Oracle Enterprise Manager.

To allow Oracle Site Guard to manage failover operations between the two Private Cloud Appliance systems, you must set up both appliances as sites. You identify the two sites by their endpoint URIs, which are used to configure the disaster recovery scripts in the failover operation plans. You also provide a user name and password to allow Oracle Site Guard to authenticate with the two appliances.

For additional information and instructions, please refer to the product documentation of Oracle Site Guard and Oracle Enterprise Manager.