High Availability and Redundancy for Oracle FastConnect

There is no inherent redundancy in a FastConnect virtual circuit. You can add redundancy by building a secondary connection across unique equipment to avoid having a single point of failure. For more information, see OCI Docs - FastConnect Redundancy Best Practices.

You can use a combination of FastConnect connections (layer 2) and IPSec VPN connections (layer 3) when building redundancy.

Connectivity model

Hosted connections follow Oracle’s Partner connectivity model.

When planning redundancy, keep in mind that you should be configuring the BGP session between your on-premises port and Oracle (not PacketFabric).

Physical redundancy on the user-end

Users should ensure physical redundancy by setting up each connection on different hardware.

Provisioning a secondary connection

When provisioning a secondary connection for redundancy, follow the process as outlined in Create an Oracle FastConnect Partner Connection.

Choosing availability zones for high availability

When a user selects an availability zone, they connect to PacketFabric’s infrastructure, which corresponds to hardware on OCI known as “availability regions”. By establishing connections to two different availability zones, you ensure that each connection is isolated from a failure in the other.

Screenshot of availability zone selection

Users must create a connection in the same location as the existing port in order to achieve redundancy.

Verifying redundancy status

Once redundant connections are established, it is important to make sure that either of the connections can function in the case of a failover in one. Users can verify whether or not connections to a dynamic routing gateway (DRG) are redundant through the Oracle Cloud Console.

Screenshot of Oracle’s DRG menu

Navigate to your DRG status from the main menu by selecting “Networking” and then “Dynamic Routing Gateway”. Source: Oracle

When to employ active/passive or active/active connections

Depending on your application requirements, you can implement these connections in either active/active or active/passive (standby) mode. In active/active mode, both connections are constantly in use and if one connection fails, its load is borne by the other. In active/passive mode, one connection is always in use and the other stands by as a backup in case the first fails.

For stateless applications, where past transactions are irrelevant, Oracle recommends configuring your connections as active/active. However, Oracle recommends active/passive connections for stateful applications where the context of previous transactions is involved.

Configuring routing correctly

In order to run connections in active/passive mode, users should configure local preference values such that one virtual circuit’s (VC) preference value is higher than the other. The VC with the higher local preference value will remain active while the other VC will only receive traffic if the primary VC is offline.

If set to active/active mode (no difference in local preference value), Oracle’s algorithm will prefer to send traffic to the oldest established route.

Process illustration

Additional resources

Oracle documentation

Connectivity Redundancy Guide

FastConnect Redundancy Best Practices

Oracle YouTube channel

YouTube: HA and DR - Level 300 - Part 1 - High Availability