A business continuity plan (BCP) is a critical part of any organization's risk management strategy.
BCP is automated for Vespa Cloud applications that are deployed to more than one zone, and are using the global endpoint to serve query requests. See deployment.xml for details on how to configure the BCP automation. The automation ensures that the application will continue to operating during an unplanned disruption in the service in a zone.
In case of a production emergency, a zone can manually be set OUT
using the
routing control to prevent it from receiving requests.
Application owners are encouraged to test their BCP setup regularly. Use this to verify that the application is able to
perform with peak traffic routed to n -1
zones.
Monitoring of the traffic shifts during the BCP is done using the monitoring dashboard available in the Cloud Console. All the zones can be
displayed in the same graph, and the traffic can be observed in real time. To monitor all zones, replace the request
parameter zone
in the URL with regexp(prod)
.
The process is as follows:
OUT
IN
for 30 minutesIN
OUT
for the one zone that remained IN
in step 3OUT
to another zone for 30 minutesIN
The BCP certification for 2024 was monitored by the Yahoo Inc BCP team, completed on 2024-02-08. The certification was successful. See the bcp_test_2024.pdf for details.