Disaster Recovery for App Engine - Multi Region

[horid121]

Hi, 

I'm using GAE standards and other services on GCP.
I realized almost GCP services support DR or I can prepare DR manually, but GAE (Google App Engine) Standard doesn't support DR automatically and even I can't prepare DR manually because I can't change the region of GAE Standard on the GCP console menu.

Here is my question. If I am using the GAE Region as "us-central1" and the us-central1 region becomes unavailable, I would like to know if there is a way to automatically recover it. If not, please let me know how I can manually recover it.

Thanks,

[System Diagram]

Related conversation: https://groups.google.com/g/google-appengine/c/3yqddF7UDy4/m/Y7BM3RCbAwAJ

[ErnestoC]

With App Engine, your application would be running in a specific region only. There is, however, zonal redundancy as the application is replicated in all zones of the chosen region. 

If you would like to use a managed service that supports multi-regional deployments for higher availability, you can try using Cloud Run, and deploy your services to multiple regions (with the help of a load balancer to route users to the nearest regions hosting your services).

You can also review the App Engine SLA for service guarantees and the general guide for Disaster Recovery for Google Cloud applications.

[horid121 · Accepted solution]

Hello, ErnestoC.
Thanks for your reply.

I understand most of your answers. But I have a few more questions.

1. I wonder if the data center (building) where us-central1 is located is automatically restored in another zone within the specific region if there is a fire or power is turned off.

2. The GAE standard of the us-central1 region seems to have a total of four zones (a, b, c, f). Are these four zones in one data center building? Or do you have more than one data center?

3. When migrating to Cloud Run, is there any problem in using the services (Mail, Memcache, Image, TaskQeue, Cron) in App Engine Standard?

Thank you.

[ErnestoC]

Google Cloud zones are not necessarily located in different physical data centers, given that zones are logical clusters of hardware, storage, and other resources in Google Cloud datacenters. You can review additional details about this abstraction in the zone virtualization documentation.

As for Cloud Run specifically, the container and storage are replicated across all zones in a region which would provide for redundancy for zonal failure. The same documentation page provides failure resilience details for a list of Google Cloud products, in addition to how to create a plan revolving around the SLAs provided by each Google Cloud service used in your organization.

[horid121]

@ErnestoC

Hi ErnestoC,
I hope you are doing well.

Please check my last questions and answer them.

Thanks,