Entra ID backup for 30K+ users

[sumeet]

Hello team,

We have a client with 30K+ users in Entra ID that is to be configured for backup. Our plan is to use v12.3 on windows. Have used Veeam calculator for Entra ID to anticipate the data usage and growth for the retention policy which is for 4+ years.

Entra ID log backup is fine, as the backup repository will S3 compatible storage. For Entra ID tenant backup, in PostgreSQL DB, we do not plan to use the default PostgreSQL installation as part of VBR (due to DB is setup on C drive). We will install PostgreSQL DB and use a different drive (with GPT and block size more than 4K, so that the drive can be expanded more than 16 TB (if required)). Use this doco https://helpcenter.veeam.com/docs/backu ... ml?ver=120 to adjust PostgreSQL configuration. The datastore for DB will be SSD.

We anticipate the first full backup to run for few days or a week. There is no option to split the job into batches of users, so a single job will run for 30K+ users. Hopefully the incremental runs are for few hours and complete through the night.

With M365 backup, we know throttling is a concern, but not sure if the same will apply for Entra ID backup.

We will also enable Conditional ACcess policies backup and also backup relationships between protected resources and management groups.

Does the above setup or design look appropriate?
Does anyone have similar experience of Entra ID backup for users which are 10K, 20K or 30K+ users? Any learning or concerns?

It will be good to know from PM if they have done any testing for Entra ID backup for such high user counts or know of any customer that have done such a backup and will be helpful to know any learnings or maybe just a smooth experience.

[PetrM]

Hello,

We conducted tests backing up test tenants with a total of 300,000 users. These backups were completed in minutes. However, in real-world scenarios where there are many relationships within a working tenant, the process takes longer, but still nowhere near days or weeks. For example, we observed that backing up a tenant with 10,000 users took an average of 30 minutes.

However, log backup performance is significantly slower, as we back up the last 30 days (this value can be changed via a registry key) during the first run. For this tenant, it took approximately 11 hours and 300 GB.

Thanks!