用于Blob存储的Azure故障转移选项

Question

Maybe someone can point me out the direction, how should I implement fail-over scenario for my blob storage.

I am working with quite huge payloads, which is send to my API. The API ingest the payload with the metadata and save it to the blob. Everything seems to be working fine, except that I have SLA requirement, which define failover strategy in case Azure Blob Storage service is unavailable, the API should fall back and persist files locally or in any other location. After Azure Blob Storage is up and running, files should be transfer back to the blob for further processing.

My first idea was that I can preserve files locally in Azure App Service file system, as soon as files are shared between all instances of the service (when scale it up to multiple instances) how stated here , but internally it is still implemented using the same Azure Storage.

So what will be the best option here?

Answer 1

As aidden mentioned, your storage accounts will fail over to a different region automatically if an outage occurs. The preview feature is so you can force a fail over on your side. The SLA for Geo-Redundant storage is 99.9%, and will continue to function even in the case a whole region of Azure is down. This is sufficient for most applications.

As you have noted, App Services also makes use of storage so writing to disk will not be sufficient in the case of a widespread storage outage. Two options come to mind when dealing with large files that will stay up even if all of Storage were to go down.

1. Locate a secondary storage option somewhere outside of Azure.
2. While not designed for file storage, CosmosDB could handle the data temporarily. It is designed to be geo-replicated, handle large amounts of data, and you could even locate it in different regions from your storage accounts.