The Blob Storage is really handy for storing temporary data structures during a scaled-out distributed processing. Yet, the lifespan of those data structures should not exceed the one of the underlying operation, otherwise clutter and dead data could potentially start filling up your Blob Storage
Temporary data in cloud computing is very similar to memory collection in object oriented languages,
when it's not done automatically by the framework, temp data tends to leak. In particular, in cloud computing, it's pretty easy to end up with storage leaks due to:
- Collection omission.
- App crash.
- Service interruption.
All those events cause garbage to accumulate into your Blob Storage. Then, it must be noted that for most cloud apps, I/O costs are usually predominant compared to pure storage costs. Enumerating through your whole Blob Storage to clean the garbage is likely to be an expensive solution.