The problem
Data grows, and classic block storage often lacks the functionality for that kind of growth. At some point the problem stops being capacity and becomes the next migration onto the next system.
The approach
Software-defined storage. We work with Qumulo, Scality RING, Cohesity and IBM Spectrum Scale, because they have different strengths and the choice should follow the requirements rather than the other way round.
Before that choice there is a requirements engineering workshop. Not as a formality: without clarity on the access patterns, availability requirements and retention periods that actually apply, every product decision is guesswork.
What it actually buys
No more migrations. Once the data is on an SDS platform, the platform owns the lifecycle. Changing the hardware underneath becomes an operation rather than a project.
Availability to match. Synchronous or asynchronous mirroring into a co-location or into the public cloud, depending on what is needed.
Growth is designed in. Extending onto further clusters is part of the model rather than the exception.