A customer is building a web cluster in which all nodes must access a shared pool of images. Which of the following storage types would be BEST for this workload?

Block storage is attached to a single VM at the OS level and cannot be natively shared across multiple cluster nodes simultaneously without an additional clustering file system layer.

File-level access storage (NFS/SMB) can be shared but introduces a single network file system endpoint that becomes a bottleneck and adds protocol overhead compared to native HTTP object access for web workloads.

Direct attached storage is physically bound to one server and cannot be shared across a cluster by definition, making it unsuitable for any shared-access workload.

Object storage (such as Amazon S3 or Azure Blob Storage) exposes each image as an individually addressable object via a REST/HTTP URL, allowing every node in the web cluster to read and serve images concurrently without contention or OS-level mount dependencies. Its flat namespace, per-object addressing, and built-in scalability make it purpose-built for large pools of binary assets like images in web workloads.