Multi Cloud Object Storage Performance Comparison for AI Data Pipelines
Storage bottlenecks—not algorithms—determine how fast your GPUs actually train.
Priya Subramaniam
Contributing Editor
Priya Subramaniam
Priya worked as a kernel developer on POSIX compliance tooling at a enterprise Linux vendor for nearly a decade before transitioning to writing and editing in 2018. Her coverage centers on the subtle correctness guarantees—and frequent violations—that emerge when agents interact with operating system interfaces.
8 stories
Checkpoint Design Patterns for Long-Running AI Agents
Preserving agent progress between sessions requires storing minimal state outside the model.
Priya Subramaniam
Unacknowledged Fsync on Cloud Filesystems Leading to Silent Data Loss
Cloud filesystems silently lose data when fsync errors go undetected and unreported.
Priya Subramaniam
Hard Links and Symlinks on Object Storage Filesystems
Hard links need inode tables that object storage's flat key model simply doesn't provide.
Priya Subramaniam
Sparse File Support in Object Storage Filesystems
Object storage can offer sparse files, but only by building the entire illusion itself.
Priya Subramaniam
Crash Consistency Models in FUSE Filesystem Implementations
Write-back caching and crash consistency are fundamentally at odds in FUSE filesystems.
Priya Subramaniam
Flock and Fcntl Locking Semantics on Distributed Filesystems
Network filesystems silently break locking assumptions built for single machines.
Priya Subramaniam
Fsync Behavior on NFS and Cloud-Mounted Filesystems
Network storage renegotiates fsync's durability promise in ways most developers never notice.
Priya Subramaniam