A Nutanix administrator is tuning storage efficiency on two containers: one hosts a random-write-heavy database, and the other hosts a sequential-write backup target whose data is written once and rarely overwritten. Which placement best matches Nutanix storage efficiency behavior to these I/O patterns?

Data-reduction features in AOS are set per container, which is why the two workloads go in separate containers with different settings. EC-X (erasure coding) replaces full replicas with data and parity strips, cutting RF2 overhead from 2x toward about 1.25x with a 4/1 strip. Curator applies it in the background to write-cold data, meaning extents that haven't been overwritten for a while, so a backup target that is written once and then read occasionally gets the full benefit. On a database with constant random overwrites, encoded strips would keep being rewritten, which adds read-modify-write and rebuild work and erodes the savings, so EC-X is not recommended there. Compression is the efficiency feature that suits almost every workload: inline compression handles large or sequential writes as they arrive, and post-process compression catches the rest, at little CPU cost. Deduplication is narrower; it pays off when many VMs share identical blocks, such as full-clone VDI or base images, and does little for a typical database. Using EC-X for random writes or for both containers ignores the overwrite penalty, while compression alone on the backup target leaves EC-X capacity savings unused. Exam caveat: EC-X needs enough nodes for the strip plus one, and it always keeps the container's failure tolerance. Operational check: review write patterns in Prism and the container's data-reduction ratio after Curator scans complete.