Amazon FSx is not one product but a family of four file systems, and the SAA-C03 rarely asks “what is FSx?” It asks which flavor a workload needs. The distractors are always other real FSx flavors or EFS, so the exam is testing whether you can read a requirement — a protocol, a platform, a performance profile — and land on the one file system built for it. The diagram maps each flavor to the use case that selects it.
The Windows counterpart to EFS
FSx for Windows File Server is the managed home for Windows workloads: SMB shares, NTFS permissions and Active Directory integration, with AWS running the servers. Think of it as the Windows counterpart to EFS’s Linux/NFS world. When a stem names Active Directory, SMB or an on-premises Windows file server, this is the answer — not EFS, which speaks only NFS.
FSx for Lustre is a parallel file system for high-performance computing, ML training and big-data processing — hundreds of gigabytes per second at sub-millisecond latency. Its signature exam trait is the S3 link: point Lustre at a bucket and objects appear as files, with results exported back to S3. That makes it the default storage layer for an S3-backed processing pipeline, with no manual copy step. Lustre also offers scratch and persistent deployment modes — scratch for cheap, regenerable, throughput-hungry data; persistent when the tier must be durable.
ONTAP for multi-protocol and NetApp continuity
FSx for NetApp ONTAP runs the full ONTAP feature set as a managed service. Its differentiators are multi-protocol access — NFS, SMB and iSCSI over the same data — and continuity for teams already invested in NetApp tooling like SnapMirror. When a scenario says “existing NetApp environment” or “serve the same data over more than one protocol”, ONTAP is the fit.
OpenZFS is the name-level answer
FSx for OpenZFS is the managed home for ZFS and NFS workloads, preserving ZFS-native snapshots and clones. The exam usually gives this one away at the name level: when the stem says a team is migrating a ZFS server and wants NFS with snapshots, choose OpenZFS.
The matching method
Every FSx scenario reduces to one requirement word. Windows, SMB, Active Directory → FSx for Windows File Server. HPC, ML, extreme throughput, S3-linked → FSx for Lustre. Multi-protocol or existing NetApp → FSx for NetApp ONTAP. ZFS migration with snapshots → FSx for OpenZFS. And when none of those apply — a plain Linux NFS share — the answer is not FSx at all, but EFS. The whole family exists to cover the cases EFS cannot, so the first fork is always “is this a Linux NFS job, or something more specific?”
How to use the seven scenarios below
Answer each one before opening the explanation, then find the requirement word that decided it. Every explanation names that word and takes apart why each other FSx flavor — or EFS — misses it. The recurring trap is a flavor that is genuinely a real file system but the wrong protocol or performance profile for the stated need. For the domain overview, see Design High-Performing Architectures, and for the block-versus-object-versus-EFS split, see EBS vs S3 vs EFS.
Updated for SAA-C03 (July 2026). The sample questions above are our own work
in the style of the exam — not real exam items. The exam itself is set and marked
by the certification body.