> For the complete documentation index, see [llms.txt](https://documentation.alluxio.io/ee-ai-en/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://documentation.alluxio.io/ee-ai-en/performance.md).

# Performance

Alluxio includes several features designed to accelerate data I/O and metadata operations, ensuring your applications run at maximum speed.

* [**Read Optimization**](/ee-ai-en/performance/file-reading.md): Use client-side prefetching and large file segmentation to maximize read throughput.
* [**Metadata Optimization**](/ee-ai-en/performance/metadata-listing.md): For directories containing millions of files, use the Index Service to create a distributed, scalable cache for directory listings, dramatically speeding up metadata operations like `ls`.
* [**S3-API Write Optimization**](/ee-ai-en/performance/s3-write-cache.md): Buffer writes in Alluxio's cache layer and asynchronously persist them to underlying storage, reducing write latency for workloads like training checkpoints and ETL pipelines.
* [**FUSE Write Optimization**](/ee-ai-en/performance/fuse-write-cache.md): Buffer POSIX writes in Alluxio's Write Cache layer and asynchronously persist to UFS, enabling low-latency writes via standard filesystem calls for write-cache workloads like model checkpoints and ETL outputs.
* [**FUSE Full POSIX Workspace**](/ee-ai-en/performance/fuse-workspace.md): Deploy FUSE Write Cache with full POSIX semantics — including random writes, overwrite, truncate, rename, and symlinks — using FoundationDB as a distributed metadata backend. Supports multi-node access with Worker NVMe or UFS PageStore.
* [**Model Loading Optimization**](/ee-ai-en/performance/model-loading.md): Accelerate model checkpoint loading for ML training and inference workloads using Alluxio's intelligent prefetching and shared memory pool.
* [**UFS Bandwidth Control**](/ee-ai-en/performance/ufs-bandwidth-limiting.md): Configure a rate limit on reads from the UFS to prevent Alluxio from overwhelming the underlying storage system during cache-filling operations.
* [**RDMA Networking**](/ee-ai-en/performance/rdma-networking.md): Configure high-speed network technologies such as IPoIB (IP over InfiniBand) for AI and HPC clusters to maximize Alluxio network throughput.