Mapfs Professional Edition
Mapfs is a small, user space file system built for Object Storage with POSIX Compatible. Mapfs is designed to speed up Cloud Object Storage access with optimized cache mechanism.
Mapfs supports Azure Blob Storage and AWS S3-like Object Storage, e.g. Wasabi, Google Cloud Storage, Oracle Cloud Storage, Aliyun, CloudFlare R2, etc.
Mapfs excels at...
- Fast data access
- Data Access Pattern: Write once, Read more;
- Lightweight solution: mapfs package only takes ~5MB;
- Keep your data intact: mapfs doesn't need third-party metadata server, doesn't need to encode your data into private format.
- Plugin-Plug out solution: It's a pure software solution. If you don't want it any more, just remove it and your production still runs as before.
Mapfs is less suitable for...
- Real-time synchronization
- Full POSIX Compatible
- Complicated ACL Control
Consistency
Mapfs supports final consistency.
If Mapfs is deployed in a cluster environment, the synchronization latency depends on the cache timeout policy.
By default, the cache timeout is 5 minutes, and that's logically the upper time latency for data changes to be synced to another mapfs client. Of course, you can reduce the cache timeout from 5 minutes to 10 seconds for example, then the maximum time latency for one mapfs client realized the data changes from another mapfs client is 10 seconds.
Anyway, mapfs doesn't support real-time synchronization. So, consider your workflow schema, if real-time sync is mandate, mapfs doesn't applies to. Otherwise, mapfs is a better choice.
Mapfs limits
- Maximum file length is 100GB
- Metadata cache size: 1000 items
- Data Cache Size: by default 25GB, at most 2TB
- Data Cache granularity: 512 bytes
- Default Cache Timeout: 5 minutes
- Mapfs consistency: final consistency
- Maximum Storage Mappings: 100
- Maximum Mapfs Agents: Unlimited, only limited by system resources
- Not Full POSIX Compatible