Global namespace
You can combine the home and cache entities to create a global namespace.
Any client node can use the same path to connect to the data within any of the IBM Storage Scale clusters that are part of the namespace.
In such a global namespace, the following AFM features can improve application performance on a remote site:
- When a file is being read into a cache, the data can be read after it arrives in the cache.
- Multiple cache filesets can share a single file system. Data can be transferred between sites in parallel.
AFM also performs on networks that are unreliable, or have high latency. The following example is of a global namespace, which implemented by using AFM, with three different sites. An IBM Storage Scale client node from any site sees all of the data from all of the sites. Each site has a single file system. Each site is the home for two of the sub directories and cache filesets that point to the data that originates at the other sites. Every node in all three clusters has direct access to the global namespace.
The following figure shows global namespace that is implemented by using AFM.
