# Distributed availability
<a name="genrel05"></a>


| GENREL05: How do you distribute inference workloads over multiple regions of availability? | 
| --- | 
|   | 

 Generative AI applications can be as simple as prompt-response workflows against a single foundation model or as advanced as multi-agent orchestration. The various components associated with a generative AI workload are required to service a region of availability. Availability could be over a well-defined zone or it could be expansive covering large geographic areas. Architecting for this variability is a complex problem. 

**Topics**
+ [GENREL05-BP01 Load-balance inference requests across all regions of availability](genrel05-bp01.md)
+ [GENREL05-BP02 Replicate embedding data across all regions of availability](genrel05-bp02.md)
+ [GENREL05-BP03 Verify that agent capabilities are available across all regions of availability](genrel05-bp03.md)