Latency could be an issue but if you were to align the region you use to a hyperscaler region that’s close geographically, it’d likely be close by ping time too. If you find one close enough, it might be effectively equivalent to a cross-availability zone network hop within a hyperscaler region to begin with.
Data transfer cost is another consideration: for larger-scale use cases that move a lot of data over the wire, this could be a problem
1 Like