Is network congestion the primary concern when load balancing Glance?
To load balance Glance in OpenStack, the book OpenStack Essentials recommends using a shared storage device. The book describes connecting this device to two different Glance repositories. I see this configuration as potentially balancing network traffic and providing redundancy there. But I see this configuration as not helping read/write demands on the shared device. I would think that load balancing Glance would be mostly important for I/O on the disk of images.
Is the load balancing of network traffic a big concern for Glance (and not disk I/O)?
Perhaps the book presumes the reader to have limited resources and balancing the I/O would involve replication and considerable work to achieve.