It's normal to expect such downtimes in artificial intelligences, like host servers, besides, your plans of migrating the entire system to a high availability cluster, to me is a huge plus.