Hello all, I am looking to build a scalable MySQL data warehouse and am considering GlusterFS. Specifically, I would like to use the Switch Scheduler to store MyISAM and Infobright tables across the parallel disk cluster. Perhaps something like option switch.case a*:brick1, 1*:brick1, b*:brick2, 2*:brick2, c*:brick3, 3*:brick3, etc.... Furthermore, I suspect the Switch Scheduler could reference a subvolume where GFS1 handles tables 0*, 3*, 6*, 9* and GFS2 handles 1*, 4*, 7* and GFS3 handles tables 2*, 5*, 8* (and alpha in a similar manner), thus somewhat abstracting the logical versus physical storage. To scale horizontally, more nodes could be added to GFS1, GFS2, and GFS3 as performance/storage requirements dictate. Are there any examples of such a thing being done? Am I WAY off base here? Any guidance would be helpful as we are still in the planning stages of this. Many thanks! --Charlie
