yes yes but I’m not speaking about that multi-tenant instancing. I am speaking of HA multiple cluster setup (where you have ONE single dgraph DB, but multiple alpha nodes (for sharding?? that’s wrong or?) and high availability). My question is, if HA multiple cluster setup won’t have bad latency since queries travel across multiple machines and that causes latency. Because I thought like you said, that dgraph balances/shardes predicates based on disk usage, this leads to my next question about that:
I saw now the video:
Every group is an own machine, right? So we have 3 machines?
Group 1 = name
Group 2 = director film + initial release date
Group 3 + netflix data
I thought dgraphs shards/balances predicates based on disk usage. Why are they now sharded/balanced based on name, directorfilm+releasedate, netflix data? I think that’s the spot where I have a knowledge gap, can you please explain that to me bro?
The brother in the video also said at 13:00 ‘‘if one machine goes down, then you can still have another alpha nodes of your group’’. So, is every alpha node now on ONE machine, or every group on ONE machine? With ‘one machine goes down’, does he mean the actual hardware machine (e.g RAM bricks or CPU bricks or SSD breaks, or electricity outage), or the VM machine?
One last question: The setup on the picture is a HA(high availability) setup, correct? Is every group one machine? My assumption: No because that doesn’t make any sense. I think that Z1 A1 A4 A7 are one machine. Z2 A2 A5 A8 one machine, and Z3 A3 A6 A9 one machine. is that true? But what do the groups mean then? Because in the video I understand that every group is one machine. but that doesn’t make sense. I am confused. (my english understanding is quite bad and the auto generated english subtitles are as you know not the best, so maybe it is an understanding mistake. I am sorry.)
thank you very much and kind regards!!!