It sounds as a student task :smile: but actually it’s hard to say what’s the best. I would say that B and C are expected to be at least no worse than A. What would you say? :wink:
A lot of it depends on the hardware and software(AV/FW). In ideal environment it would be variant C, but as I've mentioned, it depends on various factors/bottlenecks, like cache settings.
In what I have experienced, with most all hardware being equal, though admittedly not exact, I experienced the following:
B: 4- 1 Gbe LACP upload 280 Mbps, download 766 Mbps
C: 2 - 10 Gbe LACP upload 143 Mbps, download 395 Mbps
(Speeds are measured by using LANSpeedtest lite)
With this being seen, I can only imagine the only rationale behind this is the possibility of multi-threading enabling B to be faster. I don't know if Cloudberry developers are watching this, but it would be nice to hear from them if multi-threading is in fact being utilized.
If someone has the time and resources, it would be nice if someone else can simulate a similar environment and see what they get.