SVET Reports
Red Belly Blockchain Test
RBBC was tested on two VM’s and used Amazon’s high-end and low-end VMs for Intel Xeon E5–2666 v3: c4.8xlarge instances (pretty much the best what Amazon can now offer for $1.856 p/h), with 36 vCPUs, 60 GiB RAM, 10 Gbps network performance, 4000 Mbps and c4.xlarge (goes now for $0.232 p/h), with 4 vCPUs, 7.5 GiB, “moderate performance”, 750 mbts (pretty low, I agree). Still, RBBC was not tried on c4.large (2 vCPUs, 3.75 GiB, “moderate”, 500 mbts), which Amazon now offers for $0.116 p/h.
I wouldn’t, of course, argue that RBBC team did very impressive and thorough job comparing their protocol with DBFT, HBBFT and with CONS1 and running it, first, on 10 VMs (high-end) in 14 regions (140 stations) and then on 240 low-end VMs in 5 regions. They also assessed RBBC performance under two attacks scenarios.
RBBC demonstrated productivity is (depending on the number of requesters, latency, valid blocks etc.): lowest — 5,359 tps; highest — 30,684 tps.
Although, I think RBBC is more suited for and aimed by its authors at permissioned networks, we all, however, can only wish that more project teams will be willing and capable to get as far as RBBC authors in their attempt to disprove their own protocols.