I have 2 clusters
but both have several advantages/disadvantages
I am asking for advice for best architecture setup from performance/reliability perspective.
I am running 2 production clusters
- virtualised - 13nodes , each node: (8core/64G RAM/1-2TB SSD)
- dockerised - 3 containers, each container:(12core/30G/1,5TB SSD)
- no redundancy , 3 dockers on 1 physical hardware, even though the cluster is among 3 physical node, loosing I server means 3 Elastic nodes down and cluster in RED status.
- more complicated setup, I cant set specific CPU and RAM per docker,logs are not accessible as easy as in standard installation
?how to achieve redundancy if multiple dockerised nodes are on the same node. ?
- low IO storage performance (our private cloud is VMWare with VSAN storage)
- storage is expensive (it is in raid and due to Vsan architecure)
? how do you achieve high performacne + virtualisation?
Is the baremetal the choice for Elastic Clusters?
No virtualisation nor dockers means having very specific hardware to achieve same size for each node and all known complications.