It’s better to centralize everything, so its 1,5Gbps on 1 machine, or 100Mb on 15 virtual machines?
PC Specs: 128gb RAM, 2950X Threadripper, 16TB on 8HDD SATA3, 2 interfaces of ethernet, 1gb each one, 570 GPU
NEW: The thing is here, on the calculator, the max Bandwith per node is 100Mb, i am paying another 1Gb connection from my ISP so i want to use all of this for this, the question was about the limit of the storj service, and if it hasn’t got limit, which thing is better, i am calculating aprox 1 node equals 176 per month with 100Mbits bandwith up and down, no limit on bandwith used per month and 1TB per “node”
i just run one big node, like many others… but there is right answer here… both approaches work just fine
storj aims after there being no particular advantage of running multiple nodes, ofc there will always be some major differences, like overhead, administration of multiple nodes, redundancy.
i like that it’s one central system so that i can use it for other stuff, and still get the performance of all the hardware on whatever task i put it towards.
also if one node data was particularly in demand, then all my drives can help supplying that particular data at speed without putting any focused load on a single drive… or vm / container
having all the data in one big dataset also allows me to use sizable cache devices
I’d avoid the overhead of 15 VMs. If you want them on a separate VM by all means, just do one VM for all nodes. Despite the fact 100 Mbps is more than enough for one nodes a bigger shared connection is better for spikes of traffic than multiple small ones.
There is no advantage or disadvantage either way from the nodes perspective. It’s up to you.
i have seen peaks of 120-130mbit tho… but only very briefly max sustained for hours has been 50-60mbit, again a great point for keeping it all in one place and or not limiting the vm’s to a certain bandwidth.
i use windows GUI with 5 nodes on each PC, it much esier than run so many virtual machines, they will eat lot of resorses just to run, to big overhead. If you have PC with lot of HDD, just run nodes on 1 OS
didn’t there use to be like a 1gb minimum ram pr node, but then again i don’t think i’ve ever seen my node use more than 200mb, had to go check what it was on… wow 86mb for my node right now…
lol well very impressive setup, i really like how low the overhead must be… not the vm overhead, but the overhead of the “storagenode controller” value and it’s electricity costs.
linux rules for vm, you can run a container and get like 97%+ performance in the container… then it becomes basically irrelevant how many “vm’s” one makes… but in windows i’m sure its spot on… also Hyper-V sucks at hardware passthrough… which sucks… just try to get USB passed to a VM
doesn’t really solve my USB passthrough issue with hyper-v
which was what drove me mad enough to just run a windows vm on a linux hypervisor, now i can pass through a usb with a few clicks instead of spending way to much on trying to make it work…
but yeah windows is good for many things, but i don’t regret going over to linux for my server.
was hell tho… lol for a while…
This looks very inefficient to run Going to be spending alot of money just on power alone, less its free of course. There’s no need to run nodes on super high end hardware you will be throwing your money away doing this. Also running 15 nodes at the same time is a very bad idea. Each node will need to go though vetting and this will take 15x as long to get though it.
