I would like to create a NXT infrastructure advisory committee from the people that had the most to do with creating the current NXT infrastructure. I think we have the structure now to properly manage the 1 million NXT from unclaimed funds to use for NXT servers and infrastructure. We can just make the 1 million NXT a directed donation toward NXT servers and infrastructure to be used over timeframe of 2 years (or tbd?)
I am first to admit I do not know how much more servers we need to properly handle 1000 TPS network. What type of servers, how many, etc. These are the sort of things that should be solved before April so we can make sure we are ready and have it in place, just in time. I expect I will just agree to the advisory committees recommendation, unless I see something totally wrong with the proposed budget.
James
I am the second to admit that I don't know etc etc but as a systems engineer I am very interested in getting involved in this. It is the critical foundation and bedrock upon which NXT must rest. And while the NXT node network may ultimately be decentralized, a good centralized committee exam of this topic up front and right now is very important.
I have followed the discussion on this over the past twenty pages (drinking from the firehose) and am surprised by what I've read. CfB can offer a minimum bound on bandwidth required for 1000 TPS (1 MBS) but can't say how many NXT service nodes that can handle the full blockchain are required. Huh? That's a CRITICAL number to know. We have GOT to know how many nodes to deploy to support a particular level of service. If CfB doesn't know the answer to that question, my next question is : How do we go from here to figure it out?
When I first joined this thread hundreds of pages ago, the NXT support network was just getting stood up in preparation for an expected rush of traffic expected from the initial open source release in early Jan, and there was a huge DDoS attack underway. Everybody was setting up VPSs willy-nilly. Today, a month later, we don't know how many VPSs are up, how much they cost, who is running them, and when they expire. Nobody talks about DDoS anymore so I don't know if its gone away or at a lower level or if we're now just good enough to shrug off the original attack level which is still ongoing.
Inquiring minds want to know this stuff. I think James' approach of trying to define up front what is required to support 1000 TPS and make sure it exists now and in perpetuity is spot on.
I am really interested in seeing if we can standardize on Raspberry Pis as the Official Public Node of NXT and deploy a lot of those all over the world rather than renew a bunch of more expensive VPSs that got us thru the initial late Dec crunch. I have heard different things about it - it can definitely forge blocks, it's underpowered for the task, it works better if you plug a thumb drive in the USB port to give it more memory, it needs to be replaced by a Cubie. I would like to pull all this together and come up with a definitive answer on a Raspberry config that is qualified to be a Service Node without running out of horsepower. Dave, your thoughts?
Solar powered raspberries are a great idea but not for the critical backbone. I keep coming back over and over to the various articles about putting 32-64 Raspberries in Beowulf clusters for $1500 - $2000:
http://coen.boisestate.edu/ece/files/2013/05/Creating.a.Raspberry.Pi-Based.Beowulf.Cluster_v2.pdf
http://www.sdsc.edu/Events/ipp_webinars/2014_01_15/WagnerIPPWebinar.pdf
http://raspberrypicloud.files.wordpress.com/2013/06/picloud.pdf
http://meseec.ce.rit.edu/551-projects/fall2013/2-1.pdf
http://www.southampton.ac.uk/~sjc/raspberrypi/raspberry_pi_iridis_lego_supercomputer_paper_cox_Jun2013.pdf
If we can conclusively convince ourselves that a half-dozen or so of these can secure the NXT backbone for $10000 - $15000, I think we should allocate funds to build them and get them deployed in top-secret James Bond villain underground bunkers (i.e., under BCBext's bed) as soon as possible. Or maybe schools, since Pi is an educational project - we could get a lot of mileage donating Pis to schools in exchange for them running our software...
If this is a pipe dream, we need to get a consolidated VPS plan in place instead sooner rather than later.
Whether we ultimately go with Pis or VPSs, we need to keep pushing to get infrastructure requirements out in the open so everybody understands the basics of them and we've got them covered.
In an hour or two I'll put my treasurer hat back on and get who-donated-what updated in the forms etc...
