• Forum has been upgraded, all links, images, etc are as they were. Please see Official Announcements for more information

How could we make DarkTor profitable for people using amazon EC2?

QuantumExplorer

Well-known member
Dash Core Group
From reactions I have seen a lot of people are afraid about high EC2 bills (as am I).

A pay per day system could work, but I propose that a bandwidth cap be installed.
For example 50c = 1 day and up to 100mb (using 8 nodes).

Why 100mb? Well because 100mb going through 8 nodes is 800mb and 800mb costs roughly 10 cents on an ec2 instance.

I was thinking just now on how we could achieve a cap.

This is what I came up with. Tell me what you guys think.

Imagine we created a second blockchain within the masternode system. I'll call it the Darkchain and the currency would be dark credits.
This Darkchain would only be for within the masternode network and its currency would neither be mineable nor could it leave the masternode network or move between masternode, except when it is sent to clients. The currency would be created whenever it is selected by the voting system (currently when it gets a darkcoin payment).

Only 2 things would appear on the global darkchain.
*Creation.
*The transfer with expiration date. Once it is transfered one time the expiration date gets set.

Usage of the darkcredit would be an appendum in each masternode's ledger.

A user would be able to purchase dark credits with darkcoins from any masternode that had darkcredits. Once sent to a client the darkcredits will be set to expire in a little over 24 hours, this information would be visible on the darkchain.

A user would then select the anonymity he wanted (8 to 20 hops). Each DTP message he sends out would contain a small amount of darkcredits depending on how many intermediate relays he wanted to use. DTP (Darksend transport protocol) would be a tunneling protocol extended from some other tunneling protocol that would contain the dark credits.

Every node would verify that the darkcredits are still usable on its local darkchain ledger and set them as spent as it relays the message. As much as it is possible that a user double spends his darkcredits doing so would result in the loss of a lot of messages and wouldn't be practical for a user.

So what do you guys think? I am open to criticism, but I would like to start a dialogue so we can come up with a great system.
 
I wouldn't focus so much on ec2 simply because these server are supposed to be used in an elastic fashion and not for continous high bandwith use. Take a server Fom kimsufi and you get unlimited traffic as far as I know or just any root server and you get at least one terabyte of traffic.
If the interest is there I can write a thorough tutorial on setting up a root server which would not much more difficult than ec2...
 
I'd suggest we don't want to encourage more people to use Amazon EC2, as there are already a large proportion of masternodes using it, and it's best to avoid an overwhelming supermajority of masternodes being potentially in the clutches of any single government, especially one with a known history of using secret subpoenas to snoop ISP user data. Hosting diversification FTW.
 
I'd suggest we don't want to encourage more people to use Amazon EC2, as there are already a large proportion of masternodes using it, and it's best to avoid an overwhelming supermajority of masternodes being potentially in the clutches of any single government, especially one with a known history of using secret subpoenas to snoop ISP user data. Hosting diversification FTW.
It has been criticised by some in the past, that MN network is too AWS EC2 centric at the moment. When our vision of a Tor-like anonymous overlay network comes true, it's time to search for ISP alternatives anyway.

--> https://trac.torproject.org/projects/tor/wiki/doc/GoodBadISPs
 
I've been looking at some alternatives and there are some good ones and some bad ones out there. I'm sure we'll all share our experiences of who is good and bad as we look at more bandwidth friendly options than EC2.

One problem we'll have is that we currently don't know what the minimum specs will be for a masternode in the future.
 
I'll just use the UK based UK2.net as an example. I used to use them for a lot of stuff years ago but I bought a month of VPS recently just to see how it compared to EC2 despite being overpriced.

8Ghz of Xeon CPU to be split into as many VMs as you like
8GB of RAM to be split how you like
80GB of SSD to be split as you like

The above (assuming you don't want to use windows) will cost you £1030 sterling per year with unlimited bandwidth. There is a 1/3 discount for annual payments over monthly.

Halve the above spec (4GHz, 8GB RAM, 50GB SSD) and you pay £491 a year with unlimited banwidth usage.

As a provider their network is fast but they are just too expensive for our applications I think. I've also load tested some VMs on their environment and I've seen some worrying performance issues so I can't recommend them.

I then started considering building my own VPS service (I'm an enterprise hardware and network specialist) to offer out to the community (possibly accepting DRK) but then realised that I'd be no better than EC2. In fact I'd be worse. I'd be one man band with a couple of blade chassis and SANs replicating between a few locations for resiliency but without the clout of Amazon's legal team. I'd be offering less geographical diversity and less protection of service.

Ideally what we need is a load of suppliers to wake up to our hopefully growing demand and offer suitably priced and specced solutions in lots of locations. That will take time. Lets face it, we're only 600+ small clients to Amazon right now. Hardly a big chunk of their customer base.

Just getting the conversation started really.
 
As pointed out above, many people don't like that there are a lot of masternodes on EC2.

One of my masternodes is hosted here https://ddosdeflect.com/whmcs/aff.php?aff=082 (yes affiliate link :) ) It's DDOS protected also.

My 1 core VPS there is $15 a month, and comes with 2000GB bandwith, next one up is 2 core for $19.99 and comes with 4000GB bandwith. Even my single core is faster than the EC2 instance.

If the whole darktor network does come about, it was be great if it was an option on the masternode, so you could opt in or out.
 
EC2 bandwidth was only one of the issues I was thinking of, maybe my original post was poorly constructed.

If we don't put a bandwidth restriction, what is to stop people from quenching the servers to prevent the system from working? Any government could do this.

Plus so many people would start torrenting over the network leading to a loss of performance, not to mention the waste of energy which would definitively not be green friendly.
 
Why 100mb? Well because 100mb going through 8 nodes is 800mb and 800mb costs roughly 10 cents on an ec2 instance.

May advice: NEVER ever offer a public service on a host where you pay for bandwidth.

Sooner or later you'll regret it.

Most probably sooner...
 
As pointed out above, many people don't like that there are a lot of masternodes on EC2.

One of my masternodes is hosted here https://ddosdeflect.com/whmcs/aff.php?aff=082 (yes affiliate link :) ) It's DDOS protected also.

My 1 core VPS there is $15 a month, and comes with 2000GB bandwith, next one up is 2 core for $19.99 and comes with 4000GB bandwith. Even my single core is faster than the EC2 instance.

If the whole darktor network does come about, it was be great if it was an option on the masternode, so you could opt in or out.
I have my MNs on Amazon EC2 and spend ~16-17$/month.
Is Chaeplin`s guide compatible with ddosdeflect OpenVZ VPS in configuration???

May advice: NEVER ever offer a public service on a host where you pay for bandwidth.
Indeed, thats why when the DarkTor will be kicked in we need providers with NO bandwidth limits.
On DDoSDeflect for 35$ you have no limits at all...
 
Last edited by a moderator:
Ok, lets say we all switch to servers where we don't pay for bandwidth.

Since the number of masternodes has a maximum at any time, so would the available throughput of the DarkTor. Does no one else think this is an issue?
 
Back
Top