A Highly Decentralized Cloud Built with OpenStack and Bitcoin

[kordless]

Hello.

I've been working on the idea of a highly decentralized cloud on and off now for about 15 years.  My efforts started with the creation of the Grub Distributed Web Crawler back in 1999: http://en.wikipedia.org/wiki/Grub_(search_engine).  At the time we were doing Grub, I saw an opportunity where excess compute could be utilized via alternate infrastructure methodologies.

About 2 years ago I left a startup I founded and began working on OpenStack.  OpenStack is a collection of Open Source Python projects that provide cloud orchestration to one or more computers running Linux.  These projects provide a variety of services, including storage, network, and compute.  OpenStack supports QEMU/KVM and several other Open Source hypervisors.  I wrote a set of install scripts for OpenStack that are now highly ranked on Google.  Just search for 'OpenStack Install' to find me.

I've known about Bitcoin since the very early days (insert the tired old story of mining on a VM and losing the image) and I would have probably been more into it if it weren't for running a VC backed startup in SF from 2009-2012.  After having going through a calming period in 2012, I got back into the scene in late 2013 and began consuming content as fast as I could to come up to speed on where we were with the technologies.  It was a cold rainy night in late November of 2013 in the East Bay that I realized Bitcoin could solve some of the major problems with a cloud federation, and that my idea of a coop cloud just might be possible using Bitcoin.  I began coding again, after a 3 year hiatus from the craft.

Today, I have a virtual appliance built with the Flask framework that runs on top of an OpenStack cluster, which takes control of the cluster and sells instances by assigning them Bitcoin addresses. Consider the appliance a compute miner.  I'm wrapping up the pool controller in the next 30 days and will launch the service beta on 9/1/14. At that time, an individual will be able to purchase cloud instances anywhere in the world someone runs an appliance, all without having an account, or providing identity credentials, other than an injected public SSL key or, later, using BitAuth.  I can tell you that I already have providers banging on my door for this.

It's a bit like Airbnb for excess compute, all reconciled and (eventually) authenticated using Bitcoin type technologies.  I believe it is the beginning of the first globally federated cloud.

If you are a bit on the savvy side, yes this is a *little* like Maidsafe or Ethereum, but I'm operating a level below those technologies.  Much of what will be provided will be based on an untrusted to untrusted user relationships and the result is a good amount of untrusted compute. However, these relationships can have higher trust levels established (if the use case requires it) by allowing an appliance to join a private group so only you and the others in the group can share and consume resources within the group.  Future work will establish cloud-karma based on something like BitAuth. i.e. if you have never used the system before, you can only launch a handful of instances with a given address, and the system has a governor on it that limits anonymous, karmaless starts across all appliances. If you escrow BTC in a contract, maybe you get a karma boost to allow additional launches.

There's a rough whitepaper here: https://github.com/StackMonkey/utter-pool/blob/master/whitepaper.md.  There's a video of me talking about federated compute issues here: https://www.youtube.com/watch?v=Hvn0RYS93nc.  The software is 100% Open Source: https://github.com/stackmonkey.  There's a fun little Twitter bot I did last week for a marketing push on HackerNews: https://www.stackmonkey.com/blog/for-bitcoin-from-a-twitter-marketing-bot/.  There's also a video of the appliance running: http://vimeo.com/100944207

I'm a solo founder ATM.  If this interests you enough to get involved, I need help with 3 things: 1. beta test the appliance (I will help you set up OpenStack and the Appliance personally, if needed), 2. offer technical suggestions of blockchain technologies I can press into service to replace the centralized bits (Coinbase &  Ngrok for example), 3. consider getting involved by helping code, auditing code, opening tickets, or reaching out to capital sources interested in the future of infrastructure.  If you are an investor, I can provide additional clarity to the fundraise.  Contact me.

I'm fortunate to be working on this, and I'm grateful this forum exists for me to be able to share something that I passionately believe in.  Thanks for reading.

K

[taylortyler]

How will this differ from MaidSafe?

[kordless]

How will this differ from MaidSafe?

How does a hypervisor differ from bare metal?  The short answer is that they operate at different layers in the stack.  The current cloud stack is comprised of compute, storage and network offerings.  With blockchain technologies available, we now have a fourth offering (what I call the 4th knob of the cloud) called TRUST.  You take a bit of compute and the result is computed trust.

Maidsafe's claim is that they will deliver TRUSTED compute, storage and network in a platform that allows developers to write new applications.  When I say new applications, I mean new *types* of decentralized applications.  Maidsafe won't run Wordpress, or Ghost, or SMF (what powers this forum).  New software that provides feature parity to these applications might be written in the future to run on top of Maidsafe, Ethereum, etc., but for the immediate future we still need something that smells like a traditional hosting offering.

What I'm building delivers instances comparable to what you would get from Amazon.  That means you can deploy a Linux flavor of your choice, provision the server as you see fit, and then access it like you do any other server on the network.  This server could run the Maidsafe binary, Ethereum, mining programs, continuous integration tests, your blog, a Minecraft world, or just about anything else you could dream up.

I'll point out the trust levels Amazon delivers to their customers are implicit.  We have ZERO visibility into Amazon's stack, yet it runs nearly a large % of the Internet's SaaS services.

Make no mistake, the instances provided at the base layer by StackMonkey are highly untrusted.  You'll have strangers starting instances on other stranger's servers.  This provides the opportunity for a high degree of anonymization in launching instances, but it also opens up the possibility for building a karma system that provides a way for other users and providers to establish lines of trust to each other through a variety of blockchain technologies run on top of the instances.

Infrastructure is meant to be transparent, trustworthy and secure.  The only way we can achieve these goals is by starting with a base set of features and then applying computed trust to them as you move up the stack to your application.

In summary StackMonkey (and the Utter.io code) != Maidsafe, but it still smells like compute with a different type/level of trust applied to it.

[goregrind]

Why is coinbase used for payments instead of a light wallet running on the va instance (removing the need for the callback) ?

[kordless]

Why is coinbase used for payments instead of a light wallet running on the va instance (removing the need for the callback) ?

It's Open Source and you are more than welcome to code that up; I am but one person. Coinbase integration was a snap for me, didn't require any dependencies that must be periodically updated, and allows the appliance operator to focus on other matters, like running the OpenStack cluster.

Now that you've elected yourself to this work by asking a leading question, we'll need to consider how the user secures and backs up private keys to their private wallet.  Happy to discuss if you are game.  The forum won't allow me to post links to gitter.  It's /StackMonkey/utter-va on gitter dot im

Oh, and I cover addressing this eventuallity in the white paper: https://github.com/StackMonkey/utter-pool/blob/master/whitepaper.md#operation

[alexandrz]

Do you plan to implement something like s3 using openstack storage?