Search the transcripts of our AMAs. We hold fortnightly sessions with senior DADI engineers and partners and take questions from the community.
I know the DADI team has worked for 4 or 5 years on the tech and spent quite a bit of their own money on R&D including a couple years on the blockchain application. Can you describe the evolution of the tech over the past five years? Why you started working on it? What was the initial problem you were looking to solve? How it led you to current state? Basically, the story behind the technology so far please.
Joseph Denne: This is a big question! My fellow founders and I started DADI to address a problem that we saw in the technology sector: that the only options available to content owners for building digital projects were either what I describe as legacy monolith systems (large 'one size fits all' content management systems), or building a custom platform from the ground up (and having to reinvent the wheel with every iteration).
DADI was conceived as a technology stack built around the concept of microservices, and available Open Source or as a SaaS proposition, hosted in an alternative type of network. The first few years of the business were focused on the architecting of the underlying services, starting with what we see as the central building block of any digital project, the API. Along the way we picked up a number of enterprise customers through word of mouth and existing relationships, which allowed us the opportunity to create a real-world feedback loop whereby we built the future direction of the product hand in hand with our key stakeholders. This has meant that the technology has benefited from being built with real traffic and real digital products. This sets us apart as most technologies are built in a closed room with the practical application too often an afterthought.
We begin building our network a little over two years ago, initially as a small group fleshing out the architecture of the network before moving to alpha development work and eventually drawing back the curtains just ahead of our crowdsale. The history and weight of work undertaken in our privately funded R&D phase is why we've been able to move so quickly since our crowdsale. Where most companies raise capital and then start development, getting to well over a year+ down the line, we've taken our network to live in five months.
I should note that our funding was put up exclusively by the founders - we received no outside funds until the crowdsale. Of course, there's a lot more to do, and we're very excited by our roadmap over the next 18 months (and I think you're going to love it too 🙂).
We're also using Golang heavily in the network, which is a very powerful and performant stack. I'd also recommend looking at Python: it's the leading language in AI and big data right now.
What will drive demand of DADI tokens over the mid to long term? Masternodes seem to be the only feature that will drive some demand on tokens.
Joseph Denne: There are three key perspective to talk about in relation to token demand:
Macro factors (big trends)
Micro factors (our strategy for scale)
Marketplace dynamic (demand and supply and fixed token)
Mid-long term, the drivers of demand for tokens will be micro (those largely within our control) and macro factors, but also from the perspective that we have supply and demand side of the DADI network as with any marketplace. Firstly, from a macro point of view there are some big trends: shift to blockchain, IoT, general cloud adoption, web services/SaaS, network attached storage and so on. These macro trends matter because collectively they represent billions of dollars of proven value today and our technology strategy is not just about a share of the future blockchain market, it's about using blockchain to acquire share of these existing markets.
For example, our web services - which is a multi-billion dollar market alone, can be deployed as dApps in a VPC - as per Jim's note above - or Public Cloud within the DADI Network. This means we can provide customers with a credible blockchain powered alternative to their current solutions. These markets can impact the value of the DADI token as there is a huge market that exists already in which we will gain market share that can be reflected in the value of DADI tokens now.
From a micro perspective we are building out the DADI network at scale through partners. Both partners as customers of our dApps, but also partners that are providing the backbone to the DADI network in terms of Nodes and network capacity. Whilst we grow the number of Hosts in the DADI network over time we are working to ensure that we establish the largest network in terms of node count in a matter of months. The micro picture is about us delivering customer sales which of course mean tokens being purchased and distributed to Nodes.
Customer sales thus also create the demand for Nodes. In terms of customer sales, we are focused on ensuring adoption of our proven technology through outreach to enterprises directly, agencies, developers and the DADI community. Also, in the mid to long term we will be bringing other third party dApps onto the DADI network, which will grow demand massively.
Short term the big influencer of value is market size and the opportunity from blockchain and the wider macro factors. It’s important to remember that whilst DLT is new and has huge potential, a significant part of the value is going to come from servicing existing markets not new markets.
We're operating in a huge space with a unique proposition to benefit from both. This really sets us apart. And of course, in the long term the finite supply of tokens will drive up value when demand grows. In this case 1 $DADI must increase in value. I think that just about covers it!
We see a lot of different people pop through TG with different takes on what is required to convert a site to DADI and I am not sure how this question can be answered in laymen's terms but how specific are the requirements to switch a site to DADI? Is it language agnostic, database agnostic, how hard is it on scale of 1 to 10?
James Lambie: On a scale of 1-10, about a 2 🙂It really depends on the site in question, but for example we migrated and refreshed empireonline.com to DADI in 6 weeks. The previous build took the customer's team nine months. DADI allows you to spin up and API in less than 30 seconds and run existing templates in DADI Web without any structural changes. A lot of effort has gone into making the platform as simple as possible to use.
Assuming you meet your projected demand for your services, what percentage of tokens will be locked away into masternodes?
James Lambie: This is very difficult to project as the POS requirements will fluctuate significantly over time to help balance supply and demand. By way of example, if POS requirements remained static, we would need around 1,250 Empire-sized customers for DADI CDN to drive a network scale that would result in the lockup of 90% of all tokens. To keep it in perspective, that example represents about 0.005% of the total current CDN market. I imagine it will be a balance of token value, service cost and liquidity. From this perspective, anything up to 99% would seem possible.
Am I correct in saying that you pay by the amount of visits or requests to the site? How would you provide protection against a weak DDOS attack that might not actually cause any issues but will register as many requests?
Joseph Denne: For CDN we charge based on the number of requests to the service, aggregated to the nearest million. So, for example a webpage with five images on it would result in five requests. The network also pays out on the basis of requests handled by a node (factoring some items in the mix as well, such as uptime and POS).
DDOS mitigation is handled via a consensus protocol at Stargate level. Stargates look at all requests and monitor them for unexpected behaviour. Where an attack is identified the requests are blocked at DNS level, protecting the network. We also have DDOS scrubbing in place through our network partners. This provides automatic traffic routing to a scrubbing centre in the event that a specific network subset is flooded.
Saw this image "The connected living room" and loved the simple illustration of the concept. My second thought was: How about bandwidth? My Raspberry Pi latency to Google is 20ms whereas 3ms from my VPS. What about home network higher latencies? Isn't this an issue?
Arthur Mingard: Latency is considered when rating a host. However, measures have been taken to reduce it in other parts of the network. One example is the routing policy mechanics, which route requests to the nearest gateway, on the same ISP where possible, lowering the number of hops and distance travelled. Your latency to Google may be slightly higher but imagine if Google’s closest point of presence was at the end of your street. With Gateway, that’s possible, and over time, quite likely.
I’m trying to understand how many requests a typical website generates and how DADI network and CDN are able to handle them. Can you help explaining that with say one of your biggest clients compared to a website like Yahoo.com?
James Lambie: It’s not one of the largest but using empireonline.com as an example: they do ~150 million requests per month, which boils down to roughly 172 requests a second. Empire is ranked as #18,147 in the world by traffic. Yahoo! is currently #6.
How do you envisage people with the household products affecting the network? Do they need to stake? If not, how does this and non-staking nodes affect the value of the network?
James Lambie: There will always be a stake requirement for earning nodes, but how this is expressed may vary. Imagine purchasing a device that has the stake built into the initial cost. Non-staking nodes already exist in the network, in the form of some backbone components - but these are non-earning. In the future other non-staking devices may be brought on board. Imagine running a DADI app that offsets the carbon footprint of your laptop, for example. Non-earning, but still valuable to the end user, and still contributing to the network.
How far in development is the DADI wallet? Can we see any update sneak previews or features? Will it be released to initial node contributors first?
Joshua Overbye: The concept of a stand-alone wallet was merged in to account pages for the management of POS, Nodes and customer side dApp usage some time ago now. Why? As we're remaining ERC-20, there's no requirement for a bespoke wallet.
Joseph Denne: To be clear, by bespoke wallet we mean dedicated wallet, in the guise of an Exodus for example.
Using a Raspberry Pi as host with 50GB hdd, how's DADI Store dapp going work as the user files have to stored on the nodes?
James Lambie: Nodes require storage space for the running of existing dApps as well as Store. They have a default storage limit which can be configured by node owners. As an example, perhaps you configure it to allow 10GB to be used by the dApps.
What impedes a node from manipulating the output of their own computations? What happens if the computation is non-deterministic? i.e: time(), rand(), cat /dev/urandom
Arthur Mingard: The network operates a peer review system at random to cover most output formats. Non-deterministic responses are an interesting one. In a typical cloud environment, it tends to be the case that webservices run with status endpoints to determine whether the application is running as expected. This is usually the only requirement, because of the nature of the owned infrastructure, which comes with an assumed level of trust in relation to the machines in operation. To create a similar environment where trust exists, and when running trusted software on an untrusted host, the network adopts a secure container environment where the VM and containers are running in a secure enclave. This means that the owner of the hardware is unable to tamper with (DMA or other infiltration methods) even though they have full access to the machines.
Will there be apps for DADI Store for cell phones (iphone, android) and computers (windows, mac) so one can use it to share normal files/photos across all devices? i.e. something similar to OneDrive or DropBox? If so, is this also planned for Q3?
Joseph Denne: Yes, there will be in time, but this won't happen until next year. And yes, anyone will be able to build with Store because the API will be available.
Follow up question: Is DADI Store depending on DADI public network nodes or you are providing your private servers for it?
Joseph Denne: That depends on the usage type. We'll release more info on this front in the coming weeks and months.
How are bursts of load against the network going to be handled? Is there extra network power available from some datacenter that can be purchased upon need?
James Lambie: We have owned capacity in the form of the network backbone. This is a series of very high capacity nodes distributed globally that do the job of ensuring that we have the capacity needed to meet customer demand in the early stages of the network's life. As we scale out nodes we expect the dependency on the backbone will drop away.
Follow up question: Let's say customer runs a super bowl or world cup ad and the site is flooded. How does network accept burst of traffic?
Joseph Denne: There are limits in place relating to pricing tiers that act to compel extremely high - edge - use cases such as that to reach out to us first, as they would to any provider. In that sort of scenario, we would work with the customer to ensure capacity in the right geo/topographical areas. We've done this sort of thing a lot in the past - handling broadcast tx load for events like the Olympics, as well as popular TV shows, which typically see millions of users hitting a property in the space of minutes.
Is DADI Store just for files or will it be a general key-value database akin to AWS BigTable? If not, do you have plans for a dApp with features like BigTable?
Joseph Denne: Store will be used as backing for DADI API, which puts one of its primary use cases in the same ballpark as BigTable (although it's not the same thing by any measure). We have plans for big data, specifically around DADI Track, which we will be releasing much more detail on in time.
Follow up question:As most companies provide their limited free edition, do you have plans for providing DADI services like that?
Joseph Denne: Yes
With regard to DADI Store, if data is split into fragments and then stored across various nodes, what happens if some of those nodes go offline?
Joseph Denne: We have what we're calling Dual+ redundancy built in to the platform, meaning in short that no one node holds a sole copy of a file.
Follow up question:Meaning all nodes, like blockchain work? It creates problems if websites data increases but nodes remaining limited. Is there any backup solution for it?
Joseph Denne: Not all nodes, no, as that would use up all capacity pretty much immediately. But the theory is very similar. Which is where the market economics of the platform come in to play. That in hand with our backbone means that this scenario should never come to pass.
I also have a rather strange question. Are you working on an application that allows P2P streaming using the DADI cloud? If yes - when will it be available for companies/startups? If no - do you need help building it, or will it never be possible?
James Lambie: Video is on the roadmap for CDN. Live streaming isn't something we've looked at yet, but it would be great! If you'd like to get involved, drop me an email at firstname.lastname@example.org!
How will liquidity will be maintained for the customer if there is not enough supply of tokens in the market?
James Lambie: We are working with leading liquidity providers to ensure that this is not a problem and have set aside significant funds specifically for this purpose.
Any update on available community discounts for hosting services via Wirehive and other managed masternode services?
James Lambie: Not yet, I'm afraid. We're working with the team at Wirehive to button this down. We're also looking at the potential of allowing Hosts within our network backbone to be rented out - what this looks like isn't yet fully defined, but as soon as it is we'll let you know.
Question 12: When will LBX be listing $DADI?
James Lambie: Very soon! 🙂
Curious about the node ROI projections. They seem decent. Down the road though, as we look at the addressable market and the projections from the white paper, the revenue can be possibly tens of dollars per DADI token per year as far as I can tell. So the host projections kind of hockey stick at some point. I guess the non-staking devices eat into this on exchange for a more robust network. Anyway, how do you guys see this network working at huge scale?
James Lambie: The architecture as it currently stands is capable of scaling to hundreds of millions of devices. It is envisaged as a single market-driven network, for mixed use software presented through a series of SaaS layers. As it's a marketplace, supply and demand must be balanced. In the first instance this is being controlled directly (through a phased approach to onboarding and the rollout of the backbone), but it will open up in time. So as the technology scales, the model scales as well: over time we expect the POS requirements to be reduced. Plus, the current network/model configuration intentionally builds in a lot of spare capacity: as we gather data and better understand production performance, this overhead will be reduced, which will in turn increase the earning potential of nodes.
Will you take down content if required by repressive governments? Wasn’t this one of the advantages of dadi that it is harder to censor?
James Lambie: DADI operates within the boundaries of the law in the jurisdictions in which it has active business operations (currently the United Kingdom). We have no plans to set up shop in countries with repressive governments, meaning that we are not bound by the law of such countries. In this respect, yes - DADI is harder to censor. We believe that operating within existing legal frameworks is essential to the success of crypto. Individuals/projects that think they can sidestep the law are naive at best. What business or enterprise is going to use technology that takes such a stance?
Maybe a strange question: Would it be possible to have a private cloud fork of DADI that big service providers could offer business customers as something like SharePoint?
James Lambie: Yes, this is a possibility, and something that we have been exploring with a large-scale potential customer since early this year.
How far have you come with the dashboards for customers and miners/hosts? What techniques are you using when building this? What is built using DADI decentralized services and what else are you using?
Joshua Overbye: We've come a long way with the dashboards for Nodes! I'm pretty excited to start to show you the work. Our account dashboards - the very first version of which we'll be releasing after the AMA today! - will act as an account overview for communicating important information at-a-glance.
The dashboards in their first iteration exist to facilitate initial users accounts and over time will evolve as we release more features and new dApps come online.
We are currently finessing the final interface elements for how users will onboard, interact and monitor the founding nodes to their account. This coincides with the production of our interface toolkit, providing a unified experience for tools and services on the DADI network.
The process used to deliver these design decisions follow customer, community and user research, as well as expert opinion. +We have a test team in house who I work with on the day to day.
Everything we build is built with DADI.
Is DADI Store still on track for release by the end of this quarter? Do you have any clients lined up to use it once it goes live?
James Lambie: Yes. Store works with other DADI dApps, meaning that it will be supporting CDN early in its lifecycle. We have customers in place for this production use case.