DADI has rebranded to Edge. Find out more.

AMA recap: May 03rd, 2019

Friday’s AMA was stacked with interesting questions. With over 100 asked ahead and during the event, we couldn’t get to them all in the two hour session. Those not covered this time will be picked up in the next session.

Lead article image

🔗Question 1

🔗How many $DADI does the team still hold?

Joseph Denne: Almost all of our allocation remains in hand. More specifically: c.598k have been distributed to team members. (And of that a tiny amount sold.) The vast majority remain unclaimed - despite vesting periods having passed - and are in the originating wallet. Why? Because we’re in this for the long game… why would anyone want to sell this low?

🔗Question 2

🔗Can you clarify Meher Assel’s role at DADI? What’s his view on ‘decentralizing the #cloud to make the Internet fairer for everyone’?

Meher Assel: As a software engineer with lot of experiences working with the programming language Go, my role is mainly work with the network team implementing/optimize features. With lack of entertainment activities and proper education at my childhood town (Gafsa, South Tunisia), internet was a way for me to explore the world and go beyond the borders. Knowing how important a fair internet is to humanity, I’m all in with all my technical skills to help make this a reality.

Arthur Mingard: We’re really happy to have Meher on the team and his contributions have been crucial in the advancement of the network, as well as company-wide R&D efforts.

🔗Question 3

🔗With the self-onboarding getting ever closer, will you open it as a free-for-all or will you control the onboarding with a waiting list to control supply vs demand in the network? If it’s the latter will you prioritize people without existing nodes for fairness and to maximize decentralization and robustness of the network?

Joseph Denne: The latter is our current preferred approach. Why? Because as you identify, it enables the management of supply and demand, which in the early stages of the network are fairly sensitive. Assuming this is where we end up, there will be a gated approach to rollout, with individuals on boarded a way designed to ensure fairness.

🔗Question 4

🔗How are dynamic ip address changes handled for host devices?

Arthur Mingard: There are no requests made directly to a Host. This was a key part of the network design which allows Hosts to operate behind Firewalls. All requests are outbound, via GRPC.

🔗Question 5

🔗When can we expect CDN waitlist metrics?

Joseph Denne: You know I’m not sure. We’ve ever been asked for these. Happy to share. There are currently around 400+ names & domains in the list.

🔗Question 6

🔗How far along is patenting?

Joseph Denne: Well, we are many months in to the process but do not currently have an end date. However, we have started to release code and will continue to do so.

🔗Question 7

🔗Could a DADI engineer elaborate on this BFT standpoint from Consuls POV vs DADI’s proposition? In this video Jon Currey, Director of Research at HashiCorp, is explaining the reasoning for choosing for a crash failure model (SWIM protocol) rather than a full BFT model. Hashicorp says BFT is not widely adopted because there are no bad actors inside the datacenters, thus crash failure is good enough. Of course, this does not add up with DADI’s proposition with homes and offices as a datacenter. So, is DADI working on a different implementation to replace SWIM with a higher hierarchy solution?

Arthur Mingard: All of the Consul Datacenters are run on Stargate and not on machines in Homes or Offices. We may move to running Consul Agent on Gateways too and at that point there will be some considerations to make around how to handle bad actors and crash failures.

🔗Question 8

🔗Who is part of the ownership structure of the holding company? Have there been any new investors?

Joseph Denne: There has been no change on this front: my partners Will (Lebens), Chris (Mair) and I own the company outright. There are no other investors or partners.

🔗Question 9

🔗Are you going to include random stats for the founding nodes like upload time DL/UL speed, average ping, etc.?

Arthur Mingard: There are a bunch of detailed statistics that are being collected and will be visualized through the explorer in time. You can see a breakdown of what we expect to be highlighting here: https://dadi.cloud/en/updates/network/explore-the-edge/

🔗Question 10

🔗Can you tell us more about the security of the self-onboarding nodes? Does it only rely on Intel SGX, or is AMD and ARM also supported? Does the SGX solution rely on the (centralized) secure attestation servers of Intel, or is some other implementation like Golem’s Graphene-ng being considered?

Meher Assel: We’ve recently revisited some decisions relating to the potentially sensitive nature of network data and are exploring some alternative approaches. One key focus has been finding ways to shard and encode data so that a single Host would not host any meaningful and sensitive information. We’re still working on running containers under SGX and are working on AMD and ARM, but we’re more focused on Intel as the hardware enclaves are more mature. SGX does not rely on any centralized authority.

🔗Question 11

🔗Someone was able to crash the price to 230 sats with just a few hundred dollars on Kucoin today, although there is a market maker bot supposedly trading hundred-thousand dollars every day. 1) what is the point of this bot, if he can’t even sustain the price of a hundred-dollar market sell? How will clients buy and providers sell with this kind of liquidity? 2) Why is the team not buying back tokens at these prices and keeping at least a somewhat stable price at this low market cap? With not even 0.5% of ICO funds, they could not only keep the prices stable and make less people rage-quit but also save their own investments. Companies buy back their shares when they think they are undervalued all the time, but DADI seems to have 0 problems with the price crashing to 0. Please explain your rationale.

Joseph Denne: That’s a nice short question! We obviously care about token price – it’s an important instrument in the network, being central to network growth and performance. We have bought back tokens, specifically to aid the onboarding of customers through the provision of OTC for services. However, we are not interested in manipulating the price. I see no long-term benefit to this, and am honestly alarmed by the amount of manipulation that the cryptosphere tolerates. There is very little in the way of real-world use cases, business or traction for a very high percentage of coins. If we are to really drive adoption of these technologies, we need to get away from this sort of thing.

🔗Question 12

🔗Why are nodes on their second SD card reporting as offline with no resolution? Are there still fundamental issues that haven’t been resolved?

Arthur Mingard: A lot of testing and regular software releases have been made over the last few months, averaging around 1 major release per week. In this early phase there have been some breaking changes, and for machines that were disconnected during the release period we have instead chosen to send out updates cards rather than asking the owners to manually update. This service is something we expected to deliver and has been taking into account when calculating earnings. We expect the latest release of cards, which are currently shipping, to be the last.

🔗Question 13

🔗How many founding nodes have been sent out and how many of those are online? I currently see 132 FN in the network explorer. Feels like we are missing some.

Joseph Denne: I believe we’ve built and shipped 179 nodes so far. Not all of them have been plugged in, and the explorer is in Beta. There will be another set of nodes available btw - not sure how many yet, but so far, we’ve had no breakages in shipping, which is really quite remarkable, meaning that we still have spare inventory!

🔗Question 14

🔗How is the ‘security’ challenge of self-boarding going?

Arthur Mingard: We’ve been working on revamping the Access Control List process which is an important part of the wider security challenges ahead of SO. As a large piece of work, we’ll be writing more about this as we release the first version later on this month.

🔗Question 15

🔗Does DADI monitor uptime performance (+DNS lookups) for each individual region? I would love it if DADI provided a tool like Pingdom Tools to test from each DADI server locations.

Joseph Denne: Yes, this is done at Stargate level on a region by region basis. And I love the idea of a public test suite for performance like Pingdom provides. Will add that to our backlog 🙂

🔗Question 16

🔗Why is the team holding so many tokens in one contract as it is unsafe from a cyber security standpoint? Are there plans to split these tokens into different wallets or separate purposes?

Joseph Denne: Holding tokens in the originating contract reduces circulating supply and these tokens have not yet been claimed by the team. The contract has been third party audited for security, but there are plans to review the holding strategy as there are other methods of addressing this.

🔗Question 17

🔗How can we trust Stargates when they are decentralized? Aren’t they too susceptible to byzantine attacks?

Arthur Mingard: Stargates form a consensus with each other, so majority rules. DNS records are regularly checked under this process to ensure no foul play.

Joseph Denne: To elaborate on that a little, consensus for DNS means that any machine found returning unexpected results is removed from the network, with their stake forfeit. In the case of Stargates this is a highly considerable amount of money.

🔗Follow Up: Will you have an Invalid Activity Appeal process?

Joseph Denne: We’ve not specified this yet. But yes, one will be needed, at least in the early days. How this will work in a fully decentralized set up is a question.

🔗Question 18

🔗What level of diagnostics and insights is DADI team able to perform on specific problematic nodes and what tools/options does the team have to investigate specific node problems?

Meher Assel: Live logging at a debugging level can be performed on any node on the network. We also started to work on system that allows us to inspect what’s happening inside a node remotely (like listing running containers). This already helped us troubleshoot production problems in the past, and we are putting more efforts in these components. Telemetry data as part of the latest release of the telemetry application will be rolled out soon, and we’ll cover this in a write up.

🔗Question 19

🔗Does Dadi require different types of nodes? For example, are there disadvantages to having all hosts as raspberries?

Arthur Mingard: As new applications are pushed out to the network, we’ll be introducing performance bands for devices so yes, there will be a series of device types/classifications. We chose to launch with CDN because of its performance on an RPi. Believe it or not, we see better performance from the Pi than we do from some of the relatively expensive EC2 machines and we’re seeing great results already.

🔗Question 20

🔗How will payments to those whose nodes were offline be determined and when can they be expected?

Joseph Denne: There is a base payment to Founding Nodes who have been affected by the earlier issue with SD cards. This has already been factored, earnings added to account pages and the first month of payouts made

🔗Question 21

🔗As the dev team is spread over building dadi apps, building client projects and developing the network core itself, how confident is the team that they can compete against much larger teams, who are focusing on a single thing. Or more precisely, how confident is the team that they can deliver the ‘edge compute’ sdk on schedule in Q4 with their current team size and compete against others like iExec, Holo, Dfinity or Golem who have been focused on this single goal with much larger teams since multiple years?

Joseph Denne: A few of the projects mentioned “who have been focused on this single goal with much larger teams since multiple years” have technology and a proposition far less developed than ours. So I’d flip it around: how is it that so many projects in this space have failed to deliver working technology, product-market fit and real-world customer derived revenue?

Looking to our own set up, we’ve been in the process of shifting focus to core network services for some time, specifically to ensure that we have the right resource focused on the platform. See: https://dadi.cloud/en/knowledge/network/network-level-services/ We’re also working on a go to market for the individual web services that will directly support the network through the use of core services, but that does so with ring fenced resource that is paid for through its own uptake. We’ll be talking about this in much more detail soon.

🔗Question 22

🔗When can you release actual network metrics and specifically, who is using the network and for what?

Arthur Mingard: We’ve released a bunch of network metrics – check out the explorer: explorer.edge.network We can’t talk about individual customer engagements unless we have direct permission to do so. This is the case on a few fronts, and some examples have been given. And as previously indicated, we will be writing about a number of success stories. In terms of the what it’s being used for, that’s also a matter of public record. Right now, it’s content distribution through CDN (including hosting), storage and a low level of compute (in testing).

🔗Question 23

🔗How does DNS TTL work in DADI network?

Arthur Mingard: TTL has a standard of 30 seconds for all deployed applications, however this will become configurable, allowing for longer TTL times if necessary.

🔗Question 24:

🔗How long does it take for a deployment to the network to fully spread? Is it different for different services? CDN, API, Web, Store, Publish?

Joseph Denne: Deployments to the network are very quick thanks in part to the layered filesystem of Docker, which we use for the majority of network deployment.

Meher Assel: The network applications take about 30 seconds to spread. The service applications like CDN take a few minutes on first install, then a few seconds for updates.

Joseph Denne: +There is a balance to be struck, between speed of updating and disruption to services, so a staggered update is performed for each deployment.

🔗Question 25

🔗What’s the teams current plans for fiat to DADI payments of services, how will this be done?

Joseph Denne: Payments for services will be available in fiat (GBP, USD and Euro) via credit card using a traditional payment gateway. Once payments are made in fiat, we attribute the equivalent amount in $DADI from a holding account and push it in to the network. The holding account is kept topped up through a market buy service. Payments will also be possible directly in $DADI (of course), Bitcoin and Eth.

🔗Question 26

🔗When will the visualization and payout of node earnings be automated for us to view in more real time?

Joseph Denne: We don’t have a fixed deadline for this, but we are actively working on it. The first phase is done (automated processing), so moving it to a full autonomous will be fairly quick, but we’ll want to hand hold a few runs to ensure it’s working as expected first.

🔗Question 27

🔗When a stargate in the network is dedicating his rack to DADI, but DADI cannot use 100% of that rack, is the stargate provider allowed to sell this idle capacity elsewhere?

Arthur Mingard: Yes. Our Node software doesn’t mandate capacity, but the network expects that minimum availability is provided.

🔗Question 28

🔗Has there been progress on video streaming for CDN and when can we expect it to be ready?

Meher Assel: Right now, we don’t have any optimized delivery method for videos as we do have for images, but CDN does support delivering any kind of assets, which include video assets.

Joseph Denne: So, yes to video support (right now), but no to live streaming.

🔗Question 29

🔗When we see Mac OS and Windows desktop app?

Arthur Mingard: Once self-onboarding for Linux is complete, rolling out to other platforms will be pretty fast as the mechanisms are broadly the same OS to OS. We already have Host working on OSX for example and have a basic POC for iOS as well. The thing standing in the way right now is security testing of the primary build. We’re working hard on this front and soon as we have an update to give, you’ll be the first to know.

🔗Question 30

🔗Can the Dadi team give their honest opinion of Brexit?

Joseph Denne: Seriously?! I can’t speak for the entire team on this issue, but here’s my take: I don’t believe in borders, and like a significant number of people in the UK, I’m pro migration. This is worth a read: https://www.theguardian.com/world/2019/may/02/britons-more-sold-on-immigration-benefits-than-other-europeans

So, I guess I’m broadly pro EU - accepting that this is a different concept to being pro-European, which I also am - and I think that we should be a positive and leading member of the European community. However, democracy trumps my personal outlook and I find the current situation to be pretty anti-democratic. No matter how you look at it.

What a happy note to end on!

Related articles

More knowledge articles
Mail icon

Like what you are reading? Want to earn tokens by becoming an Edge Node? Save money on your hosting services? Build amazing digital products on top of Edge services? Join our mailing list.

To hear about our news, events, products and services, subscribe now. You can also indicate which services you are interested in, which we use for research and to inform the content that we send.

* You can unsubscribe at any time by emailing us at data@edge.network or by clicking on the unsubscribe link which can be found in our emails to you. Read our Privacy Policy.