Will the speed of the network degrade once we go live?

It’s slightly worrying that SafeNet cannot handle streaming* 6 Mbps most of the time. Hoping this is a testnet limitation, although I can imagine testnet has a better infrastructure than actual farmers once it goes live. I recall @dirvine mentioning somewhere that file chunks have 4 copies on the network. But most people have plain crappy connections, or at least crappy upload speeds. You can download through Bittorrent from 20 peers and still have an average download of 20 KB/s per peer. That means downloading from 4 farmers might give you a mere 80 KB/s download (unless of course you are lucky enough to have chunks allocated to farmers with a descent connection and upload speed). This is in no way competitive with centralized servers.

I think bittorrent is a fair analogy. Torrents with 600 peers and 200 leachers go lightning fast. But torrents with 4 peers can take days.

What technology, if any, does SafeNet implement to be a sound alternative to centralized and fast servers in a world where reality tells us that crappy connections are still the rule, not the exception?

*) Here’s a streaming test setup: safe://stream.tenefas.safenet/
(It only works in the SAFE Beaker Browser, not over the proxy in Chrome, for some reason.)

2 Likes

Search for opportunistic caching and network performance concerns.

2 Likes

Don’t forget it’s 4 peers per CHUNK, so it’s possibly hundreds of peers per FILE

4 Likes

In my opinion, a very valid question! In early days most people on the net was using a home computer, yes with a crappy connection but still a stationary device. Nowadays most people are on the move with mobile phones, and when at home they are using a Pad of some sort. MaidSafe’s vision of using ordinary users extra, not used, resource as memory, cpu power and bandwidth from these sort of devices is a challenge…

Another challenge is the fact that many iPhone/ iPad users must pay Apple for extra space in the iCloud, Apple are giving some space for free but this space are soon taken just to have a backup, then with some photos and such Apple will charge you for more space. In that perspective, how much extra recourse have an average iPhone user to spend…

In EU mobile operators are also moving from public IP to “black” or “fake” IP-adresses, most mobile devices are today behind an operators NAT and firewall which are not easy to “punch” and which are in the operators control to the last “bit. IP v6 are not around the corner, as it seems…

The first trials with decentralized vaults didn’t work, and I believe that this is a very big obstacle for MaidSafe’s vision to come true. Still, I am a true supporter :slight_smile: And I will be the first to contribute with a good environment for a small but efficient vault. Not so much for earning safecoins, just for fun and for what the “safe and free” network can do for me and for others :slight_smile:

4 Likes

NAT hole punching was accomplished by the team almost a year ago with a greater than 85% success rate IIRC.

Apple device owners could opt to use SAFEnet to store their files instead and use the limited space on their devices to farm for more safecoin. Those coins could be used to pay for more storage on the network. I doubt apple will allow this though since they have strategically chosen to omit SD card based storage expansion. With apple you trade UI/UX for freedom.

Cheap non apple devices are more prevalent due to cost and convenience. These are the people who will see the advantage of using their device for farming. With enough popularity apple users will protest until they are allowed to participate. These superficial consumers who follow the apple trends are also likely to seek some return after paying exorbitant fees to obtain their shiny new apple device. I believe the economics of the mobile market coupled with the safecoin incentive will balance this out.

3 Likes

I think there is no parallel downloading currently?

1 Like

Hmm, I’m not sure.

And yes, more and more people are using mobile, but as soon as SafeCoin can let people earn real $$$ with a desktop I think it will still explode.

It’s not like there’s zero BTC miners due to mobile phone popularity :stuck_out_tongue: if we build it, and there’s money to be made, people will come!

1 Like

Yes, this is definitely true. Although the smaller the chunksize - which is good speed-wise for a single user - the more chunks for many other users will be served at the same time over the same connection.

Really? Didn’t know. Do you have a link reporting on these trials and synopsis for the failure?

I’m really curious about this. I hope network speed will be factored heavily into the reward. Not only does this prevent coin leechers from limiting farming bandwidth, it would also be an interesting experiment for myself to see if the coins are worth leaving a small computer running 24/7 for harvesting. I don’t have too much disk space but I have a very fast internet connection.

Things like SHA256 and Scrypt coins (I still have a couple of miners) quickly became obsolete 10 times over because electricity in my country is crazy expensive.

1 Like

Interesting, 100 watt Solar panels are only $100 USD on eBay

I’ve been living off of those for years now

I’d say the age of free electricity is here :slight_smile:

5 Likes

Parallel uploads are possible, but not enabled currently due to issues with progress reporting in the UI (of demo app). This should be fixed in a future release and was discussed recently on the dev forum as a way to improve upload speeds.

As for download speeds, I am not sure what the status is, but there is clearly scope for optimisation if multiplexing isn’t currently enabled.

3 Likes

Test8 had vaults, so not a failure by any means. There are RFCs to prevent slow nodes from being given key routing roles too early though (and other related RFCs). Once implemented, vaults should be more stable and the network will gain performance.

2 Likes

First of all, congratulations for living in a nice house with a roof and/or a garden you must have access to a roof. When you live in an apartment building, you might get away with sticking something on the wall, but the efficiency will greatly drop due to angle.

Second, I did look into it (for ideological reasons), but there’s a lot (of costly hardware) to it. Also, these ‘panels’ come in many cells that you will have to group together in a weather-resistant fashion yourself, costing time, knowhow, and materials/money.

Third, a 200 GHash/s mining rig consumes 150 watts, or - ignoring everything said above - a $150 investment. 200 Ghash/s is about 0.0004 bitcoin per day, or $0.25 per day, or a little under two years just to break even. Ignoring the slowly declining profit margin and the investment of the rig.

For comparison: Without free energy, we pay about $0.27 per kWh. Running 150 watts for a day is about 3.6 kWh AKA about $1 per day. So modest mining, even for ideological reasons, costs about $0.75 per day nett.

Of course farming on dedicated equipment doesn’t cost nearly as much energy, but I predict it’s a lot less profitable too. Let’s do a thought experiment. Dropbox users are willing to pay $100/year for 1TB of storage. Someone writing a MegaUpload app on top of MaidSafe needs to be competative, so they will charge $50/year per terabyte, and give $40/year/TB to the harvesters. The chunks are stored at least 4 times, so that’s $10/year/harvester/TB. A 4TB drive + NAS hardware costs $300. Earning $40/year makes you need to harvest for 7.5 years to break even.

But even if you already have that NAS sitting there catching dust. No investment needed. Just 30 watts of non-idle power consumption AKA 300.024365*0.27 = $71 per year in electricity costs. Hence, earning just $40 (AKA losing $31) per year makes dedicated harvesting not profitable either.

Oh well, I was letting myself go a bit. Sorry for that. :smile:

4 Likes

10 posts were split to a new topic: Living in a van discussion

Well, I suspect the majority of Dropbox users store well under 1TB. As it is a limit few will actually be pinned at the limit either. Who knows what the average utilisation is, but I would wager many barely scratch a few gigs.

Moreover, if you have spare local capacity, farming it out is only a bandwidth/power cost - the storage device is just sitting under utilised without using its full (80%) capacity. So, the maths here is different too and I would also wager that most people don’t use even 20% of their storage device currently.

Take both of the above into consideration and the figures start to look a little more compelling.

2 Likes

@McAddison I welcome your thinking out loud and particularly working through the numbers. I’d like to see more of that :slight_smile: It is useful because it can be shown where it is right and where it is wrong, and enables us to focus in on opportunities, and of course to eliminate blind alleys (until we have light enough to go explore them more productively).

3 Likes

Moved a few posts that were off-topic to a new thread here: Vanlife is awesome

You can change the title if you want to :slight_smile:

3 Likes

It’s evident to me that this Network will rely on data centers for years before the extremeties become dominant (including mobile)

Maidsafe stated in their pitch, that they will likely be one of the largest farmers in the early days and have also mentioned forging partnerships with data centers. So even if big players dont farm initially, the network will be secured by plenty of Maidsafe owned nodes and the end user experience will be good.

I dont see a problem with the network running predominately from data centres and having Maidsafe dominant in that play. After a maximum 10 years all going to plan, it will no longer be profitable and that infrastrucure would have played it’s part.

By launch I hope we have virtualization vendors tapping the repos and offering up CDN like swarms of Safenodes for hire. To that end maybe Maidsafe can enable Webhooks on their Github repos:

Webhooks allow you to build or set up integrations which subscribe to certain events on GitHub.com. When one of those events is triggered, we’ll send a HTTP POST payload to the webhook’s configured URL. Webhooks can be used to update an external issue tracker, trigger CI builds, update a backup mirror, or even deploy to your production server. You’re only limited by your imagination.

Each webhook can be installed on an organization or a specific repository. Once installed, they will be triggered each time one or more subscribed events occurs on that organization or repository.

You can create up to 20 webhooks for each event on each installation target (specific organization or specific repository).

7 Likes

Wasn’t this for providing an in-house safenetwork so their cloud could use the private in house safenetwork for a redundant storage system rather than what is used now. Or did I read that part wrong.

I think your correct, wasn’t that the purpose of the other consulting company they have set up.

The two statements were seperate, but I’m not sure how as a compoany you could be one of the largest farmers without employing datacenters.

Maybe or just negotiate special rates for the global safenetwork. This though would not be certain the way they are going to go.

But the idea of a cloud supplier which own one/more datacentres might very well consider a private safenetwork for their cloud storage and maidsafe earn fees from them using the software commercially and is what I considered the negotiations to be about.

Farming doesn’t necessarily need this negotiation and just be a commercial deal with special rates on the servers rented.

1 Like