Ah Ok we took @dirvine’s statement of 1Mb/s as it is normally taken – small ‘b’ is bits and capital ‘B’ is bytes
Oh and having done communications development for 40 odd years, link speeds are typically in bits or baud per second. My cable is 100Mbits/sec but approx 16Mbaud (QAM64) or 12.5Mbaud(QAM256). So anything other than bits/second should be spelt out
Please tell us your shooting for sub 1 Mbit/s at launch.
We certainly hope so, the network will be setting the figures it requires at that stage. Right now we are just aggressively erring on the side of more than enough in the spot check. When the network is continuously monitoring then all that it wll require is the minimum a node needs to be helpful.
Node age / Data Chains means we will be able to have nodes of varying capabilities both in terms of cpu.b/w and storage space. Initially though when we (humans) set the figures we will make them overly aggressive. This way we can “prove” the system and then allow it to reduce the requirements to fit (i.e. we get rid of more magic numbers and let the network dynamically tell if a node is helpful or harmful).
IMHO from a network perspective it makes perfectly sense to aim for higher upload speeds as vaults (among other things) will take the role of current web servers. Imagine the current web where servers would have only 1 Mbit upload. You defenitely wouldn’t enjoy that. 
Currently average web servers have 100 Mbit to Gbit uplink speeds so a 6 Mbit limit is already very low I think. Of course in contrary it’s a P2P network where the download overall speed is the sum of all uplinks but you would still have to wait for certain requests.
In order to address this issue and also allow lower speed participants it would be an idea to add some kind of vault categories and schedule different tasks for various vault categories (e.g. time critical requests vs. not so time critical ones) Maybe even add this kind of preference as a request attribute where the user or browser could choose if it’s a time critical request. Possible use-case scenario “I have time and can wait a little bit longer to download a larger file, bundle all sources you can find”. Just a thought.
This is the key part. Many of us (me included) see data republish and upgrade handling as a key to beta. To achieve that we have the RFC’s in progress and are already working on these things in parallel, starting with formalising group consensus algorithms and message flows. Then data republish, which may be able to be short circuited, but may require data chains to be in place. This is all being debated hotly right now with pseudo code and proposals. There are many ways to achieve this and we are currently looking for the “SMART” way that can get us there safest and quickest, without any large multi month periods like we just seen with disjoint sections.
So measuring several times before we cut, is the current process
For sure… higher upload speeds and more vaults for resilience. If we don’t achieve sub 1 Mb/s by launch, I’d say there’s a big opportunity in Australia for setting up farms at data centres and ISP’s…because the vast majority of DSL connections via ISP default settings are below that.
…and it needs to be absolutely fabulous at launch or I’m going to spit the dummy 
Makes perfect sense in that we want zillions of vaults of all capacities.
Thanks David, very much appreciate your response!
The spot check describes where the network will check the quality of the resource (bandwidth and CPU) as a user attempts to set up a home run/external vault. It is anticipated that this will provide improved network performance and stability from previous test iterations that enabled user run vaults. Testing will confirm if this is the case of course.
So if the minimum requirement for bandwidth to join the network will be 6mbs, do you guys know what will be the minimum cpu wise to join the network for the upcoming test 12?
This will be tiny, we have set that difficulty to zero for this test.
Still not sure if my widdle craptop will make the cut 
lol
If a CHIP computer can run a vault with under 5-10% CPU utilization then any computer built this decade (maybe this century) will most likely be sufficient. (It might need to be 64bit initially for Intel style cpu)
still might be out of luck
lol I’ll stop
Unfortunately my mobile broadband seems to be <2Mb/s upload even with 4G. With a 12GB test file it rose to 3.8Mb/s which is still way too low. 
If my Odroids were viable I could leave them on a cable connection, but ATM it looks like I’ll be absent from the early tests.
brill, i have a computer built within the last 17 years and even better have 63mbs down and 18mbs upload!
I have at least 3 half-decent computers atm, but sadly my connection is rubbish. I can barely get 0.5 up annoyingly (I thought it was 10/2 ).
/Jabba adds ‘phone BT’ to his to-do list for tomorrow.
No wonder there were retention issues previously, if 6Mbits is erring on the side of caution simply reading this thread suggests a great many vaults were dragging the network down.
Does this matter? I think not… if it all works as planned the network will pay you to provide the resources it requires so an upgraded connection should not affect your wallet. Very few folks had closets full of graphic cards before mining for BTC started, fiber companies are going to love SAFE!
If you run a vault for an extended amount of time and it builds up a good reputation does it start from scratch if you restart your machine?
You loose half your age.
Yeah here too … 29Mbps Down / 4.21 Mbps UP on a landline. So no vault for me (not yet ) … All by all I think it’s the best solution to keep the requirements higher because I don’t think it’s affordable having a slow network.
We’re not used running on speeds that are passed and it’s irritating then.
Some p2p projects have great speeds (Try Zeronet) but yeah they’re lightweights against the routing encryption etc maidsafe has and this has to get compensated by having fast vaults imo.
Only thing I’m concerned about is: “Are they going to find enough resources to host fast vaults?”.
You can read the last “Q2 2016 State of the Internet - Connectivity Report” from Akamai. Unfortunately there is no information about upload speeds so we need to deduce, from this data, the percentage of users who have enough speed.
I try to find the percentage of users, by country, with symmetric connections but I have not found anything so I think that the information about “Average peak connection speed” is our best clue.
For example, in Europe…
My calculation, very very gross, is that about 10-15% of European, North American and some Asia-Pacific countries (Korea, Singapore, Japan, New Zealand…) users could run a vault with the 1MB upload limit. Other countries will vary widely. From percentages close to the most advanced countries to others where the number of possible users will be very low.
The good news is that the speed increases at a good pace and that the substitution of the copper lines by optical fiber can increase, besides the speed, the number of symmetrical lines.
