@Seneca, there is approximately 8 million users of cryptocurrency today. Active utilizers and also enthusiasts - around 1.5 million active users of crypto currency.
Bitcoin has laid a track for super-fast adoption of the SAFE network. Iād say that there are few people connected with Bitcoin that wonāt immediately set up vaults once it is easy on various platforms, whether thereās a lot of content on the network right away or not.
Thereāll be a bit of a chicken and egg problem as the network fleshes out with content and applications, but even if secure messaging is the only thing that really functions to start, it should spark a wild fire internationally. As providers start to put popular content there, and developers get social networking, news aggregators, whistleblower data dumps, etc., etc., going . . . Wow! Wildfire, indeed.
2 Likes
Thatās a lot, now Iām even more convinced that SAFE will have a rapid adoption rate after launch.
I think thereāll be a ton of resources available at the start, Iām not in the least worried about a space shortage. Regardless, Iāll definitely commit what I currently have available (about 500GB and a great internet connection both in upload bandwidth and latency), even if itād not be profitable. Anything extra I put in I definitely want to max profits on, since Iām still burdened with a study debt that almost makes me depressed at times.
1 Like
Youāre probably already running your computer idle for a lot of the day - in the US itāll cost you no more than .75 cents usd per day to run a desktop computer with a 400 watt power supply, not to mention you havenāt an exceptional additional expense by putting up harddisk space you likely wouldnāt have used for a few more years now.
So there is virtually no risk on your part to do this, so glad to hear!
Iām not concerned at all about available storage, etc. What will take a while to ramp up is content, indexing sites, social networks, and the like. Those who put hot public content and apps up early will help determine the growth curve. But remember: it takes safecoin to store to the network and not everyone will have it or have easy access to it. Data has to be PUT before it can be GETād, which GETs are necessary to earn safecoin. Thatās the chicken and egg Iām referring to.
The only one I know of for sure whoās preparing to put a lot of good content on the network is Freedoms Phoenix, though I hope there are others. Hopefully, others of those with purchased safecoin are preparing means of providing content, as this will really help put grist in the mill.
This is solved through the issuance of the maidsafecoins.
The exchange project - SAFE X - will use PUTs to store order books. When they are stored the app will query those PUT data to carry out exchange. This will create GETs.
There is also http://blocktech.com/ a very capable firm that has mentioned by Ryan Taylor at a Bitcoin event in SF that they are certainly interested in exploring the MaidSafe platform.
Alexandria
History, unedited.
Alexandria preserves the integrity of the historical record. It
taps into collective, on-the-ground reporting by scraping Twitter as
events unfold and prevents after the fact censorship by archiving the
information a blockchain. Alexandriaās visual word cloud and timeline
slider illuminate surprising connections. Itās history written by
everyone, not just the victors.
Such a project will have an index of content to query and access, wikipedia, twitter, wikileaks, to name a few content hosts.
This also will generate many GETs within the SAFE Network.
@dirvine, does the network reward a vault that is caching and serving cached content?
1 Like
What also helps is that the SAFE algorithms should set the PUT price extremely low if the network is practically empty. Itās probably going to be dirt cheap to store large amounts of data on launch day. Another early adopter advantage! 
3 Likes
@dirvine, does the network reward a vault that is caching and serving cached content?
No, caching is not rewarded.
Heās made it quite clear that it does not. Only GETs from vaults open the possiblity to earn.
What incentive is there to be a node that caches data? Or act as a transaction manager? Is a node downranked (and thus less likely to earn coin) unless they provide all services?
Not caching will mean more request go through the node and back, so its selfish to cache as the workload is less.
Yes all other services are required or down-ranking does happen. All account data is maintained and managed per group so failure to supply that info is a downrank situation. IT si fast in this case and likely to get a node kicked out.
3 Likes
Not about farming hardware, but safe and sound and secure trustworthy hardware for maidsafe usage?
There is a crowdfunding for an open laptop at:
Iād strongly prefer amd though over intel any time. Hate to add to monopolies. Too bad.
I made a quick and dirty analysis whether itās cost effective to use SSDās over HDDās when farming. Bear with me and let me know what you think of this, since Iām not an expert on hardware.
HDD versus SSD
A decent SSD usually has a read transfer rate of about 550 MB/s. A decent HDD 150 MB/s. Since a MaidSafe chunk is 1 MB in most cases, this translates to a read transfer time of about 1.8 ms for the SSD and 6.6 ms for the HDD.
The HDD will probably have an access time of about 9 ms. The SSD, 0.1 ms. Iām going to ignore HDD fragmentation, since a dedicated farming disk will probably not suffer much from it. I assume a chunk of 1 MB will be stored unfragmented.
Since chunks will be requested randomly, I think itās proper to assume that the access time should be taken into consideration for every single chunk. The random requests will likely also make disk caching relatively ineffective, so for simplicityās sake, Iāll ignore that as well.
So, adding the access time to the transfer time and comparing the outcomes results in the HDD being about 12-15 ms slower than the SSD.
Other factors
Whether that difference is significant depends on other factors that also influence the speed of sending a 1 MB chunk to another node in the network. Iāll assume that the farmer takes all (likely) costless measures like picking the most effective OS and filesystem for dedicated farming. I think thereās also general consensus that CPUās and most other hardware matter little in this use case. As far as I know, that leaves two other significant factors: Upload bandwidth and latency.
Latency can vary wildly and, unless the nodes have a crappy connection to the ISPās for whatever reason, it mostly depends on the real-world distance between the communicating nodes (itās also influenced by the way itās routed, but these are related). If you already have a fiber or wire connection and unless youāre willing to physically move your hardware, thereās probably little you can do to decrease your latency (maybe switch ISPās?). My experience is that latencies usually vary between 20-200 ms. A decent latency is 30 ms.
Upload bandwidth mostly depends on your ISP subscriptionās bandwidth cap, which is likely influenced by the type of connection you have (cable, dsl or fibre for example). Iām lucky enough to live in an area where good subs are relatively cheap, so I have a consistent upload speed of 10 MB/s, which is decent. Since a chunk is 1 MB, it results in a 100 ms transfer time. This is a best case scenario, since the download speed of the receiving node would also need to be 10 MB/s.
Conclusion
(Gu)estimates of a ādecentā home setup:
Disk speed: 2-18 ms
Latency: 30 ms
Upload Bandwidth: 100 ms
I think itās pretty obvious that the difference between SSD farming and HDD farming in terms of speed is rather insignificant. The latency and bandwidth values are fairly optimistic in this example, if you send a chunk to a system with a crappy download speed on another continent, your internet connection will be far slower, making the HDD/SSD difference even more insignificant.
So instead of buying expensive SSDās, I think the average home farmer is far better off buying cheaper HDDās and if possible investing in a better internet connection, especially in upload bandwidth. Itāll likely be the main bottleneck.
7 Likes
I think the analysis is good, but the focus on speed and purchase cost could be misleading.
I think the first the question to consider is, what is the cheapest way of getting near or above the ānetwork averageā? We also needs to factor in power costs.
An important question is therefore, will the network average be dominated by people using already bought, already always on equipment (i.e near zero cost of participation), or by people setting up dedicated hardware, who are incurring purchase and energy costs? Or somewhere inbetween?
We canāt know the answer, and it will change over time, but I think it is impossible to come up with cost/benefit based analysis without this info. So one way is to start with the extremes, then maybe a midpoint etc, and then see how the optimum looks for each case.
Good point. SSDs do have lower power usage than HDDs and I think they also break less quickly. At the same time, it should also be considered that in the early days more SafeCoins are farmed than in later years, so it could be an option to think short term first, getting HDDās to maximize farming rate in the early days, and after a few years adding or replacing them with SSDās that are more cost efficient on the long term.
4 Likes
Yes @Seneca, thatās my thought too.
1 Like
Great analysis - thanks.
One question about your upload speed: is it 10 MB/s (megabytes per second), or mbps (megabits per second)?
If youāre in the UK with a domestic connection, I think itāll be 10 megabits, which I think means you need 800ms instead of 100ms to upload.
If thatās the case, the SSD seems even less worthwhile.
If you actually have 10 MB/s upload speed for ārelatively cheapā; Iām jealous.
1 Like
Megabytes. Yes, Iām very lucky. 
2 Likes
This could be interesting if you plan to buy hardware for farming:
1 Like
Other than this part youāre about right on the facts. I would mention that a cheap OVH server sitting outside Paris can supply 1Mb chunks to anywhere in Europe in < 10ms. My personal suspicion is that the best earning nodes will be dedicated servers, not home servers - though with a good vDSL line you can get a < 20ms 100Mbit line into a home now for a reasonable price (still more than renting a dedicated OVH server though).
Regarding the HD being 12-15 ms slower than the SSD, that actually doesnāt matter. What matters is ability to process queue depth in constant time which really matters when serving multiple 1Mb chunks which are always going to require a seek to each. A HD will see exponential performance degradation as queue depth rises, while any modern SSD can easily cope with a queue depth of 16 or even 64 without any performance loss at all.
This is why ZFS with a SSD ARC cache is so great. If you hand 128 queue depths to ZFS, heāll intelligently use single sweeps of the hard drive head to fill the ARC cache at close to maximum performance despite the queue depth. ZFS also understands when certain files are always read in full, so heāll also prefetch full 1Mb chunks all at one go. He even understands when he can use mirror drives separately or as striped in order to either halve read latency or double read throughput according to load.
Niall
4 Likes