Verizon FIOS; 940mbps down, 880mbps up. Northeastern U.S.
Hi all,
Iâm in Hanoi now.
I just wanted to give an update on the hardware-- I should have taken a more MAID like approach to giving a date for a spec. While itâs not the most complicated build, Iâve thus far run out of time. It was my objective to give you a gorgeous, even pornographic photo of a lil box with a big hard disk and a tiny computer and so far Iâve come short.
Iâll keep at it though. The seed of this concept has me thinking about open hardware more generally again, and how to approach that. As the recent Intel bugs show, weâre sorely in need of secure hardware. Taking a page from software development, I think itâs fair to say that the only adequately secure hardware is going to be open hardware.
The BEST possible MAID farmer would be open from the bottomâ
bottom
- CPU (see: RISC-V/LowRISC/OpenRISC as examples of open cores)
- PCB (various PCB sharing communities exist but none is big because most PCB designs are made by the same companies that make the CPUs)
- Memory Modules (Thereâs no even kinda open source RAM)
- Board Components (capacitors, resistors, ports, jacks, etcâŠ)
- Software
top (maybe not in that order)
Now, what I just laid out there is in fact impossible today. I think getting there could take ten years. So, probably weâll of course start with something less-than-perfect. My main point here is that the trustworthiness of well-reviewed and understood open software has actually eclipsed that of most any hardware out there. And since software sits on top of hardware, itâs pretty tough to protect software from hardware that doesnât perform to spec. Most hardware doesnât even have an open spec, so itâs difficult to validate weather it does or does not perform according to its spec.
The above is a bit of a rumination that Iâd like to take action on. But this wonât be the ânext monthâ kind of action. Luckily, it can start by simply releasing well-understood hardware that uses proprietary components, and documenting the source of those components.
What youâd want to do is create a new standard for hardware openness and security by honing in on the reality that closed hardware is deeply damaging to the software systems that run on top of it.
Thanks for listening. Iâd love feedback.
That all sounds great but as a v1.0 I think a pi zero attached to an 2.5" hdd in a SAFE branded box would be enough. Something I could walk into each of my family membersâ home and plug in and walk away. Keep it under $20 for the base device without the HDD costs and weâre golden. Hot swappable drives with duals slots for uninterrupted service when upgrading is a must. I could map out the entire device if youâd like.
can you imagine if gamers world wide grabbed onto this tech?? multiplication of computing power to the max on a free net that just gets faster and faster!!
Keep in mind that you really should use ECC RAM for ZFS, and have at least 16GB of it.
Thereâs a lot to learn about ZFS. It is an excellent filesystem, both from a performance and reliability standpoint, but it comes with some required learning if you want true reliability.
Hey @Stark,
While it could be a $20 device thatâs not what Iâll be shooting for this time around.
Thanks for your suggestion and offer to map it out. Anything youâd like to add to your earlier description of a device based on a Pi Zero?
I do have one question-- as a consumer, youâd like to have a device with no HDD?
Iâm headed back to China soon and still have high hopes for getting a photo and a rough spec out here.
For now, I think $20 price point isnât very approachable. When the network and market are more mature, maybe.
My line of thought is still thinking that response/latency time is the critical variable if one wants âcompetitiveâ farming (and as snappy as possible a network). A long time ago in this thread I mentioned a xeon D based / 10GbE internet NAS⊠itâs built now, and the 10GbE link from the nvme drive has lower latency over the network than an âin-boxâ platter hard drive and can saturate the link (>1Gb/s). Itâs a REALLY expensive solution, and only possible for me because the computer doubles as a distributed processor that I use for work etc. and it was paid for from bitcoin profits. But 1Tb of the main nvme and 2 Tb of the HDD raid is reserved and happily waiting for safenet. The last piece of the puzzle will be a custom router that prioritizes safenet ports, and upgrading to the 10Gb home connection that is now offered where I live.
Nimbus SSD 100TB. The future of SAFE network
http://nimbusdata.com/press/nimbus-data-launches-worlds-largest-solid-state-drive-100-terabytes-power-data-driven-innovation/
Ah the first of the expected announcements. There was some inside info that said that 40x increase in SSDs was expected this year. Expect more.
This will almost put to rest the question if storage technology is still advancing. I mentioned elsewhere that its was expected to break the 10x every 5 years and this has leaped frog us some 7 years. Iâd expect the prices for smaller bigger drives to be reasonable. And existing prices to drop.
Another possible development in storage (but a bit more in the future):
Did you see the location in the press release?
âIrvine, CA, March 19, 2018 - Nimbus Data, a pioneer in flash memory solutions, today announced the ExaDriveÂź DC100, the largest capacity (100 terabytes) solid state drive (SSD) ever produced. Featuring more than 3x the capacity of the closest competitor, the ExaDrive DC100 also draws 85% less power per terabyte (TB). These innâŠâ
DĂ©jĂ vu:
I live in Southern California and that is just up a freeway about 60 miles form me.
They have a great Movie Theater there
Just a simple question, we can already farm with the alpha version of MaidSafe? And so farmer MAID and not safecoins? If yes, whatâs the actual average ?
No, farming wonât be possible until alpha 4 or 5, and that will only be with test Safecoin.
Real Safecoin will not be until beta.
Thank you for the answer. By then, the technology will have progressed well. No need to worry about that now
It seems to me that a NAS or home media type server will be already optimally designed.
A nice simple app to run on a synology/qnap/freenas box would be perfect!
Easy install, local storage, low (ish) powerâŠ
Rup