Looking at the overwhelming interest for Chia over the past months, we see two opposing methodologies for coping with the growing netspace. One option is to make a mad dash for plotting power, attempting to seed your plots as fast as possible by buying new high-end gear to support your habit. This is the option I see pursued most readily, or at least most publicly, and while some have the means and ability to do that, it just isn’t realistic for most individuals.
That’s why I’m encouraging you to look into the other philosophy: repurposing found hardware. Many perfectly-working enterprise systems get recycled for various reasons, and for years, this enterprise hardware has been a bane to system administrators, who are deluged with extra stuff that they’ve been eager to get off their hands. Now, through the farming of Chia, we have an opportunity to bring it new life.
Two weeks ago, as the netspace growth went vertical, I faced this same dilemma, planning out how to grow a sustainable farm that could last for years, while keeping up with netspace growth. Before pulling the trigger on a couple NUCs to increase my plot speeds, some fast calculations made me realize that, although it was still a solid option, there may be more cost-effective ways to get what I needed. I started hunting around eBay and Craigslist, and calling around to local electronics listings on Google Maps, seeing who might have what I’m looking for.
I’ve purchased two systems since then. One’s a 16-bay JBOD server kitted with a Supermicro X8 board, the other an 8-bay Dell tower server, both with twin older-gen Xeons and LSI SAS controllers. I paid under $300 for the lot, and am reselling the 16 drives that came with the JBOD for a couple hundred bucks, plus one of the LSI RAID cards, bringing this farm expansion to an almost zero cost, besides the high-capacity drives that will inevitably fill them.
Not only that, but buying used helped clear out some unneeded equipment from people who got some cash for their trash and will be happy to see it go to good use elsewhere. Both of these systems will serve as hybrid harvester/plotters. They have enough computing power to pump out an extra few plots per day, or slowly replot my solo space to pool space if I so choose when the time comes.
All too often, I am watching people who are excited about Chia running into stores to throw together a Threadripper with 128 GB of RAM. Before you go do that, just take a second and think about the inevitable questions that come later: how do I get drives, how do I store them, how do I power them for the long haul? Once you’re out of drives to plot to, and have no expendable cash to buy more drives, then all the plotting power you have will sit idle. It’s called farming for a reason: your crops aren’t ready to harvest in a week. As you start thinking about your long-term farming, take a look at all the options before diving in.
There’s some early writing from Bram Cohen, the veritable visionary behind Chia, in which he describes his process as being a “practitioner of evolutionary design [and] sworn enemy of waterfall methodology.” We see this practice embodied in the mechanics of Chia’s process, and this philosophy brings to mind a similar dichotomy seen in world-building between the personalities of gardener and the architect, which the author George R. R. Martin (A Song of Ice and Fire series, on which Game of Thrones is based) often refers to. To quote Martin:
Well, I have always said that I believe there are two types of writers, which I call the architect and the gardener. The architect plans everything in advance, he draws a blueprint, he knows where the plumbing is gonna run, he knows how many rooms they’re gonna be and exactly what the square footage is. Everything is finalized before you dig the hole in the ground or drive the first nail. The gardener, he may know the general shape of the garden that he wants, but he’s digging a hole in the ground, he’s planting a seed, and you know, he has some idea what’s gonna come up, he knows whether he’s planted an oak tree or a radish, so it’s not totally random but…there’s a lot to chance and other elements.
George R.R. Martin speaking to the Museum of Pop Culture (source: youtube.com)
The garden is a living thing, and I think it’s the same for writers. I think all writers are a mixture of both, but some lean much more heavily to one side then to the other…People who are inclined more to the architect’s side sometimes don’t really understand how the gardener type of writer works and they have this misapprehension that you don’t know anything about what you’re doing or where you’re going, which is not true. You do know where you’re going, at least in broad strokes. There’s a lot of things you don’t know – you discover those things on a journey, and that’s what makes the journey interesting. That’s what makes driving across the country much more interesting than just getting on an airplane: one is quicker, but the other to my mind is more an adventure.
It’s been a journey of botched bootloaders, flashing the firmware of RAID cards, and other unexpected fun hassles, but it’s also been rewarding to figure out these puzzles and discover how to make my farm grow. This is definitely the road less traveled, a road you should avoid if you aren’t technically-minded, but if you’re someone who enjoys the challenge, consider giving it a shot. Even if you’re like me and have never really engaged with high-end enterprise gear like this before, and thus experienced a boatload of growing pains, you may take away a valuable bounty of new skills and low-level computer understanding too.
Of course, none of this is to deter you from looking to new farming hardware, especially if it’s low-powered. If you want to throw a Pi together with a bunch of external USB drives for convenience and noise, you will probably have a far easier experience than I have so far. If you don’t have a solid computer to get started plotting on, and you want to put one together, you should! (I did.) But if you really want to scale up your farm, you’re probably going to want to start looking into used enterprise gear at some point.
For some, there is an intrinsic joy in hodgepodging together a disparate set of systems and parts together into something wholly new: if that’s you, you’ll find that in droves with Chia. Besides, the time to buy multiple new beast-level enterprise systems just for Chia plotting is long passed. Don’t overprovision for plotting power: when it comes to your farm, be a gardener, not an architect.
Is it a feasible, cheap and efficient plan to use the NUC build as a plotter and a Pi with external drives to farm? I’m still researching the right beginner/novice setup. And tbh, I’m kinda lost when it comes to new hardware, I haven’t put a pc together in over a decade.
This is completely of topic but I have to say, as a real architect (one that designs buildings), that Tolkien has absolutely no idea how an architect actually works (and so don’t you). Also, the way in which he uses this metaphor implies that he also does not know what an architect actually does. The blueprint is the outcome of the architect’s work and so is the writer manuscript. The architect does not build the buildings and the writer does not print the books. If the work of an architect was to simply draw a blueprint then architects would have long been replaced by some algorithm/machine that could do it much more reliably and faster.
Well, enterprise gear gets recycled for a reason. Running old equipment that draws too much power to be efficient in a datacenter is anything but ecological.