Mining hosting idea

[PulsedMedia]

An idea - Mining hosting

An idea to offer mining hosting, your feedback appreciated

First a little bit about us(??)
Pulsed Media is a smallish specialized niche hosting company. There is 2 of us working daily on tasks, and we got quite an nice market share for our niche. We do well by our automation and methods to increase stability, performance and usability above all others

I've been familiar with bitcoin for over a year, but have not been doing anything related to bitcoin, letting it to mature a little bit more Recently, i've setup my first ~1.4Ghash of power, and waiting for GPUs to arrive for another 1.7Ghash. This is just for testing before starting to roll out a cluster.

Plans for a cluster?
The plans immediately shifter towards building a cluster, autonomously managed etc. This requires a little bit of sofware development, but nothing we couldn't handle to build quite fast if we put our minds into it and put some time & effort to it. Just something like 50-80hrs of efficient work for internal system built fully

Hashing contracts ... Cluster ... This gave me an idea
I saw the hashing contracts and their outrageously high pricing, i mean, more than the mining income is! :O Like WTF? (No offense meant, but really ... Charging HW + Electricity + 100% Margin? and you get to keep the HW!)
Ok, you are free to ask for whatever you want, and if you still get business - Good for you!

They are also quite opaque offers, little background info etc. force for 3 month contracts etc.
Why not offer properly?

We got space - We got contacts - We got (some) budget - and most of all, we are established
We are an already established business so there is plenty of BG info about us, we are stable and always looking to expand. So there is certain safety about what we got to offer, and we got established routines for managing a mass of customers in terms of support, and administrative tasks.

We got plenty of space and power!
We got roughly 150m2 space which is 4.4m free height, and mostly 5m high, also we got plenty of power. Currently we got installed capacity of 14kW, but being in a industrial building means we can install more with a week of notice, and the upper limit is quite high (being a big industrial building), probably 100kW+. Despite being off-net location we can get there atleast 100Mbps connectivity as well, if not faster, if required.

This space is currently used just for building custom computer chassis', warehousing and other kind of DIY projects. It's rented as cold, so during winters we anyways need some heating there! It gets really cold here during winters, peaking out at around -30C, and regularly over -20C. Summers tho might hit 40C (on direct sunlight), but this space is slightly cooler than outside during summers Heat will be somewhat of an issue during summers, ambient hitting 35C or so, but downclocking HW should handle that without installing any kind of AC

We already got several outlets for venting air to outside in the wall too, one of them being "damn big". and ceiling being high means there is always room for the heated air to rise and transfer through the ceiling.

Contacts
We got contacts to local refurb. hardware dealers, meaning we can reseller rates(=REAL cheap) on older hardware with 2x or more PCI-E connectors, fans, memory, cpus (ie. Celerons with heatsink just 5€ inc. VAT!). So some of the hardware can be built on REAL CHEAP.

We also get access to wholesalers so we get under market pricing. As a registered business VAT will be returned to us in a few months (23%). And this being Finland -> there is quite a lot of competition in computer hardware pricing, and relative to population there is a lot of computer stores :O

Some budget
We got also a steady budget to invest into new hardware, which does not affect the day to day operations. The same budget would just go somewhere else instead (ie. marketing, or 3rd party development). We could easily add several Ghash/s each month without selling any BTC, or without any revenue from hosting.

In practice
First of we need to add to the available space some shelves etc. nothign major, some exhaust fans, cabling and acquire a steady reliable internet connection there (not some 3G shaky one). 3G could work in the interim, maybe with mining pool proxy (haven't looked into it yet).

Then we can immediately shelve around 12kW-13kW worth of hardware (leaving a bit of room for peaks ie. rebooting etc.).

Add some remote management HW (VPN Gateway, remote reboot etc.).

And build up our software and monitoring. Build an nice interface for end users. This being the major work, maybe not in terms of hours, but how much thought needs to be given to it.

How would it work for end users, AKA Customers
When you buy up, you would get an interface where you configure your pools you want to use. It would show the number of cards, and give you ways to distribute among pools, or add weight to pools and it automaticly distributes or uses just as a backup some of them.
You would also see graphs of actual Mhash/s you are getting, and can upgrade/downgrade anytime.

This could potentially work on a Mhash based billing as well, allowing us to choose the most efficient HW possible and billed hourly. Billing hourly would give the user a "upper limit" selection for Mhash/s, and actual Mhash rate would be charged, no need for separate SLA. If monthly rate, then some kind of SLA needs to be provided.

If we cannot flexibly control the rate of Mhash/s (ie. stable multiple instances per GPU and using Aggression to determine rought Mhash/s rate. Something i've not even tested), customers would choose something like Low-end GPU, Mid-Range GPU, High-End GPU and number of them. Definitions being something like under 200Mhash/s, 200 to 350Mhash/s, above 350Mhash/s. However that would complicate matters, and more fine-grained control is best idea to develop.

Pricing
Due to heavy investment required (GPUs ain't that cheap afterall), pricing would consist of a setup fee and monthly(or hourly) rate.

An example pricing mostly out of my head:
0.25€ per Mhash/s setup fee and then 0.5€ per Mhash/s per month. So for 200Mhash/s the 1st month price would be 150€, 2nd month 100€.
1Ghash/s would be 250€ setup, 500€ monthly. (afterall with 3x6950s consuming over 600W alone).

Option 2 could be higher setup fee lower monthly:
0.50€ per Mhash/s and then 0.3€ per Mhash/s per month.
1Ghash/s would then be: 500€ setup, 300€ monthly.

On monthly options: Stop the rental, and you have to pay setup all over again if you want to continue.

On hourly basis we could charge just one time setup fee, which ups your upper limit for Mhash/s.
Example:
0.35€ Per Mhash/s setup fee. Minimum 50Mhash/s increments.
0.0009€ Per Mhash/s per hour. Minimum 50Mhash/s.
1Ghash/s per hour would be: 350€ setup fee, 0.9€ per hour.

Accounts would have to preload credit, at minimum 50€ worth per time.

All payments in BTC or EUR.

Mass discounts
We could give mass discounts for those who get high numbers.
Say, if you pay setup for more than 2Ghash/s, for all above that we could lower the per Mhash setup rate by 0.05€ or something like that

Scaling
Hourly rate might be easier to handle scaling in business sense, if charging is based on the actual received. (Say averaging per 5mins and so forth).
So if we get tapped, we can allow people to get setup for their chosen amount, and then we go get more HW. Ofc, if we are completely out and cannot roll extra capacity fast enough we would need to halt new setups.

Why would we do this - looks like charity to me?
Setup fees. Simple as that. Allows us to add more HW on the line faster, and we get to keep ownership of the HW for resale if they become useless and for other customers.

Sure, you will earn more than you are paying us, but we also get multiple streams of revenue, and can tap into non-techy markets by making it simple enough! People who have no clue, or cannot setup their own mining rigs.


Sooo, what do you think?

[1715334912]

1715334912

[1715334912]

1715334912

[1715334912]

1715334912

[1715334912]

1715334912

[xunker]

I like the concept, but I think the pricing model is problematic because you're basically trading money for money, essentially buying hashes.  Where that could have problems is that as the BTC exchange rate goes down people will be less likely to pay because they can see, in black and white, the cost basis and get turned off.

Another model, one I would like to try if I had the capital (I don't) is a leasing model: lower entry fee but a larger cost in the long term.

Example:  A basic machine with 2 6990's would be about $1,800USD, more than most people are wanting to pay up front.  You could offer to lease that machine for $200/month, but require a 12-month commitment (contract), so in the end you earn $600 extra gross.  Those costs are just examples, you would change them to suit your market.

It's not even a new model, really, server hosting companies have been doing this for over a decade.  It won't appeal to everyone, especially those who like playing with hardware, but people who believe BTC is a long-term currency but don't have the space/time/skill to build or maintain miners would jump at the chance.

[PulsedMedia]

I like the concept, but I think the pricing model is problematic because you're basically trading money for money, essentially buying hashes.  Where that could have problems is that as the BTC exchange rate goes down people will be less likely to pay because they can see, in black and white, the cost basis and get turned off.

Another model, one I would like to try if I had the capital (I don't) is a leasing model: lower entry fee but a larger cost in the long term.

Example:  A basic machine with 2 6990's would be about $1,800USD, more than most people are wanting to pay up front.  You could offer to lease that machine for $200/month, but require a 12-month commitment (contract), so in the end you earn $600 extra gross.  Those costs are just examples, you would change them to suit your market.

It's not even a new model, really, server hosting companies have been doing this for over a decade.  It won't appeal to everyone, especially those who like playing with hardware, but people who believe BTC is a long-term currency but don't have the space/time/skill to build or maintain miners would jump at the chance.

Clearly it is money for money, but so is any hosting model

Customer gets in this:
 - access to any desired amount of hashing power without figuring anything out
 - No need to fully purchase the hardware upfront
 - no need to manage hardware, software etc. just "push a button"
 - no need to manage electricity capacity or scaling issues

We get:
 - Money upfront towards the Hardware
 - Steady monthly profit
 - We get to own the HW (ie. sell it if it becomes useless)

So the business model will be staying ahead of the direct HW owned profit vs. our pricing, to a degree.

With monthly our benefits are emphasized. This would allow us to grow our cluster waaaay faster than otherwise, as we get money upfront, not after the fact.

While at first glance it might look like it does not make sense for us to offer this, ie. we'd be doing an charity. That's actually a PRO for the business.
Truth is that if we spend like 10k € on hardware day 0. We have to wait atleast till day 11 till we can add capacity by mining ourselves (whatever the return happens to be). Selling that same capacity to end customers at day 0 for 1/3rd of the cost + mohtly rate, allows us to add ~3300€ worth of hardware online, which we can use for mining, or yet again sell, and cycle for 1100€ more capacity.

That would bring us at day 30 capacity of at least 14400€ in terms of investment.
At which time we get again monthly payments.

Assume monthly payments total to 1/5th of HW price after costs, it would mean we have gained 33% + 11% + 2.2% + 20% + 20% on the hardware at end of month 1, totaling 86.2%.
Assume difficulty raises so much that HW ROI drops to 25% monthly in the next couple of weeks, this is significantly higher rate of growth.

Assume atleast 10% turnover monthly too, ie. at day 1 of month 2, we can again sell 1k € worth of investment to new customers for new setup fees.

At which point HW has devalued itself only by 20-25% worth.

New HW gets released, we instantly invest into it, and start recycling our HW, and after enough has been swapped, lower prices accordingly, meantime profitting from old rates with new HW (increased profit margin).

The model you describe would require customer to pay HW + Operating costs for 1 year upfront, with no way of backing out, meaning it's actually worse than just doing it yourself, way more expensive, way less flexible. You'd pay full HW price + full operating cost, but end up owning nothing but the gained BTC, meaning you loose immediately 100%+our margin on the HW investment. Makes absolutely no sense.

This should be win-win-win situation, customer wins (lower initial investment, lower risk), we win (risk is offset by customers, operational costs covered if BTC goes haywire), bitcoin network wins (commercial mining service!).

The example pricing on my initial post should be actually quite close to the actual.

[vivithemage]

If you can make it work, go for it. Do you have some rough numbers for setup fees? daily/hourly fees?

[PulsedMedia]

The examples were kinda close guesstimates. Not actual maths, i need to do couple days of maths to verify everything at first.
Also need to do some kind of estimate on buyer amounts, scaling requirements etc.

But i guess we will have at least 12Ghash by launch date for offer. That would be quickly scaled to 20Ghash during the first month, if not faster. That is already quite an nice capacity! Using new 5830s that would be about 41 cards online.

[airdata]

At which point HW has devalued itself only by 20-25% worth.

Not very long.

I have been thinking that it's fair to consider a 40-50% resale value on all mining hardware purchased, more in cases of super high end things like 6990s.

Some pool-mates and I were talking today about the idea of colocating mining equipment.  Considering things like electricity and cooling, it's a great idea.

[DaMayan]

At which point HW has devalued itself only by 20-25% worth.

Not very long.

I have been thinking that it's fair to consider a 40-50% resale value on all mining hardware purchased, more in cases of super high end things like 6990s.

Some pool-mates and I were talking today about the idea of colocating mining equipment.  Considering things like electricity and cooling, it's a great idea.

Yep the high end videocards can probably be sold.
But who is going to buy your Sempron 140, your pci-e extenders, your 2x650w psu or 1x1200w psu, your outdated motherboard and your 2gb low speed value ram?

[NetTecture]

At which point HW has devalued itself only by 20-25% worth.

Not very long.

I have been thinking that it's fair to consider a 40-50% resale value on all mining hardware purchased, more in cases of super high end things like 6990s.

Some pool-mates and I were talking today about the idea of colocating mining equipment.  Considering things like electricity and cooling, it's a great idea.

Yep the high end videocards can probably be sold.
But who is going to buy your Sempron 140, your pci-e extenders, your 2x650w psu or 1x1200w psu, your outdated motherboard and your 2gb low speed value ram?

That pretty much is the weak point. Mining is not a normal computer, unless you overspend. it is a pathethic little low perforamcne hardware item with a lot of high end graphics cards attacked, and a very high energy densigty, comparable only to blade chassis.

[PulsedMedia]

At which point HW has devalued itself only by 20-25% worth.

Not very long.

I have been thinking that it's fair to consider a 40-50% resale value on all mining hardware purchased, more in cases of super high end things like 6990s.

Some pool-mates and I were talking today about the idea of colocating mining equipment.  Considering things like electricity and cooling, it's a great idea.

Yep the high end videocards can probably be sold.
But who is going to buy your Sempron 140, your pci-e extenders, your 2x650w psu or 1x1200w psu, your outdated motherboard and your 2gb low speed value ram?

My contacts come into play, i can be certain for anything that can be used in mainstream i will get at least something.

Then comes the local forums, auctions etc. which will take the 650W PSUs, 1200W PSUs etc. for better prices.

Even 5+ yo mobos have market here, so does even original DDR or even earlier! Infact, very old RAM is the most expensive around here (10+yo), and there is still people needing those oO;
and in most cases, just upgrade CPU + RAM and you got a highend machine

All you listed, except PCI-E extenders are easy sales. PCI-E extenders are a minority cost (under 5$ each) and does not matter even if i end up throwing them out. If i get x16 PCI-E extenders, those should rather easy sale to gamers with cooling issues.

The low end vid cards will sell as well, no probs.

One thing about Finland: People are poor. REAAALLY poor. 40€ might be an painful investment for some. And i'm serious! Many people working full time earn barely enough to feed themselves, even if not at minimum wage because you loose the social benefits and acquire extra costs to be able to goto work. Some for being already broke and paying off debt can barely afford anything. Even at mean wage you can't afford much, 200€ will be painful for many middle wage people.

Also stuff costs insanely much here, you are lucky to make a daily budget for food for single person below 3€ a day, and if you want to eat proper foods and properly, without skimming it's easily 8€ a day. Pizza costs regularly 8-10€ and Big Mac lunch is 7.5-8€. 1.5L bottle of coca cola costs 2.5-3.5€ (cheapest knockoffs are 1.3-1.5€/bottle). Small cup of coffee at a gas station 2€. Pint of beer at cheap bar is 2-3€, regular place 3-5€, expensive 5-9€. 1Liter of milk is about 1.5€, 400gr (0.4kg) of macaroni is atleast cheap at 0.28€. Cheese at least 6€ per Kg. Tuna costs about 1.2€ for 150gr (0.15kg) or cheapo with fishbones and stuff 0.5€ for 150gr. Coffee packet of 400gr is about 4.5€ for good brand, cheapo about 2.9€.

I just bought Antec 300 case yesterday, cheapest place asked 58€ for it, inc. VAT 23%. I just had read on one forum someone bought the same case new at US for 40$(28.37€). USB Sticks cost atleast 6€ each. DDR3 RAM costs close 10€ per Gig (vs. at US you can get 2x4G for 65$). It makes very often sense to order from US/Hong Kong/China even with insane shipping fees and customs duties, esp. for small stuff like cables. DVI-I to VGA adapters cost 8€ each here! (vs. order from HK 0.99$ inc. shipping). Even on enthusiast forums it's kind of rare to see high end computer hardware being sold oO; Highend is REALLY hard to sell here, when 8800Ultras were not that old, next gen was just released, people were bitching my asking of 300€ per card (at the time they were listed at around 650€ per card new). I tried to sell a Quad Core Extreme CPU, 2nd fastest, listed at the time at 950€, people were bitching why i won't sell it for 200€ because a 60€ core2duo is so much more worth for money.

So the low end stuff has their market, and it's a huge market. I get constantly question "can you get me a HDD for even 5€ cheaper??", "Can you get me a brand new computer, i got 100€ to spend?". Most Finns are trying to nickel & dime everything they can.

Median US income (2007) is about the same in euros as mean FI income (2009 or 2010), but taxes are through the roof. At US i hear a gallon (3.97L) of fuel costs about 2.50$ (~1.79€), so you get almost 4 times as much for just a bit higher price. My belief are rents are about the same in USD than here in EUR, so ~39% difference. etc etc etc. the list just goes on.

Because everything costs so damn much here, when people go abroad a major portion of the trip is shopping for regular items (not souvenirs or novelties), because even expensive stuff abroad is dead cheap for FI prices! Before Estonia had to raise prices due to EU, there was a lot of people constantly taking trips there to bring alcohol and tobacco, as they could get 100% profit margin, and it would still be like 1/3rd of the price here! :O