H/w Hosting Directory & Reputation

[jtoomim]

Ha ha, impeded airflow! One of my coworkers had actual numbers for MTTR (Mean Time To Repair/Reconfigure/Replace) and MCTR ('Cost' instead of 'Time') for tight racks vs. slack racks (after a merger of two companies with vastly different cultures) and the numbers aren't supporting the claims of the OCD crowd.

I however understand your position when marketing yourself to the owners of Spondoolies' hardware which is in large part "status symbol" crowd.

I just went and measured. In the wake of the mess of power cables, airflow is around 2.3 m/s. In a similar position above, not in cable wake, airflow is about 3.8 m/s. If you feel around with your hand, the effect is quite distinct. There is no noticeable effect from ethernet cables, though.

The performance and reliability of most servers is not significantly affected by airflow resistance. If a server gets a little bit too hot, it increases the fan speed. With SP30s, the fans are running full blast anyway, and the performance is affected by internal temperature. I would think that this would be a bit more sensitive than MTTR/MCTR for some 10 kW racks.

[1714943343]

1714943343

[1714943343]

1714943343

[1714943343]

1714943343

[1714943343]

1714943343

[1714943343]

1714943343

[2112]

I just went and measured. In the wake of the mess of power cables, airflow is around 2.3 m/s. In a similar position above, not in cable wake, airflow is about 3.8 m/s. If you feel around with your hand, the effect is quite distinct. There is no noticeable effect from ethernet cables, though.

The performance and reliability of most servers is not significantly affected by airflow resistance. If a server gets a little bit too hot, it increases the fan speed. With SP30s, the fans are running full blast anyway, and the performance is affected by internal temperature. I would think that this would be a bit more sensitive than MTTR/MCTR for some 10 kW racks.

Imagine what you could achieve if you actually opened the Spondoolies' cases and blew the air over their radiators using large centrifugal fans. Obviously, don't do that with customer's equipment, experiment with your own. I understand that as a prepaid hoster, you care very little MTTR/MCTR, downtime isn't hitting your pocket. Also, the expected useful lifetime of Bitcoin miner is much shorter than a general purpose server, so you may in fact see very little failures before your customers liquidate their equipment.

I've posted couple of months ago on the Spondoolies' thread regarding just that:

Just take away the external metal casing. Flip the machine on its side. Borrow a good centrifugal "air mover" from a neighborhood water damage repair contractor and some baffling that they use to direct the air. Also borrow a contact-less thermometer to understand the why the SP10 casing is badly designed for cooling and creates unnecessary temperature gradients. I don't know if Spondoolies' firmware has a "seized fan" shutdown programmed in, so you'll have to experiment with which fans can be removed.

The alternative is just to dress in your best cold weather clothes and photograph yourself near your Spondoolies' machines. You'll have a nice memento.

I haven't used Spondoolies' hardware personally, but I do have relevant experience of restarting bankrupt batch data processing facilities filled out with racks of 1U and 2U hardware from Dell and Sun. It had the same symptoms: the bottom was getting hot and the intake air had to be really cool. Neither Dell nor Sun field service technicians were giving us any trouble after seeing our temporary facility w.r.t. warranties and service contracts. We've actually lowered the rate of faults due to seized fans and accumulation of dust and debris. Only hard drive failures had increased.

The physics of it is really simple: it doesn't make sense to first concentrate the heat only to dissipate it right afterwards. But my point of view is different than yours, you are just providing the hosting service, whereas I talk like end-user minimizing the overall costs. Your customers obviously do care about the appearances, whereas I cared much more about things like avoiding downtime, minimizing staff workload, maximizing their morale and nearly zero about how it looked.

[jtoomim]

Hey, we've got a p2pool node up for testing. Anyone want to try it out?

74.82.237.66:9332

0% fees, 0% author contribution, hosted in central WA on one of our 100 Mbps symmetrical fiber links. IP address will change eventually, and we might be rebooting it, so don't expect it to be reliable just yet.

[blunthammer]

I am cross posting this in the right place on purpose

Wanted to let the community know that there is available MEGAWATT SCALE capacity out there ready to go, inexpensive (by data center standards), and finished space (you can move in this weekend).

The Coin Foundry has 3.6MW of available capacity right now in Texas at 5 cents (may be gone by 9/8/2014)

The Coin Foundry has 8MW in Washington at 5 cents

The Coin Foundry has 2 MW in Virginia at 7 cents.

The facilities are prepped and you can move into the Texas and Virginia ones this weekend.

The Coin Foundry has 6 other buildings at 20+ MW each around the US at 5-7 cents power they are performing due diligence on.

Stop the hair splitting over 1-2 cents for the guys at megawatt scale.

You know what kills profitability fastest?

NOT STARTING TO MINE TODAY.

http://www.thecoinfoundry.com - only Dallas is up on the site, they just got control of Virginia and Washington Friday. Dallas may be gone by next week with a miner shipping 100 rigs this month and 300 next month, just FYI.

I am a data center guy, and I tell everyone that there is limited product available (Tier I ) that is cost effective for miners. It's not the power but the cooling overhead that adds $$$ to the equation. The big data center players all overbuilt for mining operations (redundant electrical & mechanical systems that aren't needed for mining (no one dies if a facility loses power for a few hours). So they lose money to attract the miners and can't sustain keeping the miners as customers because you can lose money at scale too.

happy to answer questions - mark at blunthammer.com

[Arctic]

Hi Mark,

Your earlier post https://bitcointalk.org/index.php?topic=717049.0 suggests it's 5c + rent of $50-$100kW.

 is that 5c per kW/h plus $50-$100 per kW per month?

[blunthammer]

It's $125,000 per month per megawatt all inclusive of power, cooling, rent. The whole building is $250,000 per month all inclusive.

There are no tenants in the building so there is no need to chop up power/rent, etc. as I did before. Because there are no other leases to provision resources for now, it is simpler to build out revenue and cost models. I wanted to figure out the all in number for the building, and so that is what I did.

[Biodom]

Want to give kudos to sidehack at Gekkoscience hosting.
FEDEX did not deliver to correct address, so he drove to depot to pick up my SP-30.
Everything was set up fast and is chugging along great so far (at ~4.5Th).
Thanks

[Arctic]

It's $125,000 per month per megawatt all inclusive of power, cooling, rent.

OK got it.  That's why we went to Sweden.  

Our all inclusive price for 1MW is $79,000.   I assume the reason why thecoinfoundry is 58% more is that you are having to run conventional aircon in traditional DC space.


As difficulty ramps it will become less and less possible to mine profitably anywhere that PUEs are not very close to 1.

[blunthammer]

It's $125,000 per month per megawatt all inclusive of power, cooling, rent.

OK got it.  That's why we went to Sweden.  

Our all inclusive price for 1MW is $79,000.   I assume the reason why thecoinfoundry is 58% more is that you are having to run conventional aircon in traditional DC space.


As difficulty ramps it will become less and less possible to mine profitably anywhere that PUEs are not very close to 1.

You can mine profitably so long as you build a business model that insures it. Thinking you can buy a couple of rigs and put them in your garage in scottsdale AZ from June-September and profitably mine is not going to work. Mining at scale and placing bigger bets to buy in is what is required. Fewer and fewer people will have the balls resources or desire to go bigger and bigger to get a seat at the table.

The coin foundry value prop as I understand it is:

The sites are there ready to go in a real building.

You overnight rigs today, you are mining in 24-36 hours with whatever rigs you can mount and turn on
How many BTC can you mine with 1 MW in 24 hours? (I don't know)
Is it enough to cover the cost for that day and the weeks required to squeeze 58% premium over mining 1MW of BTC?
Those are the business questions that need answering.

Someone who wants to put 500 rigs in the site over the next 8 weeks seems to think it's a screaming deal. Dallas will be gone by 9/30 or sooner. New rigs are shipping and they want the jump on the hash rate to get going now vs, trekking around eastern Washington for 2 weeks to find a building. Then another 4 weeks to have a utility design a distribution system for them and drop in gear to step down the voltage, and then coordinate rack implementations, cable runs, and wiring installs (2 weeks) before the rigs get there and then have the people, process, and space to unpack rigs and get em racked.

It's all super easy until you start...

[jimmothy]

You can mine profitably so long as you build a business model that insures it. Thinking you can buy a couple of rigs and put them in your garage in scottsdale AZ from June-September and profitably mine is not going to work. Mining at scale and placing bigger bets to buy in is what is required. Fewer and fewer people will have the balls resources or desire to go bigger and bigger to get a seat at the table.

The coin foundry value prop as I understand it is:

The sites are there ready to go in a real building.

You overnight rigs today, you are mining in 24-36 hours with whatever rigs you can mount and turn on
How many BTC can you mine with 1 MW in 24 hours? (I don't know)
Is it enough to cover the cost for that day and the weeks required to squeeze 58% premium over mining 1MW of BTC?
Those are the business questions that need answering.

Someone who wants to put 500 rigs in the site over the next 8 weeks seems to think it's a screaming deal. Dallas will be gone by 9/30 or sooner. New rigs are shipping and they want the jump on the hash rate to get going now vs, trekking around eastern Washington for 2 weeks to find a building. Then another 4 weeks to have a utility design a distribution system for them and drop in gear to step down the voltage, and then coordinate rack implementations, cable runs, and wiring installs (2 weeks) before the rigs get there and then have the people, process, and space to unpack rigs and get em racked.

It's all super easy until you start...

I'm sorry but you sound like a used car salesman. Are you the same person who designed the website? (who clearly has no clue how bitcoin mining works)

Your company is offering a horrible deal, no question about it.

A 12 month contract locked in with such shitty rates would mean a guaranteed loss of ~$0.5m assuming you have free hardware and it starts mining today.

Nobody gets free hardware so a more reasonable estimation would be a loss of ~$1.3m.

http://btcinvest.net/en/bitcoin-mining-profit-calculator.php?diff=27428630902&dcosts=0&diff_mincrease=15&blpbtc=25&dhsmhs=1000000000&diff_mincreasedecrease=3&btcusd=487.49&dpowcon=1000000&btcusd_mincrease=1&pcost=0.175&calcweeks=32&dleadtime=0&action=calc

[blunthammer]

You can mine profitably so long as you build a business model that insures it. Thinking you can buy a couple of rigs and put them in your garage in scottsdale AZ from June-September and profitably mine is not going to work. Mining at scale and placing bigger bets to buy in is what is required. Fewer and fewer people will have the balls resources or desire to go bigger and bigger to get a seat at the table.

The coin foundry value prop as I understand it is:

The sites are there ready to go in a real building.

You overnight rigs today, you are mining in 24-36 hours with whatever rigs you can mount and turn on
How many BTC can you mine with 1 MW in 24 hours? (I don't know)
Is it enough to cover the cost for that day and the weeks required to squeeze 58% premium over mining 1MW of BTC?
Those are the business questions that need answering.

Someone who wants to put 500 rigs in the site over the next 8 weeks seems to think it's a screaming deal. Dallas will be gone by 9/30 or sooner. New rigs are shipping and they want the jump on the hash rate to get going now vs, trekking around eastern Washington for 2 weeks to find a building. Then another 4 weeks to have a utility design a distribution system for them and drop in gear to step down the voltage, and then coordinate rack implementations, cable runs, and wiring installs (2 weeks) before the rigs get there and then have the people, process, and space to unpack rigs and get em racked.

It's all super easy until you start...

I'm sorry but you sound like a used car salesman. Are you the same person who designed the website? (who clearly has no clue how bitcoin mining works)

Your company is offering a horrible deal, no question about it.

A 12 month contract locked in with such shitty rates would mean a guaranteed loss of ~$0.5m assuming you have free hardware and it starts mining today.

Nobody gets free hardware so a more reasonable estimation would be a loss of ~$1.3m.

http://btcinvest.net/en/bitcoin-mining-profit-calculator.php?diff=27428630902&dcosts=0&diff_mincrease=15&blpbtc=25&dhsmhs=1000000000&diff_mincreasedecrease=3&btcusd=487.49&dpowcon=1000000&btcusd_mincrease=1&pcost=0.175&calcweeks=32&dleadtime=0&action=calc

Jimmothy, thanks for the guidance. I know the coin foundry guys and I know data centers. I clearly don't know as much about mining as most of the folks on here, I haven't had the time in Bitcoin that I have had in data centers.

If it's a totally shitty deal, what's a good deal? What makes it a good deal?
I have to assume that not everyone owns the fastest rigs. What if they did? What if they owned more efficient rigs? Why doesn't everyone just run those?  Why aren't those the standard? What do you use?
Is the only unit of measurement the Kw or Kwh?
What does an efficient design of a rig or a facility do to the profitability?

I appreciate the link to the mining calc. Where do you mine? At what scale? How do you get a good deal?

What I don't see calculations for in that calculator are CapEx allowances, rent, taxes (ad velorum, sales, utility), debt service, insurance, infrastructure/tenant improvements, wiring, ducting, cabinets, labor, permits, security (physical & logical), and legal.  Telecom services, shipping, travel, and all of the other things that are material costs of a business.

So what makes a great deal and why? I know you don't speak for the global community, I am curious...

[jimmothy]

Jimmothy, thanks for the guidance. I know the coin foundry guys and I know data centers. I clearly don't know as much about mining as most of the folks on here, I haven't had the time in Bitcoin that I have had in data centers.

I have to assume that not everyone owns the fastest rigs. What if they did? What if they owned more efficient rigs? Why doesn't everyone just run those?  Why aren't those the standard? What do you use?

What does an efficient design of a rig or a facility do to the profitability?

Sure you can buy more efficient hardware like SP31 or S3+ but none are shipping now.

Is the only unit of measurement the Kw or Kwh?

What I don't see calculations for in that calculator are CapEx allowances, rent, taxes (ad velorum, sales, utility), debt service, insurance, infrastructure/tenant improvements, wiring, ducting, cabinets, labor, permits, security (physical & logical), and legal.  Telecom services, shipping, travel, and all of the other things that are material costs of a business.

The important units are $/GH(capex), W/GH(efficiency), and $/kwh(opex).

I appreciate the link to the mining calc. Where do you mine? At what scale? How do you get a good deal?

I stopped mining and sold my hardware because I pay ~$0.15/kwh which is not viable in the long run.

Giant scale mining operations are already beginning to take advantage of <$0.05/kwh locations.

So what makes a great deal and why? I know you don't speak for the global community, I am curious...

A good deal would be less than $0.1/kwh without a ridiculously long term contract. (less than 3 months would be ideal.)

Nobody wants to pay for 12 months of hosting when you stop earning a profit after 4-5.

[QuiveringGibbage]

couple of new listings in the directory from the past fortnight:

WIP Last update: 2014-09-05
UNITED STATES of AMERICA
=====================
URL: http://www.omnisdatacenters.com/colocation/
Price: Contact Us For Pricing: sales@omnisdatacenters.com (877) 393-4678
Location: Phoenix, Arizona, USA
Comment: Representative is Omnis
Thread: https://bitcointalk.org/index.php?topic=762112.0
=====================
URL: http://www.thecoinfoundry.com/
Price: $125,000 per Megawatt per month. You can lease the entire facility for $250,000 per month.
Location: Dallas Texas, USA
Comment: Representative is blunthammer
Thread: https://bitcointalk.org/index.php?topic=767595.0
=====================

I am back in the game, SP30 arrived at Sidehacks. Will double that next month with another SP30, and I've decided to also go with MichColo again.
Stats from Ozcoin:

Code:

[10:58:18] <+QG> .w crassisquama
[10:58:19] <@ozbot> This round (0d 3h 36m) QG.crassisquama has 13310019 shares (8230 stales; 0.062%), avg 4.91 TH/s 

Cheers,
QG

[RoadStress]

A few remarks:

=====================
URL: http://stealthyhosting.com/colocation.html
Price: USD$450/mth, no setup, mth 2 mth
Location: USA, Seattle, WA 98124
Comment: We could do 208v if needed.

This is not clear. I have contacted them and they told me that their price is ~175$/kW. It may go down with a bulk order, but it wasn't stated.

URL: http://www.omnisdatacenters.com/colocation/
Price: Contact Us For Pricing: sales@omnisdatacenters.com (877) 393-4678
Location: Phoenix, Arizona, USA
Comment: Representative is Omnis
Thread: https://bitcointalk.org/index.php?topic=762112.0
=====================

Domain was registered just a few months ago and only for 1 year. I can't find any information about them or their history and they haven't replied on their thread and I suspect them to be a shady business. I would advise people to double check them if they want to send some miners over there. This entry should have a warning sign until we get some more information/proofs of them being a real and trustworthy business.

[jtoomim]

Toomim Bros is currently sold out of capacity. We will have more capacity coming online around the beginning of October.

http://toom.im/images/three_full_racks.jpg


(Many more machines are en route.)

[Arctic]

The important units are $/GH(capex), W/GH(efficiency), and $/kwh(opex).

Hit the nail on the head jimmothy.


The first two are down to your hardware supplier - kit is getting better and better power efficiency wise down to about 600W/GH from 1000W/GH last generation.

The last part is down to your hosting facility. 

At Hydro66 we have plenty of spare capacity available now at prices as low as $79/kw/m including rent, connectivity etc. 


A

[QuiveringGibbage]

A few remarks:

=====================
URL: http://stealthyhosting.com/colocation.html

This is not clear. I have contacted them and they told me that their price is ~175$/kW. It may go down with a bulk order, but it wasn't stated.

URL: http://www.omnisdatacenters.com/colocation/
=====================

Domain was registered just a few months ago and only for 1 year. I can't find any information about them or their history and they haven't replied on their thread and I suspect them to be a shady business. I would advise people to double check them if they want to send some miners over there. This entry should have a warning sign until we get some more information/proofs of them being a real and trustworthy business.

thanks for the remarks have updated directory accordingly. Got my stealthyhosting quote from April this year, so yer, little out dated.

will be using your affiliate link again shortly looking to pick up another Spondoolies-Tech SP3x.

Cheers,
QG

[dumo]

After careful review we decided to send our equipment to Toomim Bros for hosting. So far the results have been fantastic. Jonathan is incredibly accommodating and will go out of his way to assist you. I think the key is that he also owns and operates the same hardware and has a genuine interest in it and wants to see yours (the customer’s) performing to the best of its capabilities. Immediately upon hook-up, we realized a constant 2+ TH/s  increase over the original factory hosting hash rate i.e the best we saw at the factory hosting was 82 TH/s over the past several months, now we are seeing constant 84 TH/s. So much so we should have moved there a long time ago !

Highly recommend Toomim Bros. for your hosting needs.

[RoadStress]

After careful review we decided to send our equipment to Toomim Bros for hosting. So far the results have been fantastic. Jonathan is incredibly accommodating and will go out of his way to assist you. I think the key is that he also owns and operates the same hardware and has a genuine interest in it and wants to see yours (the customer’s) performing to the best of its capabilities. Immediately upon hook-up, we realized a constant 2+ TH/s  increase over the original factory hosting hash rate i.e the best we saw at the factory hosting was 82 TH/s over the past several months, now we are seeing constant 84 TH/s. So much so we should have moved there a long time ago !

Highly recommend Toomim Bros. for your hosting needs.

+1 I was very impressed by their assistance too. This is my top option right now for hosting.

Also in other notes Advania is out of rack space for the moment.

[QuiveringGibbage]

WIP Last update: 2014-09-20
EUROPE
=====================
URL: https://bitcointalk.org/index.php?topic=776290.0
Price: For a small Fee, we will offer - 1: Free Electricity, 2: Free cooling, 3: Free Storage, 4: Free ''return shipping'', 5: Free Repairs
Location: SWEDEN
Comment: Operator is Sakarias-Corporation Sakariasisak@hotmail.com
Thread: https://bitcointalk.org/index.php?topic=776290.0
=====================

new listing this week, new venture, good luck to him, I hope it all works out.

* QuiveringGibbage patiently awaits 2nd SP30 to ship... anyone willing to sell theirs? i want more, now!! .. 5BTC?!?

Cheers,
QG