They had almost 2 years prior to launch to work this out. ~36000 work units per iteration at 60MB per work unit would be about 2.1TB per iteration. Then there is about 182.5 iterations(let's use 183) so, 183 time 2.1TB is ~395TB but then there is minimum quorum of 2 so that 395TB becomes 790TB. If I look around the room and don't see 790TB of storage there, I don't launch. I assume what happened though, is the techs said "Wait, we don't need all that because we will transmit data to the researchers at the end of each iteration.." That's when there should have been a discussion with the research team, yes, prior to launch saying we are going to be sending you 2.1 TB (or 4.2TB if quorum of 2) approximately every 3 weeks. Do you have a place for this data and oh, BTW this project is going to generate about 1 PB of data. Do you have the resources? If the answer is NO, don't launch. Do the researchers need to second copy of the results? If the work unit validated, aren't the results identical copies? If the researchers say they want both for comparison purposes, then fine but they have to have a place to put it. The fact that this project just started and is throttled just screams out "DON"T LAUNCH...". If it would have come up after a year of processing one could reasonably argue that more data was generated than what was anticipated and we need to adjust, so we adjust. So, in contrast to your statement, I am surprised and I have been since the first month of launch. This just seems like project management 101

Also know as the canonical copy. If there is bit by bit matching then that would be enough to retain and transmit.

But if one is in a storage/budget crunch, where do we trim? That seems like a good spot and can cut your requirements in half. Additionally, what does the extra copy really get you anyway? In the event that one needed to go back and re-compare results(which seems like the only reason to keep it), which is the correct copy? All you would know is that they are or aren't different. Also, if the BOINC system did a bit by bit compare to validate the WU, wouldn't they be identical? So why would you need to keep it? Also, in the rare case something seemed a little "off", it might be easier to just rerun the WU locally and be done with it rather than digging through a string of bits.

Just to add to what others are saying, a quick google search shows that in 2017, storing a petebyte on a server cost $10,000 a month. Let's make that two for $20,000.

That's a lot of electricity, but again, probably less then IBM spends on incidentals in a week.

You can save costs by unplugging backup drives once they fill, until or unless they are needed to remedy an error.

Just about everybody knows somebody who's gotten cancer, and due to the societal costs of treatment just about everybody would see themselves better off if chronic, debilitating, and expensive conditions like TB & HIV/AIDS were a thing of the past.

I don't judge people who pick other charitable causes, I absolutely take note the folks who give almost nothing. The ones who ask me how much BOINC will pay them and cannot comprehend a donation that does not somehow work its way back to themselves.

Millions of Africans aren't getting enough food. That's been true for what - most of our lifetimes? It's been so bad for so long that we're used to it. That caring 'too much' is seen as un-serious. "Let's do what we can to help, so long as it's minimal. Or comfortable. Or cost effective. And goodness, don't judge those radiant, positive people who are too busy enjoying our status quo to give much of a care to anybody else's. Their greed is innocent and pure, and precious and vulnerable and mustn't be confronted".
Edit: Clarity.

I don't think you're personally disturbed by a lack of a reference link or an emoji in an internet post.

You may want to reexamine your feelings on that one on what's actually troubling you.

Internet bandwidth is not the stated reason for the throttling.
We're all free to ask IBM for more support.

At this point the Africa Rainfall Project is barely alive.
Ladies and Gentlemen, IBM can expand it.
They have the technology.
They have the capability to build a two petabyte storage solution.
Africa Rainfall Project will be that catalyst.
Better than it was before.
Better, stronger, faster.