Most large sites have large numbers of spinning hard-drives. These are replaced in a planned way, rather than waiting for a disk to fail.
I’m looking to partner with others on an idea i have to provide a structure to this retirement of disks.
The idea is to arrange the physical disks into logical queues, from oldest to newest.
The standard practice is as follows:- when the oldest is retired a new disk is put in place, but the new disk is, say, twice the capacity.
The idea is to reverse capacity and disk count, so that the oldest is classified by a standard size, say 1 TB, and a calculation is made to calculate what fraction of a disk of that generation provided 1 TB. You replace 1 TB of old with 1 TB of new, but can see a reduction in the calculated quantity of disks that were required to achieve that.
We then have measurable reduction in emissions from electricity used and embodied CO2e.
I’m not sure I entirely follow your logic. Is this about replacing what has been used for storage rather than the next in line to be replaced per some standard depreciation manifest?
Hi James
I haven’t explained the idea very well, have I.
Unfortunately I can’t get onto the forum because my browser is blocking access due to HSTS being enabled and a self-signed certificate being used. And I don’t have an email address to tell anyone about it.
To go through the idea a bit better -
This is a procedure to reduce carbon emissions by using “Disks per Terabyte” rather than Terabytes per disk.
Disks are arranged in generations, usually in sequence of year of installation.
The oldest disks are replaced as per the organisation’s normal policy on retiring old disks before they fail.
But when newer disks are installed to replace them a calculation is done. The calculation is of disks per one TB. Normally we would think in terms of x number of disks with y TBs per disk. But this approach reverses that.
So when the oldest generation is discarded it is replaced by a newer generation of disks that usually have a larger capacity per disk. From that we calculate disks per one TB and it will usually become fewer disks per one TB.
The calculation disks per one TB normalises the amounts to make comparison easier and tells you how much of each disk drive has been saved in terms of embodied carbon and saved in electricity not used per disk.
An objection to this i have heard so far is that the total quantity of installed disk increases every year so no comparison can be made. In my opinion disks per one TB normalises it to a scale of TB which is a valid comparison for any quantity.
Another objection i have heard is that disks are of different types so no comparison can be made. In my opinion if the disks are broadly similar the differences won’t make much difference to the calculation. If they are quite different then the generations can be subdivided so that like is compared to like.
I’m not sure if this helps.
I would be interested to hear any comments.
regards
Cheney ketley
RAL STFC, UKRI.
Right. I think I can see the logic here.
Hi James,
Thanks for responding.
Sounds like I’m making progress although I can see ways to make it even clearer and to sell the benefits a bit more.
If you were running a big disk array do you think it would be something that would help with carbon accounting and reduction?
Regards
Cheney ketley
SC dept
STFC, UKRI.