Any drive on our list of the Best SSDs would wipe the floor with any HDD in performance and power efficiency metrics. However, HDDs have historically had one saving grace compared to SSDs: their storage density, which leads to an advantage in the $/GB race. Now, according to a new report, the declining HDD market has environmental concerns on its side as well: a study carried out by the University of Wisconsin-Madison and the University of British Columbia has tried to measure just how exactly SSDs impact the environment. The verdict is that SSDs have twice the environmental impact of an HDD, but it might not be that clear.
According to the study, SSD's excessive carbon footprint compared to HDDs comes from the manufacturing process itself. The latest SSDs use multiple NAND, DRAM, and controller chips, each manufactured with cutting-edge silicon manufacturing techniques and multi-layer bonding processes, requiring both expensive materials and high electricity usage. So while HDDs have a higher carbon footprint throughout their operational lifetime due to increased energy consumption, SSDs generate the brunt of their emissions before they ever write the first byte.
The study itself did a meta-study (a study analyzing studies' results) on LCA reports from several manufacturers. According to it, the average SSD has a Storage Embodied Factor (SEF, representing the rate of CO2 emissions relative to the capacity of the storage medium) of 0.16. HDDs, on the other hand, have a SEF of 0.02. The lower the SEF, the lower the environmental impact.
The researchers also mention energy sources for semiconductor manufacturing as being mostly based on high-impact mediums such as coal. Still, the lack of data on this domain (namely, what exact percentage of energy is derived from renewable and non-renewable sources) means that the results have to be taken with a grain of salt. Further studies are definitely needed here.
The study further plotted out what the best storage option would be in certain workload scenarios and concluded that a 1TB HDD would beat a 1TB SSD by emitting 99 kg and 199 kg of CO2 against the SSD's 184 Kg and 369 Kg (over five and ten years, respectively).
However, their choice of scenarios was very limited. In fact, they plotted out a single scenario where the devices' operation was distributed throughout 80% idle and 20% active time. HDDs, in particular, typically have double the power consumption of SSDs while active (typically 8/9 W against the best SSD's 4/5 W) and much higher idle consumption (the latest SSDs consume below the 400 milliwatt/h range; HDDs have typically bottomed-out around 3 W/h). The researchers further assumed that the SSD would consume around 1.3 W, while the HDD was assumed to consume just 4.5 W, thus losing out on clarity of data.
Here's the issue: workloads matter, as does the speed at which the workload is concluded. So, for example, let's imagine we're copying a 10 GB file towards both an SSD and an HDD, and let's assume our SSD features a 2,200 MB/s write speed (which most released in the past two years already do). Our HDD, in the meantime, still features a maximum sequential write speed of 220 MB/s (such as Seagate's IronWolf Pro 4 TB does (opens in new tab)).
It's easy to see the difference: the SSD would take around four and a half seconds to transfer the file, at an average power consumption of 4.5W, before returning to its milliwatt-sipping idle state. On the other hand, the HDD would chew through the file transfer in around 45 seconds (ten times longer), at an average power consumption of 8.5 W (consuming 89% more power). All in all, the HDD would consume around 15 times more power than the SSD to process the same workload.
Measuring a component's environmental impact is a much more complex endeavor than it might initially seem. While power efficiency is generally a good metric (by indicating the amount of energy needed per unit of work), upstream carbon impacts, those that happen throughout the manufacturing process itself and until the product is in your hands, are generally much more impactful. You have to measure the carbon footprint of the materials extraction process, the logistics that get those materials through the factory and into your hands, and even the carbon footprint of the workers' activities.
The impact of discarding a piece of hardware too can vary wildly, depending on if you send it towards a specialized recycling center or if you just throw it in a garbage can (please don't do this, especially not when you have 8,000 Bitcoin inside the HDD itself). Manufacturers usually capture these metrics by creating LCA (Life Cycle Analysis) reports.
These are just some elements that must be considered when planning an optimized solution. Datacenters are well aware of both the $/GB cost and power consumption metrics for both SSDs and HDDs. That's why all recent supercomputing designs feature both a hot data node (composed of fast SSDs that ferry in-processing information) and cold data storage nodes (composed of HDDs for data that's simply being stored, where the power consumption difference matters less).
It's exceedingly hard to study and calculate something as complex as the environmental impact of a single component — particularly when there are logistics, supply chains, and workload variables to consider. We must remember not to take the data out of context. While the research provides insight into the need to properly account for the entirety of an HDD and SSD's carbon footprint, there are just too many variables to consider in order to stake any concrete claims of the increased environmental impact of HDDs against SSDs.
In general, we'd say that consumers have many more benefits to take away from investing in SSDs compared to HDDs. The concerns about SSDs wearing out aren't too relevant for most users — most don't write nearly enough data on their SSDs to render them inoperable. Of course, environmental concerns are extremely important, and we should always strive to reduce our impact as much as possible. But you can always cut back on the time it takes for your file copies to finish while helping the environment elsewhere with the extra hours of your time you'll have available.
My PC almost 2 years old, both nvme & hdd same age. nvme is boot drive and yet its been on 3000 less hours in the same time.
So ssd may not be cheaper in long run but nvme might?
Your choice of storage device, or your plastic straw from Starbucks IS going to ruin the planet.
Just because CO2 is "natural" is utterly meaningless in regards to its effects. Mass extinction events are natural, deadly famines and floods are natural, neurotoxins are natural. None are particularly fun or desirable. "Natural!" (like "organic!") makes for great marketing for slapping onto breakfast cereals or shampoo bottles, but is meaningless when it comes to any positive or negative impact in and of itself and implies nothing in terms of health or harm.
If you're intending to spool up another 10-20 datacentres storing a few tens of petabytes each (e.g. Facebook alone is producing 4-5 petabytes per day, and that has to go somewhere, and the be backed up somewhere else, and stored in duplicate at multiple locations to minimise access latency and meet bandwidth demands, etc) every year, each sucking down tens to hundreds of megawatts each, then the emissions from your operations are not insignificant and minimising them more complex than it first appears. And how different technologies result in emissions are also of concern as to how those datacentres are designed.
Using the article's example: if you are already using 100% renewable power for your datacentres, then HDDs may be the greener option despite their higher power consumption due to their lower emissions during manufacture. On the other hand, if you do not have access to reliable renewable power that can meet 100% capacity demand (e.g. lack of installed capacity, lack of local battery buffer capacity, etc) then the reduced total energy demand for SSDs may make the total lifecycle emissions lower overall despite the higher emissions during manufacture.