Capacity fill/empty rate

We often pay too much attention to the static details backed by concrete (often highly-impressive) numbers rather than consider the dynamics of the underlying behavior during operation. For instance, disks in the size of a couple of terabytes suggest that we have more than enough space for our content, but we rarely think in terms of the relatively low average write transfer speeds. This means that the time one needs to fill such a drive with data or empty it to its initial state can become too long. It is not hard to notice that the problem of lacking capacity may be of our own making and pattterns of use. We could free more space, but we can't regain any time lost during the use of a slow device.

Time to fill an HDD: 1TB space = (10^6MB / 50MB/s avg. write speed) / 3600s = 5.55 hours
Time to fill an SSD: 500GB = (500000MB / 250MB/s avg. write speed) / 3600s = 0.55 hours

In this example, the second undertaking is much faster than the first, yet it still exceeds the threshold someone would be willing to wait. This is not the same as waiting fifteen minutes to obtain your passport photos, for instance. "Capacity whose time can't come" could be a possible usability problem. This means that most devices cannot simply be shipped by "neglecting the experience". Having more sensible fill/empty rates relative to the offered capacity is a good step in the right direction.

Consider four memory capacities integrated in USB sticks (8GB, 16GB, 32GB and 64GB) with average write speeds of 4MB/s, 12MB/s, 21MB/s and 36MB/s. In theory, the times it would take to fill these devices would be 0.55, 0.37, 0.423, 0.493 hours respectively. We notice that doubling the capacity in many cases is not accompanied by doubling the transfer rate, effectively leading to longer fill-up times. Writing many small files can become especially slow, which means the usage scenario is more suitable for small collections of files rather than a complete backup.

Some older tablets had to be charged for extended periods of time after usage (or even lack of it). Someone could use the tablet to watch dynamic content for 2 hours, after which they had to spend 3 hours to charge it. The battery capacity fill rate was quite low, which coupled with the fast empty (or discharge) rate practically made the device unusable. But if felt that these two rates were two separate engineering decisions, which worked together in combination with other factors to decrease the value of the device.

The empty rate could be particularly important for fire brigades for instance. If it is slower than the rate with which a fire spreads, it would be unsuitable for the task. Adjustable water debit could then be seen as much more convenient.

Filling the entire capacity of a bus with people is relatively easy, but then they may not easily find their way out. The empty rate slows down due to various obstacles: other people, bags/suitcases, strollers, etc. Trying to balance both rates by letting people come in only through the front door (after checking tickets) and letting them go out through the other doors has been shown to increase the time spent at the bus stop.

All these cases show that capacity should not be increased blindly, but in a thoughtful way, so as to be useful within the context and under maximum (not merely expected) load from usage. Integrating one component with extreme capacity at the expense of many others does not resemble a well-balanced system.