When looking to buy a new hard drive or SSD for your PC we often just consider capacity. If performance is important to you, however, there are a few metrics you can look at: sequential and random read and write performance.
But what are these performance metrics, and should you put as much stock in them as the manufacturers do?
Storage Basics: What Are Reads and Writes?
When you’re putting new data on the drive it’s called writing. This is when you carry out operations such as saving a new file, or modifying an old one. Reading, then, is when you access that data. This can be to open a text document, a photo, a program, or whatever else is stored on your drive.
How these operations work changes based on whether it’s a hard drive or an SSD. Hard drives have mechanical parts including a read/write head and a spinning platter that stores the information. To retrieve data the head has to be positioned at the point on the platter where the data is stored much like how a record player has to drop on the right spot of a record to play the song you want. The difference is that the hard drive is working much faster and more accurately than a record player.
SSDs don’t have mechanical parts. Instead, these drives are made up of data-saving space called cells, which are grouped together to make pages, which are then combined again to make blocks. While SSDs are very fast at reading and writing data quickly on newer drives, they can be a bit slower when overwriting data–replacing old data with new data. That’s because an SSD can only write data to available pages meaning pages that don’t have any data on them. If your drive doesn’t have enough free space available then the SSD has to erase data in blocks. Since it needs to erase at the block level it may need to copy an entire block and then rewrite the whole block including the new data you asked it to save. This all happens in fractions of a second, but the reason for such a roundabout process is that if the SSD tried to erase data at the lower page level it would risk corrupting nearby data not queued for erasure.
What We Do With Our Storage Drives
The idea behind sequential and random read/write metrics is to reflect how we use our storage drives on a day-to-day basis. If you are transferring a large file to your drive or accessing that large file then we’re talking about sequential read and write operations.
When you’re using a hard drive, sequential reads or writes make life easier. The drive’s read/write head hits the part of the drive where the file is saved or going to be written and gets to work. If you have an SSD, sequential operations can also go a bit faster since you’re writing or reading from a clump of blocks.
Random read/write performance, on the other hand, is about reading or writing small files scattered throughout the drive. This can reflect what happens when you do things like open a Word document and a spreadsheet while launching Chrome. Hard drives have a harder time than SSDs with random operations since it increases seek time, which is when the read/write head has to position itself to get the requested data.
Sequential Versus Random
So now we understand the difference between sequential and random reads and writes, how does this apply to you and your purchasing decisions? As with anything PC-related, it all depends on what you’re doing.
If you’re using a PC to mostly read and write large files for one user then sequential performance becomes important. For most of us, however, paying closer attention to random performance (when the metric is available) will be more helpful since it’s often more reflective of how we use our computers on a day-to-day basis.
The problem is that not every drive on the market will show you random read/write metrics since sequential metrics often look more impressive. When you can’t find random metrics for the drive you’re interested in you can either read third-party reviews, look at alternative drives, or just buy it anyway and hope for the best. If you choose the latter strategy then it’s highly recommended to go with a known brand, especially when we’re talking about SSDs.
When you do find random metrics they will usually be expressed in input/output operations per second (IOPs). The basic idea is that the more operations per second that a storage drive can do then the better it will perform. The problem is there are a number of tests that can come up with some substantial IOPs numbers that may not reflect what you’ll see at home. In general, you want to look at IOPs tests that have what’s called a queue depth (QD) of 1 or, at the most, 8. Queue depth is how many operations are lined up and waiting to be processed by the drive. The way storage drive firmware works is that the bigger the queue depth the more efficient the drive becomes. The problem is most home users would struggle to get up to a queue depth of 8 never mind 32. So that kind of efficiency you would never see, which is why a measure of 1 to 8 often provides a better understanding of what kind of performance you can expect to see.
So What Should I Buy?
So what do we take away from all this? Just as we’ve always understood things an SSD performs better than a hard drive. So the first step is to buy an SSD when performance is the most important consideration. If you need to find differences in SSD performance look to random and sequential read/write benchmarks to compare SSDs and focus on random performance for daily uses. The only caveats are if you are using your computer to continually move around and process large files.
Ultimately, however, most people don’t really need to sweat all of this. Just buy an SSD from a reputable storage maker in the capacity you need at the price you can afford. If capacity is more important then get a hard drive as they provide better value for higher capacity storage—at least for the next few years.
To see what specific SSDs we recommend, check out our PS5 SSD buying guide. Though we picked them with PS5s in mind, they’re great choices for PCs as well, since speed is a primary factor in both cases.