Sequential & Random Read or Write Speed – Explained
Mainly the performance of a drive is measured by two different types of data operations, sequential and random operations. So, by using the ‘Sequential/Random Speed Stats’ you can easily calculate any drive’s performance which you can use to compare it with others and make a wise decision.
As we all know, every data that we saved in our computer is written on blocks. ‘Sequential Read/Write Speed’ simply means that how fast that drive can write or read the data from a series of blocks. On the other hand, when we turn on our computer, it starts to process numbers of files from multiple locations, which simply means that the drive needs to access the data from random blocks, again and again;
And that’s where the ‘Random Read Speed’ comes in.
To understand ‘Random write speed’, let suppose you are installing software or an OS update (a little one). You’ll see that a minor update or an installation will surely take a while, however, a 40GB 4K video will load easily to play or seak. Generally, updation or installation takes time because the software needs to write multiple files at multiple locations, which basically take time as compare to write a big file in a series of blocks.
So, In sequential operations mainly large data chunks are operated in a collective manner without repeated seeks. Sequential operations work for bigger and serially managed data blocks whereas random operations work with small and randomly located, scattered data blocks, causing more latency time.
To understand ‘Sequential and Random Read/Write Speed’, first, you’ll need to understand ‘IOPS’.
IOPS is a commonly used technical acronym for the phrase ‘Input/Output Operations Per Second’, it is a scale for measuring the performance of a storage device or network;
More value in the IOPS signifies the capability of executing more operations per second. In simple words, data operations speed is proportionate to the IOPS value of the drive. More read and write operations per second enables the drive for performing smoother and faster.
In solid-state drives or you can say SSDs, IOPS is always a lot higher than hard disk drives or HDDs. HDDs work on the basis of RPM or revolutions per minute (being circular disks the drive head moves around inside) which is usually 7200 RPM with a value of 90 IOPS, but SSDs are made of solid-state flash memory banks and can provide an IOPS of more than 400,000.
You know that by using a simple equation you can easily convert the IOPS digits to MB/s for the better understanding.
IOPS = (MB/s Throughput / KB per IO) * 1024
MB/s = (IOPS * KB per IO) / 1024
Let’s say that you have an SSD claiming a Random 4K write speed of 40,000 IOPS and it achieves 155MB/s in the CrystalDiskMark with the QD32 write test. So to convert the 155MB/s to IOPS, we perform the following calculation:
- IOPS = (155 / 4) * 1024
- IOPS = 38.75 * 1024
- IOPS = 39,680
To see what throughput we need to achieve to match the actual 40,000 IOPS claim, we can do this calculation in reverse:
- MB/s = (40,000 * 4) / 1024
- MB/s = 160,000 / 1024
- MB/s = 156.25
Sequential & Random operations in SSDs
In SSDs, operations are always faster as there are no moving drive heads for data operations; all the data cells are in solid state and working simultaneously in a lightning fast speed. For this flash drives, it all depends upon the device’s internal controller logic and memory interface speed.
In sequential and random operations the NAND package of SSD does data locating and arranging it all in a serial manner to execute the task without seeking, which majorly transforms operations looking sequential and smoother.
Sequential & Random Operations in HDDs
In HDDs both the operations go slow because of the moving drive heads, when the fastest HDD can accomplish 15k RPM and for operational activities to read and write the disk keeps rotating. IOPS of the HDDs are dependent on random seek time. While operating in a sequential manner, first it seeks the storage location on the disk and then operates the data blocks serially, otherwise, in random operations it seeks and operate time and again to execute the task.
#Ultimate Performance Comparison
In our real-world activities it never shows up significantly, but SSDs are a lot faster in executing tasks with smaller data blocks than HDDs. In sequential data write sector SSDs can perform approximately 3.4 times faster whereas in the read sector it is faster by about 3.5 times to 5 times.
But when it comes to random operations the only latency time SSDs take is to reprogram the controller which makes it more than 50 times faster to process read requests and almost 100 times faster to process the write ones.
In transfer operations, SSDs always read and write upon the basic unit of pages. Otherwise, it consumes less power and works more efficiently as a storage device. As I said, booting up an OS can show users the proper speed differences of random operations in SSDs and HDDs.
“That’s all for now, thanks for sticking with the article. It is always good to let me know about your views, in the comments below.” 🙂
Madhur, great post, clearly explained and informative.. Thanks
So that’s why SSD always faster than HDD, thank you Madhur!
Great content! Super high-quality! Keep it up! 🙂
Thanks man! Definitely appreciate you taking time off your day to inform us!
Best explanation that I have been able to find on the net so far.
But still I am trying to understand here what the specified performance of a HDDs means in a real life situation. As I see it:
HDD will perform random write operations when installing and uninstalling programs or updating.
HDD will perform sequential write operations when copying a large zip (or mov or ISO) from D:\ (SSD) to C:\ (HDD)
Is this correct Madhur?
Hi Madhur, nice post. Thanks!
I have a RAM-disk that perform sequential data (read and write) very fast (up to 2096 MB/s) but random operations (read & write) only 60 MB/s and often less. (read operations are in both cases even slower)
I always though that a RAM-disk (that is only a separation of your normal RAM) should be no so much effected by type/structure of data and I thought read should be faster than write.
Thanks & have a nice day,
winsat disk -seq -write -drive F
winsat disk -ran -write -drive F
winsat disk -seq -read -drive F
winsat disk -ran -read -drive F