Published 8 minutes ago

Abhinav pivoted from a career in banking to pursue his first love in writing. Even while working full-time, he continued contributing as an editor-at-large, a role he has held for more than 7 years. A lifelong tech enthusiast who has built three gaming and productivity powerhouse PCs since 2018, his passion for technology keeps him closely following the semiconductor industry, from NVIDIA and AMD to ARM. His MSc dissertation explored how artificial intelligence will reshape the future of work, reflecting his curiosity about the wider social impact of emerging technologies.

When SSDs slow down or wear out earlier than expected, most users attribute it to hardware and quality issues. What many are unaware of, however, is that an important driver of premature wear is the constant "micro-write" workload they endure over a course of time.

Modern operating systems are like living, breathing ecosystems of digital activity, generating an endless stream of micro-writes that transpire inside your SSD. They log files, browser caches, telemetry data, index, update, and execute background services that constantly write tiny chunks of data to your drive. While each individual write is insignificant, they cumulatively create a high-frequency noise that SSDs, due to their architecture, handle with massive inefficiency. There are a few reasons why this can lead to your SSD giving up on you long before its expiration date.

Small files worsen write amplification

Tiny files cause big NAND wear

SSDs do not store data the same way operating systems request it. While an OS and its file system (such as NTFS, APFS, and ext4) typically manage data in 4KB clusters, NAND flash hardware is constrained by much larger ‘blocks’ that typically range from 256 KB to several megabytes, which can *only *be rewritten after a full erasure. When a small file changes, the SSD must read an entire block into the cache, modify a tiny portion, and then rewrite the whole block elsewhere. So, while data is stored in smaller ‘pages’, it can only be erased in ‘blocks’. To do so, the block must be copied and written to a new location before the old block is erased.

This process gives rise to a phenomenon known as write amplification (WA), which is a silent SSD killer that wears down drive endurance. To understand the severity of this, Micron analyzed various application benchmarks from data analytics and cloud operations. The research showed that random 4 KB writes generate the highest write amplification factor of any workload. Over time, this dramatically accelerates cell wear, especially on consumer-grade SSDs with limited over-provisioning.

Credit: Micron Technology.

The reason why this matters is that the desktop software has effectively become a micro-write factory. Major web browsers like Chrome and Microsoft Edge constantly update small SQLite databases for history, cookies, and cache state, while also rewriting session restore files at short intervals. At the OS-level, security processes from Windows Defender also add to the load by committing frequent small writes for event logs and signature updates. Similarly, system services such as Windows Search, macOS Spotlight, and background telemetry engines generate a stream of writes that continuously stress NAND flash.

Related

Your SSD has to work harder to move or change small files

It seems counterintuitive, but SSDs are actually much happier moving big files than dealing with a number of tiny ones. When you copy a large video or game folder, the drive can stream data neatly and efficiently, but every time the details of a small file change, your operating system kicks off a chain reaction behind the scenes.

The reason is that small files can generate disproportionate metadata traffic, leading to what is known as ‘metadata churn’. Each file write touches directory entries, allocation tables, journaling logs, and file indexes. On a modern OS like Windows, this can easily result in dozens of writes for a tiny change to a single file.

In a related study, Linux kernel researchers at Google found that, on certain workloads, standard file system journaling can consume up to 76% of a disk’s write bandwidth, thereby significantly accelerating drive wearthrough metadata copies.

In the study cited above, the figures were observed during a multi-threaded mail-server workload. This specific environment represents the worst-case scenario for metadata churn, but in most cases this level of overhead won’t be observed in everyday desktop use.

Budget SSDs suffer disproportionately

DRAM-less and QLC variants can compound the issue

Budget consumer SSDs without dedicated DRAM have to rely on Host Memory Buffer (HMB) or NAND-based mapping tables. When subjected to a stream of micro-writes, these drives suffer a performance penalty far greater than the ones equipped with DRAM, as the controller must constantly read, modify, and rewrite the FTL map on slow NAND cells rather than on high-speed DRAM. This essentially leads to a "double write" effect that spikes internal latency and accelerates cell wear.

This problem is further compounded with QLC NAND due to their lower native endurance (typically offering only 500–1,000 Program/Erase cycles). Because micro-writes generate a higher net Write Amplification (WA), they exhaust these limited cycles slightly faster than the sequential Total Bytes Written (TBW) ratings would suggest.

To alleviate micro-write wear, prioritize drives that feature dedicated DRAM and TLC NAND, which are equipped to handle metadata churn much more efficiently than QLC models.

What does this mean for you?

Your SSD’s longevity is determined more by the nature of your data than the total volume you write and erase. To alleviate micro-write wear, prioritize drives that feature dedicated DRAM and TLC NAND, which are equipped to handle metadata churn much more efficiently than QLC models.

Beyond the architecture and hardware, you can expand your drive’s lifespan through better storage hygiene, too. Maintaining about 20% of free space helps the controller with ‘garbage collection’, which somewhat mitigates the problem of write amplification. Additionally, if you think your workflow is heavy on tasks such as video encoding or editing, you may consider using secondary storage to keep metadata traffic away from your primary boot drive. As most users rarely (if ever) change their boot drive, it’s prudent to get an SSD that serves you for longer.

Related

4 software utilities that quietly ruin your SSD lifespan

Watch out for these if you want your SSD to have a long life