Failure Rate Analysis - Best 10Tb+ hard drives: Seagate, Western Digital or Toshiba?

Summary

In this video, the presenter analyzes real-world drive reliability data from Backblaze, covering 350,000 drives over a 10-year period. The focus is on large-capacity drives (10TB and above) from Seagate, HGST, Toshiba, and Western Digital. The video compares Annualized Failure Rates (AFR) across 18 different models, discussing early failures, wear-out phases, and the overall reliability of each manufacturer's drives. Key insights include the impressive performance of Western Digital drives, while Seagate faces reliability challenges with specific models. The video provides useful data for making informed decisions on drive selection based on long-term reliability.

Takeaways

• 😀 The video analyzes 200GB of real-world data covering 350,000 hard drives over 10 years, sourced from Backblaze.
• 😀 Focus is on Annualized Failure Rate (AFR) analysis for drives with capacities of 10TB and above, using data from 18 drive models with over 500 units deployed.
• 😀 The AFR is used to determine the likelihood of a drive failing within a year, but actual failure rates can vary depending on the specific model and age.
• 😀 The analysis looks at drives from four manufacturers: Seagate, Western Digital, Toshiba, and HGST (a subsidiary of Western Digital).
• 😀 Early drive failures are common, especially if quality control or logistics issues exist, but random failures tend to occur as drives age.
• 😀 The AFR increases over time as drives wear out, particularly due to physical components such as motors, bearings, and platters degrading.
• 😀 The video stresses the importance of large datasets in AFR analysis, as more data provides a clearer view of drive reliability over time.
• 😀 Seagate’s x12 model had significant AFR problems early in its deployment, leading to its decommissioning, but newer x16 models show improved reliability.
• 😀 Toshiba drives tend to show higher early failure rates but stabilize to below 1% AFR after a few months, with some drives showing better performance over time.
• 😀 Western Digital’s drives from the HC530 and HC550 ranges have the best AFR performance, with some models even falling below their quoted AFR of 0.35%.
• 😀 The reliability of the drive models varies, with Seagate having the largest fleet of drives, but Toshiba and Western Digital emerging as leaders in reliability for the models analyzed.

Q & A

What is AFR and why is it important in this analysis?

AFR stands for Annualized Failure Rate. It indicates the percentage chance that a given drive will fail in a year. It is important because it helps assess the reliability of drives over time, providing a clearer picture of which drives perform better and which may fail sooner.

What does the 'bathtub curve' represent in terms of drive reliability?

The bathtub curve is a model that shows how drive failure rates typically follow a pattern: high failure rates in the early stages (due to manufacturing issues or logistics), stable failure rates during the middle years (when the drive is functioning properly), and an increase in failure rates towards the end of its life due to wear-out failure.

Why are Seagate's X12 models highlighted as problematic in the analysis?

Seagate's X12 models are highlighted due to high failure rates early in their deployment, reaching over 2% AFR. This led to the decommissioning of most of these drives by Backblaze after identifying the issue, though the remaining units stabilized at a constant failure rate.

What are the key findings regarding Western Digital (WD) drives in the analysis?

Western Digital (WD) drives, particularly from the HC530 and HC550 ranges, showed the best reliability. They performed well with low AFRs, even surpassing the manufacturer's quoted failure rates. These drives were the most impressive in terms of long-term performance.

How does Backblaze handle the data when calculating AFR?

Backblaze calculates AFR based on cumulative data, which means that for each week, they consider the entire historical dataset. This smooths the data to provide a more coherent view of AFR development over a drive's lifetime and helps reduce the impact of individual failures.

What is the significance of using cumulative data in AFR analysis?

Using cumulative data smooths out the spikes in failure rates that can occur over short time frames. It provides a more consistent and reliable trend over time, giving a clearer picture of the overall drive cohort's performance.

How do Toshiba drives compare in terms of early failure rates?

Toshiba drives, particularly the MG07 and MG08 series, show higher early failure rates, especially in the first few weeks after deployment. However, their AFR stabilizes below 1% after the initial failure period, making them reliable in the long term despite early issues.

Why does Seagate have the largest deployment base among the manufacturers analyzed?

Seagate has the largest deployment base because they provide a wide variety of models, including drives with different capacities and features. Their drives are also generally more affordable, which makes them an attractive option for large-scale deployments.

What does the analysis suggest about the reliability of HGST drives over time?

The analysis suggests that HGST drives, especially those from the UltraStar HC520 series, have stable performance over time. While they do show some early failure rates, they quickly stabilize, and there are no significant signs of wear-out failure, making them reliable over the long term.

What was the performance of the Seagate X16 drives compared to the older X12 models?

The Seagate X16 drives performed better than the older X12 models, with a much lower AFR of around 1%. They showed signs of improving reliability, and after 3-3.5 years of use, their failure rates remained stable, making them a more dependable choice compared to earlier Seagate models.