Backblaze's ``Statistical data by manufacturer/model 2023 Q2 edition'' that shows models that are prone to failure and highly reliable models from the operational track record of over 240,000 HDDs is released



Backblaze , a cloud storage company, analyzes the failure rate of over 240,000 HDDs operated worldwide in the second quarter of 2023 (April to June) based on the manufacturer, model, and years of use. Another statistical data 2023 Q2 edition' has been released.

Backblaze Drive Stats for Q2 2023
https://www.backblaze.com/blog/backblaze-drive-stats-for-q2-2023/



Backblaze, which provides cloud storage services, operates a total of 245,757 drives in data centers around the world. Of these, the failure rate report for the second quarter of 2023 was released for 240,940 data HDDs, excluding boot drives and test HDDs.

The table below summarizes the failure rate of data HDDs by model, 'MFG' is the HDD manufacturer, 'Model' is the model name, 'Drive Size' is the capacity, 'Drive Count' is the number of units in operation, ' 'Avg.Age' is the average number of months of operation, 'Drive Days' is the cumulative number of operating days in the quarter, 'Drive Failures' is the number of failures, and 'AFR' is the annual failure rate. There are 7 types of HDD models, 14 types of Seagate , 7 types of Toshiba , and 3 types of WDC , and Backblaze operates a total of 31 types of models.



In the second quarter of 2023, the cumulative number of operating days for data HDDs was 21,408,175 days, with a total of 1339 failed HDDs and an annual failure rate of 2.28%. The annual failure rate for

the first quarter of fiscal 2023 was 1.54%, and the annual failure rate for the full year of fiscal 2022 was 1.37%, so it can be said that the failure rate for this fiscal year is slightly higher.

The model with the highest annual failure rate was Toshiba's HDWF180 (8TB) with an annual failure rate of 19.63%, but this model has only 61 units in operation, and the cumulative number of operating days is statistically significant 50,000 days. Note that it is 5577 days, which is significantly lower than Seagate's ST14000NM0018 (14TB) also has a cumulative operating days of 5111 days, and HGST's HUH728080ALE604 (8TB), which has the next highest failure rate, also has a cumulative operating days of 8094 days. Seagate's ST10000NM0086 (10 TB) has the highest failure rate among models with a cumulative operating days exceeding 50,000 days, with a cumulative operating days of 100,772 days and a failure rate of 12.31%.

On the other hand, the following six data HDDs did not fail in the second quarter of 2023. Of these, Toshiba's MG08ACA16TEY (16TB), WDC's WUH721816ALE6L0 (16TB), and HGST's HUH721212ALE600 (12TB) all have significantly exceeded 50,000 cumulative operating days, which is a statistically significant standard of 50,000 days. meet. According to Backblaze, these three HDDs are highly reliable models with an annual failure rate of 0.13 to 0.45% from the start of operation, so it is no wonder that failures did not occur this quarter.



As for the reason for the high failure rate in the second quarter of 2023, Backblaze points out that there may have been many HDD failures that have been in use for a long time. The graph below shows the failure rate for the past 3 years of 'old HDD' that has passed more than 5 years since purchase, by capacity, '10 TB' shown in red and '8 TB' purple HDD It can be seen that the failure rate has increased in recent years.



Below is a graph comparing the increase in the failure rate of each capacity, with the ``increase in the annual failure rate from the first quarter to the second quarter of 2023 (0.05%)'' as the red baseline. The 8TB and 10TB models had significantly increased failure rates compared to the average and were shown to be the major contributors to the increase in failure rates in Q2.



Although Backblaze is concerned about the increase in the failure rate of data HDDs, it is not possible to escape HDD failures as long as it operates a cloud storage service, and by continuously monitoring the failure rate in this way claim to be ready. ``We will continue to monitor our systems, take action as necessary, and share with you what we can do along the way,'' he said.

in Hardware, Posted by log1h_ik