After four months, the DSS began to show signs of fatigue. Let's recap. In December 2010, we decided to test the lifetime of an SSD, to see what was happening exactly when a SSD comes to the limits of its flash memory. We have therefore launched a test: write e continuously on a DSS based on a controller and memory Indilinx Samsung MLC, given to 10,000 write cycles.
Yesterday, the DSS has reached 10,000 cycles on the most worn his cell - a value returned by the SMART technology - whereas the average wear was announced at 8473 cycles. This morning we restarted the computer to check a detail, and Windows has launched an audit of storage media, finding errors.
Result of the first audit: 4229 Rated as bad clusters and MFT (table of contents to NTFS, simplifying) corrupt. We re a second check after formatting the SSD: 12 927 bad clusters. Lacona We then test our classic, and we stop after copying about 600 GB of SSD. Result of third test: 540 bad clusters.
As seen, the SSD really starts to die, and to repeat the mistakes are there to prove it. In conventional use, a storage medium that errors of this type would leave immediately after-sales service, do not play with fire and its data. In our case, we decided to continue writing on the SSD and periodically noted the number of errors.
What will happen when all cells exceed 10 000 cycles? The paris open. 
Yesterday, the DSS has reached 10,000 cycles on the most worn his cell - a value returned by the SMART technology - whereas the average wear was announced at 8473 cycles. This morning we restarted the computer to check a detail, and Windows has launched an audit of storage media, finding errors.
Result of the first audit: 4229 Rated as bad clusters and MFT (table of contents to NTFS, simplifying) corrupt. We re a second check after formatting the SSD: 12 927 bad clusters. Lacona We then test our classic, and we stop after copying about 600 GB of SSD. Result of third test: 540 bad clusters.
As seen, the SSD really starts to die, and to repeat the mistakes are there to prove it. In conventional use, a storage medium that errors of this type would leave immediately after-sales service, do not play with fire and its data. In our case, we decided to continue writing on the SSD and periodically noted the number of errors.
What will happen when all cells exceed 10 000 cycles? The paris open.
No comments:
Post a Comment