There is a result of the analysis of two weeks ago

Old sda screenshot

and now

New sda screenshot

It is seen that the average speed decreased from 169.7 to 166.2, and the access time increased from 14.57 to 14.98. If you put screenshots on top of one another (the red line and the blue dots - the old graph, the gray line and the green dots - the new one), then you can see that the reduction in speed is in several specific areas (the piece in the area of ​​80% - 95% is alarming), and the large part of the graph is the same:

Screenshot overlay - sda differences

Since the main doubts are caused by the end of the disk, I took a measurement on the sda9 section (there are also measurements for two weeks ago in all sections separately). I draw your attention that on the first screenshot the maximum of the speed scale is 180, and on the second 170, but the maximums on the time scale are the same (50). On the combined screenshot, this is taken into account.

Old sda9 screenshot

New sda9 screenshot

Screenshots overlay - differences sda9

Questions

  • Should these changes be considered simply as a statistical error, or do they show a worsening of the condition?
  • What generally deviations are normal for a statistical error?
  • What may indicate a deterioration in the performance of the disk in the area closer to its end?
  • Does this look like a reduction in head positioning accuracy?
  • What is the state of the disk?
  • If something happens to him, then what will happen to the boot area in the most likely way? That is, everything will be covered at the same time, or will it be among the last to fly, or the first?

PS: During both measurements the disk was not mounted and was not used differently.

  • I think that on two tests it is difficult to build statistics and some theory. Perform the test N times (with the same break in 2 weeks). - Alexey Sarovsky
  • @ AlekseySarovsky, in general, I agree, so I hoped that some standard errors were known so that you could compare them. And then the measurements were not just made. - Qwertiy ♦
  • one
    Yes, it is the same disk. But where is the guarantee that you provided the same testing conditions? What was at least the same core and the same background activity? Run this utility ten to twenty times in a row and adjust. There will be a similar situation, but also in the direction of improvement. But this does not mean that your disk began to heal. - KoVadim
  • one
    I do not know how 10 percent, but I think that 5 is fully justified. If you need more, it is better to find the documentation for your disk and test it according to it. - KoVadim
  • one
    @KoVadim is right. Measure 10 times in a row. If the measurement of two weeks ago will be close to one of these, then everything is fine with the disk. Otherwise, the question remains open, since it is unlikely to succeed in exactly reproducing the conditions of 2 weeks ago. - avp

0