Solved SSD overheating during btrfs data scrubbing

Currently reading
Solved SSD overheating during btrfs data scrubbing

just curious how you measure the consumption because for SSD is different consumption for Read and for Write I/O (and the write is frequently 7x more than average from white papers). Don’t be driven by white papers. You can be surprised, that there isn’t such heavy difference. But you need connect measure device what can measure objective dynamic values and calculate them include Power factor to VA, instead to W.

Finally, it’s proven by science, that SSD dynamic power consumption is worse than HDD.
“Contrasting a common presumption that, SSDs can offer high performance with less power and energy consumption and at low operating temperature, our empirical evaluations reveal that dynamic power consumptions of a many-resource SSD is, on average, 5x and 4x worse than a conventional SSD and HDD, respectively.
more you can read here. Source: University of Texas.
I like science and facts.
Be sure, this is about dynamic consumption measurement, not about linear, because linear doesn’t exist in real operation.
 
Last edited:
Thanks for the article. It is comparing soms use cases that are definitely not typical ssd/Sata use is not applicable to my use case anyway.
SSD (sata at least) definitely consume less average power than HDD, the 970 NVME in my desktop is another story, the power use of that one is in the HDD range and needs active cooling. Certainly in burst mode ( dynamic load) It uses 9W max.

VA is a non SI unit and mathematically the same as watt ( Voltage*Amperage). Do not understand why power factor is better, it is used in AC not in DC.
I was not accurate using 0.27W for my SSD (= average power use) max is 1.5 v. I should also have taken typical nas drives, modern drives typically take 6-8W) in real world use. In stead of my old 7200 HGST that is running so well after all these years. Stil a factor 4-5 In power consumption separating my hdd and SSD.


I am a bit reluctant to fight a battle of the nerds here on what is more important for heating up a system: dynamic, average , max power use. Finally thermodynamics tell us energy in = energy out.
It all started with your observation that the nas switched off at 61 deg C disk temp and my assumption that that is too early :) as I did not see any signs that vurnerable parts of the nas get overheated.

thanks anyway for the nice discussion...
 
I understand your point of view. And it's right. No doubt. And don't take it as battle, this is also a platform for experiences handling.

From my (science) perspective I don't like "average" parameters, because it's inaccurate info and frequently also misleading. Specially when we don't know how was measured and by whom (tracked operation conditions and tracked tasks). And white papers don't contain such description.

So we need to specify the conditions under which electricity is consumed, which is close to the stated "average" value. Because if the disk overheats to the near-limit value (as was in this thread case), it is highly likely that the power consumption will also be pretty above average = Newton's laws.

Then we need to decide if the data Scrubbing is normal service in BTRFS. When yes, then we can take behavior of the disk drive as normal = more energy changed to the thermal energy = more electricity consumption = above the white paper conditions. There was my wording about dynamic consumption principle, confirmed by the science link above.

Another question is about the energy efficiency. When SSD overheating during the data scrubbing is really common situation (use google) in case of HDD there isn't such behavior. Finally wee need to calculate for the energy consumption also increased demand from airflow (fans) what we need to use for a normalize of the SSD temperature, what could affect rest of NAS environment.

Finally we have to take into consideration no just DC, but AC, because our bill for the NAS consumption is based on AC consumption. Then we have take into consideration also power factor of the NAS. And power factor is about energy efficiency of all NAS parts. When VA is mathematically same as W, in energy consumption case isn't same. Because Apparent power (demand power) is measured/calculated in VA. It's same as you can pay for a glass of beer = your bill is about price of the entire glass, include foam part (Reactive power) and beer part (Real power). Then VA is part of SI.
We can speak about el. energy till morning.

Enjoy your time here!
 

Create an account or login to comment

You must be a member in order to leave a comment

Create account

Create an account on our community. It's easy!

Log in

Already have an account? Log in here.

Similar threads

D
  • Solved
Awesome setup (923 over 10G that is)! BTRFS vs EXT4 is also something to consider depending on the...
Replies
3
Views
1,863
  • Locked
CLOSING THREAD, as this was resolved here: Oh, No... DSM 7.1 *REMOVED* critical SSD/NVME Cache...
Replies
1
Views
3,634
I have used a 512Gb NVMe card for cache, and soon will acquire one more. I have run cache advisor which...
Replies
0
Views
984
  • Question
You need 1 SSD if you want a Read-only cache and 2 SSD if you want a Read-write cache. The capacity of the...
Replies
1
Views
2,022
This setup only applies to the writeback cache mode. Writethrough and writearound modes store no cache...
Replies
37
Views
12,698
D
Generally from my tests, nvme with SSD array will not be super faster than a single SSD array, but in the...
Replies
3
Views
4,574
It's because there's a tradeoff. Every 1GB you add to the SSD cache utilizes 416KB of system memory to...
Replies
3
Views
3,249

Welcome to SynoForum.com!

SynoForum.com is an unofficial Synology forum for NAS owners and enthusiasts.

Registration is free, easy and fast!

Back
Top