E
E
Evgeny Zalivadny2020-12-22 11:41:33
linux
Evgeny Zalivadny, 2020-12-22 11:41:33

Is it right to mount /var on HDD instead of SSD so it will last longer?

When installing Linux, is it rational to specify a mount point for the /var partition on the HDD (if available), instead of mounting on the SSD in order to not write logs to the SSD and thereby increase its lifespan? Perhaps there are other partitions that should be mounted on the HDD?

Answer the question

In order to leave comments, you need to log in

7 answer(s)
A
Armenian Radio, 2020-12-22
@gbg

No, all of this "SSD life saving" advice applies to the earliest generations of drives. Modern drives will become morally obsolete much earlier than the moment when their write resource is exhausted.

K
ky0, 2020-12-22
@ky0

Almost always - this is saving on matches. Except, of course, when you have a really healthy stream of logs - but then they are usually not stored locally, but centrally.

A
Adamos, 2020-12-22
@Adamos

If you care about optimization and don't care about logs, then /tmp and /var/log are perfectly mounted into memory (tmpfs). The downside is that you won't be able to see what was in the logs before the reboot. Plus - the costs will rush to zero (by mounting /var on the HDD, you will noticeably slow down access to it for writing).

A
Artem @Jump, 2020-12-22
Tag curated by

  • Firstly, it is not clear what the service life has to do with it? What side is he?
  • Secondly, if you decide to separate partitions across different disks, then the principle is simple - all data that you often work with is on SSDs, which are rarely accessed on HDDs. Since /var is frequently changed data - its out of the question on an SSD.

P
Puma Thailand, 2020-12-22
@opium

Not anymore

V
Victor Taran, 2020-12-22
@shambler81

NO, this is no longer relevant
. Trim technology is already on all disks, their wear is no less than that of hard ones, and sometimes even more.
Children's sores ssd already asked

V
Vladimir Korotenko, 2020-12-24
@firedragon

It all depends on the loads
Let's say for a HEAVY DB server
I would do the following
RAID 1 - OS
RAID 10 - database data
RAID 10 - database logs
Application server (all in one)
RAID 1 - OS
RAID 10 - database + logs + data
Application server (web muzzle)
RAID 1 - OS - data
A separate issue is ssd accelerators and nvme cache
Look, all this is present and allows you to keep the budget within reasonable limits.
In the end, a lot of IOPS and random reading is not always necessary.
In general, make a plan for resource consumption, and choose everything according to this plan.

Didn't find what you were looking for?

Ask your question

Ask a Question

731 491 924 answers to any question