LSI 2208 and TRIM - live without?

[alexc]

Here is setup I try to do:
I have LSI 2208 raid controller with cache and bbu, and 4 SSD disks connected to it. I can not change the firmware, so HBA is not an option and I also don't want to loose cache and bbu. I would like to use RAID10 setup out of these 4 ssds.
The probles is that 2208 not aware of TRIM so I'm afraid the overall performance will become lower and lower over time (as it does on desktop ssd setup if no trim is used).
But I know I can stay on the safe side by provide extra spare space on each disk so garbage collection will be effective. Sound good, but I have no way to tell 2208 card not to use whole disks but only, say, 80% of ots space - it'll ise whole space it can see.
So how can I manage to keep raid volume performant over time, what you cat advice, please?
To be specific, I plan to put LVM over the volume and put VM disks on it.

 

[Whatever]

The only right way is to increase spare-space. Creating RAID volume leaving some space "for garbage purposes" does not work as you expect (think: are you sure that SSD'd controller understand's LSI "partitioning" semantic? I'm not so sure).
Read this: https://www.thomas-krenn.com/en/wiki/SSD_Over-provisioning_using_hdparm

However the method from the link above does not work for all SSD vendors (I've succeeded with some Intel ones)

 

[alexc]

The only right way is to increase spare-space. Creating RAID volume leaving some space "for garbage purposes" does not work as you expect (think: are you sure that SSD'd controller understand's LSI "partitioning" semantic? I'm not so sure).
Read this: https://www.thomas-krenn.com/en/wiki/SSD_Over-provisioning_using_hdparm

However the method from the link above does not work for all SSD vendors (I've succeeded with some Intel ones)

Really interesting post, but my problem is this: I can only see 'whole' volume from OS (like /dev/sdb as exported by RAID card, not one-by-one SSDs disk), so not sure if changing HPA will results in "real" disks HPA change. Moreover, my RAID reports this:

# hdparm -N /dev/sdb

/dev/sdb:
SG_IO: bad/missing sense data, sb[]:  70 00 05 00 00 00 00 0d 00 00 00 00 20 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
SG_IO: bad/missing sense data, sb[]:  70 00 05 00 00 00 00 0d 00 00 00 00 20 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
 max sectors   = 0/1, HPA is enabled

Note it says HPA is enabled, but max sectors is under questions as for me (too low value).

Disks are Intel's s3610 so the only hope is that they aware of there problems internally. Not that many RAID cards nowadays are TRIM-enabled (btw, it's already 10 years of SSD are on market), so my choice is low.

 

[Whatever]

Well, the right algorithm (from my perspective) should be:
1. Change spare-space on each SSD drive separately (with no RAID array)
2. Create RAID volume with all SSDs (with all available space) without any additional space room
3. Use it...