Hi again,
We finally resized an mounted the cache on a raid0 over an vdo lvm. This where de secuential write results:
WRITE: bw=247MiB/s (259MB/s), 247MiB/s-247MiB/s (259MB/s-259MB/s), io=4947MiB (5187MB), run=20004-20004msec
While directly on the raid0 being used as cache we got this:
WRITE: bw=1610MiB/s (1688MB/s), 1610MiB/s-1610MiB/s (1688MB/s-1688MB/s), io=13.6GiB (14.6GB), run=8661-8661msec
The cache was created like this:
wbcreate cache /dev/mapper/vgtest-vdo_test /dev/md999p1 --reformat --writeback_threshold=70 --read_cache_threshold=127
Where md999p1 is a 800G partition and underlying vdo is of 4T and virtually 500T (that's what writeboost will see, just doing tests, don't expect that much deduplication).
Is this difference expected?
This is the dmsetup:
cache: 0 1073741824000 writeboost 0 213042754 1677502 1 0 0 0 0 261 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 10 writeback_threshold 70 nr_cur_batched_writeback 0 sync_data_interval 0 update_sb_record_interval 0 read_cache_threshold 0
``
Hi again,
We finally resized an mounted the cache on a raid0 over an vdo lvm. This where de secuential write results:
While directly on the raid0 being used as cache we got this:
The cache was created like this:
Where md999p1 is a 800G partition and underlying vdo is of 4T and virtually 500T (that's what writeboost will see, just doing tests, don't expect that much deduplication).
Is this difference expected?
This is the dmsetup: