Creating an XFS filesystem can use a partition or the logical device directly.
mkfs.xfs /dev/sda1 mkfs.xfs /dev/sdb
Do a dry-run and see what filesystem attributes it would set:
mkfs.xfs -N /dev/sda1
If the device already has another filesystem, mkfs will stop. You can safely force it to run:
mkfs.xfs -Nf /dev/sda1
For the sake of clarity (and not making mistakes) when using additional arguments, use mkfs.xfs -d name=/dev/sda syntax when creating filesystems instead of mkfs.xfs /dev/sda
The journal log can be put on a separate device to improve performance.
mkfs.xfs -d name=/dev/sda -l logdev=/dev/sdb mount -o logdev=/dev/sdb /dev/sda /mnt/xfs
In addition to having the journal log on a separate device, the metadata and file data can be split up as well.
Create the new XFS filesystem and separate log journal at the same time:
mkfs.xfs -d name=/dev/sda -l logdev=/dev/sdb -r rtdev=/dev/sdc mount -o logdev=/dev/sdb,rtdev=/dev/sdc /dev/sda /mnt/xfs
In both cases, the logdev option will want a block size. See man mkfs.xfs for syntax, suggestions.
Alternatively, see what mkfs.xfs would use by default, and use that:
mkfs.xfs -Nf -d name=/dev/sda
meta-data=/dev/sda isize=256 agcount=4, agsize=524288 blks
= sectsz=512 attr=2, projid32bit=1
= crc=0 finobt=0
data = bsize=4096 blocks=2097152, imaxpct=25
= sunit=0 swidth=0 blks
naming =version 2 bsize=4096 ascii-ci=0 ftype=0
log =internal log bsize=4096 blocks=2560, version=2
= sectsz=512 sunit=0 blks, lazy-count=1
realtime =none extsz=4096 blocks=0, rtextents=0
mkfs.xfs -d name=/dev/sda -l logdev=/dev/sdb,size=2560b