4 ms·
If you take say old i7 4770k's, how many of those along with how many disks would you need to get 1GB/s sustained sequential access with Ceph? My single ZFS bo
by magicalhippo 4y ago
If you take say old i7 4770k's, how many of those along with how many disks would you need to get 1GB/s sustained sequential access with Ceph?
My single ZFS box does that with ease, 3x mirrored vdevs = 6 disks total, but I'm curious as the flexibility of Ceph sounds tempting.
- antongribok 4y agoI just setup a test cluster at work to test this for you: 4 nodes, each node with 2x SAS SSDs, dual 25Gb NICs (one for front-end, one for back-end replication). The test pool is 3x replicated with Snappy compression enabled. On a separate client (also with 25Gb) I mounded an RBD image with krbd and ran FIO: fio --filename=/dev/rbd1 --direct=1 --sync=1 --rw=write --bs=4096K --numjobs=1 --iodepth=16 --ramp_time=5 --runtime=60 --ioengine=libaio --time_based --group_reporting --name=krbd-test --eta-newline=5s I get a consistent 1.4 GiB/s: write: IOPS=357, BW=1431MiB/s (1501MB/s)(83.9GiB/60036msec)
- magicalhippo 4y agoThank you very much for the test. That gives me a good idea of what I could expect from my hardware, which is not quite that beefy.