{ "version": "https://jsonfeed.org/version/1", "title": "Ceph", "description": "Ceph is a massively scalable, open source, distributed storage system.", "home_page_url": "go/ceph", "feed_url": "feed/ceph.json", "icon": "https://cdn.v2ex.com/navatar/a532/400e/637_large.png?m=1355371763", "favicon": "https://cdn.v2ex.com/navatar/a532/400e/637_normal.png?m=1355371763", "items": [ { "author": { "url": "member/c1462066778", "name": "c1462066778", "avatar": "https://cdn.v2ex.com/gravatar/8480f99b17957a3c303164987fc10c68?s=73&d=retro" }, "url": "t/912582", "title": "Ceph \u8bfb\u5199\u6027\u80fd\u95ee\u9898\uff0c\u8bfb\u5165\u5feb\uff0c\u5199\u51fa\u6162", "id": "t/912582", "date_published": "2023-02-02T06:43:42+00:00", "content_html": "

Ceph \u8bfb\u5199\u6027\u80fd\u95ee\u9898\uff0c\u8bfb\u5165\u5feb\uff0c\u5199\u51fa\u6162

\n

\u5927\u5bb6\u597d\uff0c\u6211\u4eec\u9700\u8981\u5c06\u6240\u6709\u4e91\u73af\u5883\u8fc1\u79fb\u5230 Proxmox \u3002\u76ee\u524d\u6211\u6b63\u5728\u8bc4\u4f30\u6d4b\u8bd5 Proxmox+Ceph+OpenStack \u3002

\n

\u4f46\u662f\u73b0\u5728\u9047\u5230\u4ee5\u4e0b\u56f0\u96be\uff1a

\n
    \n
  1. VMware vSAN \u8fc1\u79fb\u5230 ceph \u65f6\uff0c\u6211\u53d1\u73b0 hdd+ssd \u5728 ceph \u4e2d\u7684\u8868\u73b0\u975e\u5e38\u7cdf\u7cd5\uff0c\u5e76\u4e14\u5199\u6027\u80fd\u975e\u5e38\u5dee\u3002\u6027\u80fd\u8fdc\u4e0d\u53ca vSAN
  2. \n
  3. \u5168\u95ea\u5b58\u7ed3\u6784\u4e2d\u7684 ceph \u987a\u5e8f\u5199\u5165\u6027\u80fd\u8fd8\u4e0d\u5982\u5355\u5757\u786c\u76d8\uff0c\u751a\u81f3\u4e0d\u5982\u5355\u5757\u673a\u68b0\u786c\u76d8
  4. \n
  5. \u5728\u4f7f\u7528 bcache \u4e2d\u7684 hdd+ssd \u7ed3\u6784\u4e2d\uff0cceph \u7684\u987a\u5e8f\u5199\u5165\u6027\u80fd\u8fdc\u4f4e\u4e8e\u5355\u5757\u786c\u76d8\u5199\u5165
  6. \n
\n

\u6d4b\u8bd5\u670d\u52a1\u5668\u53c2\u6570\uff08\u8fd9\u4e0d\u91cd\u8981\uff09\uff1a

\n

CPU\uff1aDual Intel\u00ae Xeon\u00ae E5-2698Bv3

\n

Memory\uff1a8 x 16G DDR3

\n

Dual 1 Gbit NIC\uff1aRealtek Semiconductor Co., Ltd. RTL8111/8168/8411

\n

Disk\uff1a

\n

1 x 500G NVME SAMSUNG MZALQ512HALU-000L1 (\u540c\u65f6\u4e5f\u662f PVE \u4e2d\u7684 ssd-data Thinpool)

\n

1 x 500G SATA WDC_WD5000AZLX-60K2TA0 (\u7269\u7406\u673a\u7cfb\u7edf\u76d8)

\n

2 x 500G SATA WDC_WD5000AZLX-60K2TA0

\n

1 x 1T SATA ST1000LM035-1RK172

\n

PVE\uff1apve-manager/7.3-4/d69b70d4 (running kernel: 5.15.74-1-pve)

\n

Network Configure\uff1a

\n

enp4s0 (OVS Port) -> vmbr0 (OVS Bridge) -> br0mgmt (192.168.1.3/24,192.168.1.1)

\n

enp5s0 (OVS Port,MTU=9000) -> vmbr1 (OVS Bridge,MTU=9000)

\n

vmbr2 (OVS Bridge,MTU=9000)

\n

\u6d4b\u8bd5\u865a\u62df\u673a\u53c2\u6570 x 3 (\u4e09\u53f0\u865a\u62df\u673a\u662f\u4e00\u6837\u7684\u53c2\u6570)\uff1a

\n

CPU\uff1a32 (1 sockets, 32 cores) [host]

\n

Memory\uff1a32G

\n

Disk\uff1a

\n

1 x local-lvm:vm-101-disk-0,iothread=1,size=32G

\n

2 x ssd-data:vm-101-disk-0,iothread=1,size=120G

\n

Network Device\uff1a

\n

net0: bridge=vmbr0,firewall=1

\n

net1: bridge=vmbr2,firewall=1,mtu=1 (Ceph Cluster/Public Network)

\n

net2: bridge=vmbr0,firewall=1

\n

net3: bridge=vmbr0,firewall=1

\n

Network Configure\uff1a

\n

ens18 (net0,OVS Port) -> vmbr0 (OVS Bridge) -> br0mgmt (10.10.1.11/24,10.10.1.1)

\n

ens19 (net1,OVS Port,MTU=9000) -> vmbr1 (OVS Bridge,MTU=9000) -> br1ceph (192.168.10.1/24,MTU=9000)

\n

ens20 (net2,Network Device,Active=No)

\n

ens21 (net3,Network Device,Active=No)

\n

\u57fa\u51c6\u6d4b\u8bd5\u5de5\u5177:

\n
    \n
  1. fio
  2. \n
  3. fio-cdm ( https://github.com/xlucn/fio-cdm)
  4. \n
\n

\u5bf9\u4e8e fio-cdm \uff0c\u5982\u679c\u4e0d\u586b\u5199\u4efb\u4f55\u53c2\u6570\uff0c\u90a3\u4e48\u5bf9\u5e94\u4e8e fio \u7684\u914d\u7f6e\u6587\u4ef6\u5982\u4e0b

\n

\u4f7f\u7528 python fio-cdm -f - \u53ef\u4ee5\u5f97\u5230

\n
[global]\nioengine=libaio\nfilename=.fio_testmark\ndirectory=/root\nsize=1073741824.0\ndirect=1\nruntime=5\nrefill_buffers\nnorandommap\nrandrepeat=0\nallrandrepeat=0\ngroup_reporting\n\n[seq-read-1m-q8-t1]\nrw=read\nbs=1m\nrwmixread=0\niodepth=8\nnumjobs=1\nloops=5\nstonewall\n\n[seq-write-1m-q8-t1]\nrw=write\nbs=1m\nrwmixread=0\niodepth=8\nnumjobs=1\nloops=5\nstonewall\n\n[seq-read-1m-q1-t1]\nrw=read\nbs=1m\nrwmixread=0\niodepth=1\nnumjobs=1\nloops=5\nstonewall\n\n[seq-write-1m-q1-t1]\nrw=write\nbs=1m\nrwmixread=0\niodepth=1\nnumjobs=1\nloops=5\nstonewall\n\n[rnd-read-4k-q32-t16]\nrw=randread\nbs=4k\nrwmixread=0\niodepth=32\nnumjobs=16\nloops=5\nstonewall\n\n[rnd-write-4k-q32-t16]\nrw=randwrite\nbs=4k\nrwmixread=0\niodepth=32\nnumjobs=16\nloops=5\nstonewall\n\n[rnd-read-4k-q1-t1]\nrw=randread\nbs=4k\nrwmixread=0\niodepth=1\nnumjobs=1\nloops=5\nstonewall\n\n[rnd-write-4k-q1-t1]\nrw=randwrite\nbs=4k\nrwmixread=0\niodepth=1\nnumjobs=1\nloops=5\nstonewall\n
\n

\u73af\u5883\u6784\u5efa\u6b65\u9aa4:

\n
# prepare tools\nroot@pve01:~# apt update -y && apt upgrade -y\nroot@pve01:~# apt install fio git -y\nroot@pve01:~# git clone https://github.com/xlucn/fio-cdm.git\n\n# create test block\nroot@pve01:~# rbd create test -s 20G\nroot@pve01:~# rbd map test\nroot@pve01:~# mkfs.xfs /dev/rbd0\nroot@pve01:~# mkdir /mnt/test\nroot@pve01:/mnt# mount /dev/rbd0 /mnt/test\n\n# start test\nroot@pve01:/mnt/test# python3 ~/fio-cdm/fio-cdm\n
\n

\u73af\u5883\u6d4b\u8bd5:

\n
    \n
  1. Network Bandwidth
  2. \n
\n
root@pve01:~# apt install iperf3 -y\nroot@pve01:~# iperf3 -s\n-----------------------------------------------------------\nServer listening on 5201\n-----------------------------------------------------------\nAccepted connection from 10.10.1.12, port 52968\n[ 5] local 10.10.1.11 port 5201 connected to 10.10.1.12 port 52972\n[ ID] Interval Transfer Bitrate\n[ 5] 0.00-1.00 sec 1.87 GBytes 16.0 Gbits/sec \n[ 5] 1.00-2.00 sec 1.92 GBytes 16.5 Gbits/sec \n[ 5] 2.00-3.00 sec 1.90 GBytes 16.4 Gbits/sec \n[ 5] 3.00-4.00 sec 1.90 GBytes 16.3 Gbits/sec \n[ 5] 4.00-5.00 sec 1.85 GBytes 15.9 Gbits/sec \n[ 5] 5.00-6.00 sec 1.85 GBytes 15.9 Gbits/sec \n[ 5] 6.00-7.00 sec 1.70 GBytes 14.6 Gbits/sec \n[ 5] 7.00-8.00 sec 1.75 GBytes 15.0 Gbits/sec \n[ 5] 8.00-9.00 sec 1.89 GBytes 16.2 Gbits/sec \n[ 5] 9.00-10.00 sec 1.87 GBytes 16.0 Gbits/sec \n[ 5] 10.00-10.04 sec 79.9 MBytes 15.9 Gbits/sec \n- - - - - - - - - - - - - - - - - - - - - - - - -\n[ ID] Interval Transfer Bitrate\n[ 5] 0.00-10.04 sec 18.6 GBytes 15.9 Gbits/sec receiver\n\n
\n
    \n
  1. Jumbo Frames
  2. \n
\n
root@pve01:~# ping -M do -s 8000 192.168.10.2\nPING 192.168.10.2 (192.168.10.2) 8000(8028) bytes of data.\n8008 bytes from 192.168.10.2: icmp_seq=1 ttl=64 time=1.51 ms\n8008 bytes from 192.168.10.2: icmp_seq=2 ttl=64 time=0.500 ms\n^C\n--- 192.168.10.2 ping statistics ---\n2 packets transmitted, 2 received, 0% packet loss, time 1002ms\nrtt min/avg/max/mdev = 0.500/1.007/1.514/0.507 ms\nroot@pve01:~# \n
\n

\u57fa\u51c6\u6d4b\u8bd5\u5206\u7c7b:

\n
    \n
  1. Physical Disk Benchmark \uff0c\u7269\u7406\u78c1\u76d8\u57fa\u51c6\u6d4b\u8bd5
  2. \n
  3. Single osd, single server benchmark \uff0c\u5355\u4e2a OSD \u5355\u4e2a\u670d\u52a1\u5668\u57fa\u51c6\u6d4b\u8bd5
  4. \n
  5. Multiple OSDs, single server benchmarks \uff0c\u591a\u4e2a OSD \u5355\u4e2a\u670d\u52a1\u5668\u57fa\u51c6\u6d4b\u8bd5
  6. \n
  7. Multiple OSDs, multiple server benchmarks \uff0c\u591a\u4e2a OSD \u591a\u4e2a\u670d\u52a1\u5668\u57fa\u51c6\u6d4b\u8bd5
  8. \n
\n

\u57fa\u51c6\u6d4b\u8bd5\u7ed3\u679c\uff08 Ceph \u548c\u7cfb\u7edf\u6ca1\u6709\u8fdb\u884c\u8fc7\u4efb\u4f55\u8c03\u4f18\uff0c\u6ca1\u6709\u4f7f\u7528 bcache \u52a0\u901f\uff09

\n

Benchmark Result (Ceph and the system have not been tuned or bcache accelerated. ):

\n

1. Physical Disk Benchmark (Test sequence is 4)\uff08\u6d4b\u8bd5\u987a\u5e8f\u662f 4 \uff09

\n

step.

\n
root@pve1:~# lsblk\nNAME MAJ:MIN RM SIZE RO TYPE MOUNTPOINT\nsda 8:0 0 465.8G 0 disk \n\u251c\u2500sda1 8:1 0 1007K 0 part \n\u251c\u2500sda2 8:2 0 512M 0 part /boot/efi\n\u2514\u2500sda3 8:3 0 465.3G 0 part \n \u251c\u2500pve-root 253:0 0 96G 0 lvm /\n \u251c\u2500pve-data_tmeta 253:1 0 3.5G 0 lvm \n \u2502 \u2514\u2500pve-data-tpool 253:3 0 346.2G 0 lvm \n \u2502 \u251c\u2500pve-data 253:4 0 346.2G 1 lvm \n \u2502 \u2514\u2500pve-vm--100--disk--0 253:5 0 16G 0 lvm \n \u2514\u2500pve-data_tdata 253:2 0 346.2G 0 lvm \n \u2514\u2500pve-data-tpool 253:3 0 346.2G 0 lvm \n \u251c\u2500pve-data 253:4 0 346.2G 1 lvm \n \u2514\u2500pve-vm--100--disk--0 253:5 0 16G 0 lvm \nsdb 8:16 0 931.5G 0 disk \nsdc 8:32 0 465.8G 0 disk \nsdd 8:48 0 465.8G 0 disk \nnvme0n1 259:0 0 476.9G 0 disk \nroot@pve1:~# mkfs.xfs /dev/nvme0n1 -f\nroot@pve1:~# mkdir /mnt/nvme\nroot@pve1:~# mount /dev/nvme0n1 /mnt/nvme\nroot@pve1:~# cd /mnt/nvme/\n
\n

result.

\n
root@pve1:/mnt/nvme# python3 ~/fio-cdm/fio-cdm\ntests: 5, size: 1.0GiB, target: /mnt/nvme 3.4GiB/476.7GiB\n|Name | Read(MB/s)| Write(MB/s)|\n|------------|------------|------------|\n|SEQ1M Q8 T1 | 2361.95| 1435.48|\n|SEQ1M Q1 T1 | 1629.84| 1262.63|\n|RND4K Q32T16| 954.86| 1078.88|\n|. IOPS | 233119.53| 263398.08|\n|. latency us| 2194.84| 1941.78|\n|RND4K Q1 T1 | 55.56| 225.06|\n|. IOPS | 13565.49| 54946.21|\n|. latency us| 72.76| 16.97|\n\n
\n

2. Single osd, single server benchmark (Test sequence is 3)\uff08\u6d4b\u8bd5\u987a\u5e8f\u662f 3 \uff09

\n

\u4fee\u6539 ceph.conf \u4e2dosd_pool_default_min_size\u548cosd_pool_default_size\u4e3a 1 \uff0c\u7136\u540esystemctl restart ceph.target\u5e76\u4fee\u590d\u6240\u6709\u62a5\u9519

\n

step.

\n
root@pve01:/mnt/test# ceph osd pool get rbd size\nsize: 2\nroot@pve01:/mnt/test# ceph config set global mon_allow_pool_size_one true\nroot@pve01:/mnt/test# ceph osd pool set rbd min_size 1\nset pool 2 min_size to 1\nroot@pve01:/mnt/test# ceph osd pool set rbd size 1 --yes-i-really-mean-it\nset pool 2 size to 1\n\n
\n

result

\n
root@pve01:/mnt/test# ceph -s\n cluster:\n id: 1f3eacc8-2488-4e1a-94bf-7181ee7db522\n health: HEALTH_WARN\n 2 pool(s) have no replicas configured\n \n services:\n mon: 3 daemons, quorum pve01,pve02,pve03 (age 17m)\n mgr: pve01(active, since 17m), standbys: pve02, pve03\n osd: 6 osds: 1 up (since 19s), 1 in (since 96s)\n \n data:\n pools: 2 pools, 33 pgs\n objects: 281 objects, 1.0 GiB\n usage: 1.1 GiB used, 119 GiB / 120 GiB avail\n pgs: 33 active+clean\n \nroot@pve01:/mnt/test# ceph osd tree\nID CLASS WEIGHT TYPE NAME STATUS REWEIGHT PRI-AFF\n-1 0.70312 root default \n-3 0.23438 host pve01 \n 0 ssd 0.11719 osd.0 up 1.00000 1.00000\n 1 ssd 0.11719 osd.1 down 0 1.00000\n-5 0.23438 host pve02 \n 2 ssd 0.11719 osd.2 down 0 1.00000\n 3 ssd 0.11719 osd.3 down 0 1.00000\n-7 0.23438 host pve03 \n 4 ssd 0.11719 osd.4 down 0 1.00000\n 5 ssd 0.11719 osd.5 down 0 1.00000\nroot@pve01:/mnt/test# python3 ~/fio-cdm/fio-cdm\ntests: 5, size: 1.0GiB, target: /mnt/test 175.8MiB/20.0GiB\n|Name | Read(MB/s)| Write(MB/s)|\n|------------|------------|------------|\n|SEQ1M Q8 T1 | 1153.07| 515.29|\n|SEQ1M Q1 T1 | 447.35| 142.98|\n|RND4K Q32T16| 99.07| 32.19|\n|. IOPS | 24186.26| 7859.91|\n|. latency us| 21148.94| 65076.23|\n|RND4K Q1 T1 | 7.47| 1.48|\n|. IOPS | 1823.24| 360.98|\n|. latency us| 545.98| 2765.23|\nroot@pve01:/mnt/test# \n\n
\n

3. Multiple OSDs, single server benchmarks (Test sequence is 2)\uff08\u6d4b\u8bd5\u987a\u5e8f\u662f 2 \uff09

\n

\u4fee\u6539 crushmap \u4e2dstep chooseleaf firstn 0 type host\uff0c\u5c06host\u4fee\u6539\u4e3aosd

\n

OSD tree

\n
root@pve01:/etc/ceph# ceph osd tree\nID CLASS WEIGHT TYPE NAME STATUS REWEIGHT PRI-AFF\n-1 0.70312 root default \n-3 0.23438 host pve01 \n 0 ssd 0.11719 osd.0 up 1.00000 1.00000\n 1 ssd 0.11719 osd.1 up 1.00000 1.00000\n-5 0.23438 host pve02 \n 2 ssd 0.11719 osd.2 down 0 1.00000\n 3 ssd 0.11719 osd.3 down 0 1.00000\n-7 0.23438 host pve03 \n 4 ssd 0.11719 osd.4 down 0 1.00000\n 5 ssd 0.11719 osd.5 down 0 1.00000\n\n
\n

result

\n
root@pve01:/mnt/test# python3 ~/fio-cdm/fio-cdm\ntests: 5, size: 1.0GiB, target: /mnt/test 175.8MiB/20.0GiB\n|Name | Read(MB/s)| Write(MB/s)|\n|------------|------------|------------|\n|SEQ1M Q8 T1 | 1376.59| 397.29|\n|SEQ1M Q1 T1 | 442.74| 111.41|\n|RND4K Q32T16| 114.97| 29.08|\n|. IOPS | 28068.12| 7099.90|\n|. latency us| 18219.04| 72038.06|\n|RND4K Q1 T1 | 6.82| 1.04|\n|. IOPS | 1665.27| 254.40|\n|. latency us| 598.00| 3926.30|\n\n
\n

4. Multiple OSDs, multiple server benchmarks (Test sequence is 1)\uff08\u6d4b\u8bd5\u987a\u5e8f\u662f 1 \uff09

\n

OSD tree

\n
root@pve01:/etc/ceph# ceph osd tree\nID CLASS WEIGHT TYPE NAME STATUS REWEIGHT PRI-AFF\n-1 0.70312 root default \n-3 0.23438 host pve01 \n 0 ssd 0.11719 osd.0 up 1.00000 1.00000\n 1 ssd 0.11719 osd.1 up 1.00000 1.00000\n-5 0.23438 host pve02 \n 2 ssd 0.11719 osd.2 up 1.00000 1.00000\n 3 ssd 0.11719 osd.3 up 1.00000 1.00000\n-7 0.23438 host pve03 \n 4 ssd 0.11719 osd.4 up 1.00000 1.00000\n 5 ssd 0.11719 osd.5 up 1.00000 1.00000\n
\n

result

\n
tests: 5, size: 1.0GiB, target: /mnt/test 175.8MiB/20.0GiB\n|Name | Read(MB/s)| Write(MB/s)|\n|------------|------------|------------|\n|SEQ1M Q8 T1 | 1527.37| 296.25|\n|SEQ1M Q1 T1 | 408.86| 106.43|\n|RND4K Q32T16| 189.20| 43.00|\n|. IOPS | 46191.94| 10499.01|\n|. latency us| 11068.93| 48709.85|\n|RND4K Q1 T1 | 4.99| 0.95|\n|. IOPS | 1219.16| 232.37|\n|. latency us| 817.51| 4299.14|\n\n
\n

\u7ed3\u8bba

\n
    \n
  1. \u53ef\u4ee5\u770b\u5230 ceph \u7684\u5199\u5165\u6027\u80fd\uff08 106.43MB/s \uff09\u4e0e\u7269\u7406\u78c1\u76d8\u7684\u5199\u5165\u6027\u80fd\uff08 1262.63MB/s \uff09\u4e4b\u95f4\u7684\u5dee\u8ddd\u662f\u5de8\u5927\u7684\uff0c\u751a\u81f3 RND4K Q1 T1 \u7684\u60c5\u51b5\u4e0b\u76f4\u63a5\u53d8\u6210\u4e86\u673a\u68b0\u786c\u76d8
  2. \n
  3. \u4e00\u4e2a\u6216\u8005\u591a\u4e2a osd \u4ee5\u53ca\u4e00\u4e2a\u6216\u8005\u591a\u4e2a\u673a\u5668\u5bf9 ceph \u7684\u5f71\u54cd\u5e76\u4e0d\u5927\uff08\u53ef\u80fd\u662f\u6211\u7684\u96c6\u7fa4\u6570\u91cf\u4e0d\u591f\uff09
  4. \n
  5. \u4e09\u4e2a\u8282\u70b9\u6784\u5efa\u7684 ceph \u96c6\u7fa4\uff0c\u4f1a\u5bfc\u81f4\u78c1\u76d8 read \u6027\u80fd\u4f1a\u4e0b\u964d\u5230\u539f\u6765\u7684\u4e00\u534a\uff0cwrite \u6027\u80fd\u4e0b\u964d\u5230\u539f\u6765\u7684\u56db\u5206\u4e4b\u4e00\u751a\u81f3\u66f4\u591a
  6. \n
\n

\u9644\u5f55

\n

\u4e00\u4e9b ssd \u57fa\u51c6\u6d4b\u8bd5\u7ed3\u679c

\n
Micron_1 100_MTFDDAK1T0TB SCSI Disk Device
\n
G:\\fio>python \"E:\\Programing\\PycharmProjects\\fio-cdm\\fio-cdm\"\ntests: 5, size: 1.0GiB, target: G:\\fio 228.2GiB/953.8GiB\n|Name | Read(MB/s)| Write(MB/s)|\n|------------|------------|------------|\n|SEQ1M Q8 T1 | 363.45| 453.54|\n|SEQ1M Q1 T1 | 329.47| 404.09|\n|RND4K Q32T16| 196.16| 212.42|\n|. IOPS | 47890.44| 51861.48|\n|. latency us| 10677.71| 9862.74|\n|RND4K Q1 T1 | 20.66| 65.44|\n|. IOPS | 5044.79| 15976.40|\n|. latency us| 197.04| 61.07|\n
\n
SAMSUNG MZALQ512HALU-000L1
\n
root@pve1:/mnt/test# python3 ~/fio-cdm/fio-cdm\ntests: 5, size: 1.0GiB, target: /mnt/test 3.4GiB/476.7GiB\n|Name | Read(MB/s)| Write(MB/s)|\n|------------|------------|------------|\n|SEQ1M Q8 T1 | 2358.84| 1476.54|\n|SEQ1M Q1 T1 | 1702.19| 1291.18|\n|RND4K Q32T16| 955.34| 1070.17|\n|. IOPS | 233238.46| 261273.09|\n|. latency us| 2193.90| 1957.79|\n|RND4K Q1 T1 | 55.04| 229.99|\n|. IOPS | 13437.11| 56149.97|\n|. latency us| 73.17| 16.65|\n
\n

\u200b

\n

bcache

\n

\u4f7f\u7528 bcache \u52a0\u901f\u540e\u7684 hdd+ssd \u6df7\u5408\u78c1\u76d8 ceph \u67b6\u6784\u7684\u6d4b\u8bd5\u7ed3\u679c

\n

\u53ef\u4ee5\u770b\u5230 read \u6709\u660e\u663e\u63d0\u5347\uff0c\u4f46\u662f write \u4ecd\u7136\u975e\u5e38\u5dee\u52b2

\n
tests: 5, size: 1.0GiB, target: /mnt/test 104.3MiB/10.0GiB\n|Name | Read(MB/s)| Write(MB/s)|\n|------------|------------|------------|\n|SEQ1M Q8 T1 | 1652.93| 242.41|\n|SEQ1M Q1 T1 | 552.91| 81.16|\n|RND4K Q32T16| 429.52| 31.95|\n|. IOPS | 104862.76| 7799.72|\n|. latency us| 4879.87| 65618.50|\n|RND4K Q1 T1 | 13.10| 0.45|\n|. IOPS | 3198.16| 110.09|\n|. latency us| 310.07| 9077.11|\n
\n

\u5373\u4fbf\u662f\u4e00\u5757\u78c1\u76d8\u4e0a\u591a\u4e2a osd \u4e5f\u65e0\u6cd5\u89e3\u51b3 write \u95ee\u9898

\n

\u8be6\u7ec6\u6d4b\u8bd5\u6570\u636e\uff1a https://www.reddit.com/r/ceph/comments/xnse2j/comment/j6qs57g/?cOntext=3

\n

\u5982\u679c\u4f7f\u7528 VMware vSAN \uff0c\u53ef\u4ee5\u5f88\u8f7b\u677e\u7684\u8ba9 hdd \u52a0\u901f\u5230 ssd \u7684\u901f\u5ea6\uff0c\u800c\u4e14\u51e0\u4e4e\u611f\u77e5\u4e0d\u5230 hdd \u7684\u5b58\u5728\uff08\u5e76\u672a\u8be6\u7ec6\u5bf9\u6bd4\uff0c\u6211\u53ea\u662f\u51ed\u611f\u89c9\u7684\uff09

\n

\u5176\u4ed6\u4e13\u4e1a\u7684\u6d4b\u8bd5\u62a5\u544a\u5206\u6790

\n

\u6211\u5206\u6790\u6bd4\u8f83\u4e86\u51e0\u4e2a\u62a5\u544a\uff0c\u6458\u8981\u5982\u4e0b

\n

Proxmox-VE_Ceph-Benchmark-201802.pdf

\n

Proxmox-VE_Ceph-Benchmark-202009-rev2.pdf

\n

Dell_R730xd_RedHat_Ceph_Performance_SizingGuide_WhitePaper.pdf

\n

micron_9300_and_red_hat_ceph_reference_architecture.pdf

\n
pve 201802
\n

\u4ece\u62a5\u544a\u4e2d\u5f97\u77e5\uff0c\u6d4b\u8bd5\u89c4\u6a21\u4e3a 6 x Server \uff0cEach server 4 x Samsung SM863 Series, 2.5\", 240 GB SSD, SATA-3 (6 Gb/s) MLC.

\n
# Samsung SM863 Series, 2.5\", 240 GB SSD\n# from https://www.samsung.com/us/business/support/owners/product/sm863-series-240gb/\n|Name | Read(MB/s)| Write(MB/s)|\n|------------|------------|------------|\n|SEQ?M Q? T? | 520.00| 485.00|\n|RND4K Q? T? | ?| ?|\n|. IOPS | 97000.00| 20000.00|\n
\n

\u62a5\u544a\u7ed3\u679c\u663e\u793a

\n
# 3 Node Cluster/ 4 x Samsung SM863 as OSD per Node\n# rados bench 60 write -b 4M -t 16\n# rados bench 60 read -t 16 (uses 4M from write)\n|Name | Read(MB/s)| Write(MB/s)|\n# 10 Gbit Network\n|------------|------------|------------|\n|SEQ4M Q? T16| 1064.42| 789.12|\n# 100 Gbit Network\n|------------|------------|------------|\n|SEQ4M Q? T16| 3087.82| 1011.63|\n
\n

\u53ef\u4ee5\u770b\u5230\u7f51\u7edc\u5e26\u5bbd\u5bf9\u6027\u80fd\u7684\u5f71\u54cd\u662f\u5de8\u5927\u7684\u3002\u867d\u7136 10 Gbit Network \u4e0b\u7684\u6027\u80fd\u4e0d\u8db3\uff0c\u4f46\u662f\u81f3\u5c11\u8bfb\u5199\u6027\u80fd\u90fd\u903c\u8fd1\u4e86\u5e26\u5bbd\u6781\u9650\u3002\u7136\u800c\u770b\u770b\u6211\u7684\u6d4b\u8bd5\u7ed3\u679c\uff0cWRITE \u975e\u5e38\u7cdf\u7cd5(296.25MB/s)

\n
pve 202009
\n

\u4ece\u62a5\u544a\u4e2d\u5f97\u77e5\uff0c\u6d4b\u8bd5\u89c4\u6a21\u4e3a 3 x Server; Each server 4 x Micron 9300 Max 3.2 TB (MTFDHAL3T2TDR); 1 x 100 GbE DACs, in a full-mesh topology

\n
# Micron 9300 Max 3.2 TB (MTFDHAL3T2TDR)\n|Name | Read(MB/s)| Write(MB/s)|\n|------------|------------|------------| \n|SEQ128KQ32T?| 3500.00| 3100.00| (MTFDHAL12T8TDR-1AT1ZABYY-Micron-LBGA-2022.pdf)\n|RND4K Q512T?| 3340.00| 840.00| (\u6839\u636e\u516c\u5f0f\u4f30\u7b97,throughput ~= iops * 4k / 1000)\n|. IOPS | 835000.00| 210000.00| (MTFDHAL12T8TDR-1AT1ZABYY-Micron-LBGA-2022.pdf)\n|------------|------------|------------| \n|RND4K Q1 T1 | | 205.82| (\u4ece\u62a5\u544a\u4e2d\u5f97\u77e5)\n|. IOPS | | 51000.00| (\u4ece\u62a5\u544a\u4e2d\u5f97\u77e5)\n|. latency ms| | 0.02| (\u4ece\u62a5\u544a\u4e2d\u5f97\u77e5)\n
\n

\u62a5\u544a\u7ed3\u679c\u663e\u793a

\n
# MULTI-VM WORKLOAD (LINUX)\n# \u6211\u4e0d\u7406\u89e3 Thread \u548c Job \u6709\u4ec0\u4e48\u533a\u522b\uff0c\u6587\u6863\u4e2d\u4e5f\u6ca1\u6709\u6807\u8bc6\u961f\u5217\u6df1\u5ea6\n|Name | Read(MB/s)| Write(MB/s)|\n|------------|------------|------------|\n|SEQ4M Q? T1 | 7176.00| 2581.00| (SEQUENTIAL BANDWIDTH BY NUMBER OF JOBS)\n|RND4K Q1 T1 | 86.00| 28.99| (\u6839\u636e\u516c\u5f0f\u4f30\u7b97)\n|. IOPS | 21502.00| 7248.00| (RANDOM IO/S BY NUMBER OF JOBS)\n
\n

\u540c\u6837\u7684\uff0cRND4K Q1 T1 WRITE \u6d4b\u8bd5\u7ed3\u679c\u975e\u5e38\u7cdf\u7cd5\uff0c\u53ea\u6709 7k iops\uff0c\u800c\u7269\u7406\u78c1\u76d8\u62e5\u6709 51k iops \uff0c\u8fd9\u6837\u7684\u5dee\u8ddd\u6211\u611f\u89c9\u662f\u65e0\u6cd5\u63a5\u53d7\u7684\u3002

\n
Dell R730xd report
\n

\u4ece\u62a5\u544a\u4e2d\u5f97\u77e5\uff0c\u6d4b\u8bd5\u89c4\u6a21\u4e3a 5 x Storage Server; Each Server 12HDD+3SSD, 3 x replication 2 x 10GbE NIC

\n
# \u4ece\u62a5\u544a\u4e2d\u6458\u6284\u7684\u6d4b\u8bd5\u7ed3\u679c\n# Figure 8 Throughput/server comparison by using different configurations\n|Name | Read(MB/s)| Write(MB/s)|\n|------------|------------|------------|\n|SEQ4M Q64T1 | 1150.00| 300.00|\n
\n

\u8fd9\u6837\u7684\u60c5\u51b5\u4e0b SEQ4M Q64T1 \u6d4b\u8bd5\u7ed3\u679c\u4e2d write \u53ea\u6709\u5927\u7ea6 300MB/s \uff0c\u8fd9\u5927\u6982\u53ea\u662f\u5355\u5757 SAS \u7684\u4e24\u500d\uff0c\u4e5f\u5c31\u662f 2 x 158.16 MB/s (4M blocks)\u3002\u8fd9\u8ba9\u6211\u96be\u4ee5\u7f6e\u4fe1\uff0c\u5b83\u751a\u81f3\u5feb\u8fc7\u4e8e\u6211\u7684 nvme \u78c1\u76d8\u3002\u4e0d\u8fc7\u53e6\u4e00\u4e2a\u91cd\u8981\u4e8b\u5b9e\u662f 12*5=60 \u5757 hdd \u53ea\u6709 300MB/s \u7684\u987a\u5e8f\u5199\u5165\u901f\u5ea6\uff0c\u8fd9\u6837\u7684\u6027\u80fd\u635f\u8017\u662f\u4e0d\u662f\u592a\u5927\u4e86\uff1f

\n
Micron report
\n

\u4ece\u62a5\u544a\u4e2d\u5f97\u77e5\uff0c\u6d4b\u8bd5\u89c4\u6a21\u4e3a 3 x Storage Server \uff1b Each Server 10 x micron 9300MAX 12.8T \uff0c2 x replication \uff0c2 x 100GbE NIC

\n
# micron 9300MAX 12.8T (MTFDHAL12T8TDR-1AT1ZABYY) \u7269\u7406\u78c1\u76d8\u6d4b\u8bd5 \n|Name | Read(MB/s)| Write(MB/s)| (? \u662f\u672a\u7ed9\u51fa\u53c2\u6570)\n|------------|------------|------------|\n|SEQ?M Q? T? | 48360.00| ?| (\u4ece\u62a5\u544a\u4e2d\u6458\u6284)\n|SEQ128KQ32T?| 3500.00| 3500.00| (MTFDHAL12T8TDR-1AT1ZABYY-Micron-LBGA-2022.pdf)\n|RND4K Q512T?| 3400.00| 1240.00| (\u6839\u636e\u516c\u5f0f\u4f30\u7b97)\n|. IOPS | 850000.00| 310000.00| (MTFDHAL12T8TDR-1AT1ZABYY-Micron-LBGA-2022.pdf)\n|. latency us| 86.00| 11.00| (MTFDHAL12T8TDR-1AT1ZABYY-Micron-LBGA-2022.pdf)\n|------------|------------|------------|\n|RND4K Q? T? | 8397.77| 1908.11| (\u6839\u636e\u516c\u5f0f\u4f30\u7b97)\n|. IOPS | 2099444.00| 477029.00| (\u4ece\u62a5\u544a\u4e2d\u6458\u6284\uff0cExecutive Summary)\n|. latency ms| 1.50| 6.70| (\u4ece\u62a5\u544a\u4e2d\u6458\u6284\uff0cExecutive Summary)\n
\n

\u5728 WRITE \u6d4b\u8bd5\u7ed3\u679c\u5982\u4e0b\uff0c

\n
# (\u4ece\u62a5\u544a\u4e2d\u6458\u6284)\n|Name | Read(MB/s)| Write(MB/s)|\n|------------|------------|------------|\n|RND4KQ32T100| ?| ?|\n|. IOPS | 2099444.00| 477029.00| (\u4e0d\u77e5\u9053\u662f\u4e0d\u662f\u5b98\u7f51\u62a5\u544a\u5b58\u5728\u95ee\u9898\uff0c\u8fd9\u91cc\u5c45\u7136\u6ca1\u6709\u4efb\u4f55\u6027\u80fd\u635f\u8017)\n|. latency ms| 1.52| 6.71|\n
\n

\u4e0d\u5f97\u4e0d\u8bf4 Micron \u5b98\u65b9\u7684\u6d4b\u8bd5\u5e73\u53f0\u8fc7\u4e8e\u9ad8\u7aef\uff0c\u4e0d\u662f\u6211\u4eec\u4e2d\u5c0f\u578b\u4f01\u4e1a\u8d1f\u62c5\u5f97\u8d77\u3002

\n

\u4ece\u7ed3\u679c\u4e2d\u5f97\u77e5\uff0cWRITE \u63a5\u8fd1\u4e8e\u5355\u5757\u7269\u7406\u78c1\u76d8\u6027\u80fd\u3002\u90a3\u4e48\u662f\u5426\u8bf4\u660e\uff0c\u5982\u679c\u53ea\u4f7f\u7528\u5355\u4e2a\u8282\u70b9\u5355\u4e2a\u78c1\u76d8\uff0c\u90a3\u4e48 WRITE \u6027\u80fd\u5c06\u4f1a\u4e0b\u964d\u5230 477k / 30 = 15.9k iops ? \u5982\u679c\u662f\u7684\u8bdd\uff0c\u90a3\u8fd9\u5c06\u662f sata ssd \u7684\u6027\u80fd\u3002

\n

\u6700\u540e\u7684\u6700\u540e\uff0c\u6211\u60f3\u4e86\u89e3\u7684\u95ee\u9898\u662f\uff1a

\n
    \n
  1. \u5982\u4f55\u4fee\u590d ceph \u4e2d\u7684 write \u6027\u80fd\u95ee\u9898\uff1f ceph \u80fd\u4e0d\u80fd\u505a\u5230\u548c VMware vSAN \u540c\u6837\u7684\u6027\u80fd\u3002
  2. \n
  3. \u7ed3\u679c\u4e2d\u770b\u5230\u5168\u95ea\u5b58\u78c1\u76d8\u7684\u6027\u80fd\u8fd8\u4e0d\u5982 hdd+ssd \uff0c\u90a3\u4e48\u5982\u679c\u4e0d\u4f7f\u7528 bcache \u7684\u8bdd\uff0c\u8981\u600e\u4e48\u505a\u624d\u4fee\u590d ceph \u5168\u95ea\u5b58\u76d8\u4e0b\u7684\u6027\u80fd\u95ee\u9898\uff1f
  4. \n
  5. \u5bf9\u4e8e hdd+ssd \u67b6\u6784\u662f\u5426\u8fd8\u6709\u66f4\u597d\u7684\u65b9\u6848\uff1f
  6. \n
\n" }, { "author": { "url": "member/Livid", "name": "Livid", "avatar": "https://cdn.v2ex.com/avatar/c4ca/4238/1_large.png?m=1778546420" }, "url": "t/73330", "date_modified": "2013-06-23T07:34:38+00:00", "content_html": "http://ceph.com/docs/master/", "date_published": "2013-06-23T07:33:02+00:00", "title": "\u6709\u4eba\u5728 production \u73af\u5883\u4e2d\u7528 Ceph \u4e48\uff1f", "id": "t/73330" } ] } ubao msn snddm index pchome yahoo rakuten mypaper meadowduck bidyahoo youbao zxmzxm asda bnvcg cvbfg dfscv mmhjk xxddc yybgb zznbn ccubao uaitu acv GXCV ET GDG YH FG BCVB FJFH CBRE CBC GDG ET54 WRWR RWER WREW WRWER RWER SDG EW SF DSFSF fbbs ubao fhd dfg ewr dg df ewwr ewwr et ruyut utut dfg fgd gdfgt etg dfgt dfgd ert4 gd fgg wr 235 wer3 we vsdf sdf gdf ert xcv sdf rwer hfd dfg cvb rwf afb dfh jgh bmn lgh rty gfds cxv xcv xcs vdas fdf fgd cv sdf tert sdf sdf sdf sdf sdf sdf sdf sdf sdf sdf sdf sdf sdf sdf sdf sdf sdf sdf sdf sdf sdf sdf sdf sdf sdf sdf sdf sdf sdf sdf sdf sdf sdf sdf sdf sdf sdf sdf sdf sdf shasha9178 shasha9178 shasha9178 shasha9178 shasha9178 liflif2 liflif2 liflif2 liflif2 liflif2 liblib3 liblib3 liblib3 liblib3 liblib3 zhazha444 zhazha444 zhazha444 zhazha444 zhazha444 dende5 dende denden denden2 denden21 fenfen9 fenf619 fen619 fenfe9 fe619 sdf sdf sdf sdf sdf zhazh90 zhazh0 zhaa50 zha90 zh590 zho zhoz zhozh zhozho zhozho2 lislis lls95 lili95 lils5 liss9 sdf0ty987 sdft876 sdft9876 sdf09876 sd0t9876 sdf0ty98 sdf0976 sdf0ty986 sdf0ty96 sdf0t76 sdf0876 df0ty98 sf0t876 sd0ty76 sdy76 sdf76 sdf0t76 sdf0ty9 sdf0ty98 sdf0ty987 sdf0ty98 sdf6676 sdf876 sd876 sd876 sdf6 sdf6 sdf9876 sdf0t sdf06 sdf0ty9776 sdf0ty9776 sdf0ty76 sdf8876 sdf0t sd6 sdf06 s688876 sd688 sdf86