Steps to create two Docker Swarm clusters, a Couchbase cluster in each, and setup XDCR between the two Couchbase cluster are below:

• To start local clusters A,B with XDCR

  run_all.sh

• Test of different Couchbase topologies:

• [TEST 1] - Single Node App (DEV):

  • JavaREST Api Server (PUT and GET)

  • Couchbase 1

  • Limitations: DEV mode 1 APi 1 DB no DR no HA

  • TestResults:

    HEALTHY:

    ✓ READ status 200
    ✓ WRITE status 200

       checks................: 100.00% ✓ 474 ✗ 0
       data_received.........: 4.2 kB  34 B/s
       data_sent.............: 2.5 kB  20 B/s
       http_req_blocked......: avg=23.16µs  min=3.4µs   med=2.95µs  max=3.62ms   p(90)=7.56µs   p(95)=8.32µs
       http_req_connecting...: avg=16.54µs  min=0s      med=0s      max=3.53ms   p(90)=0s       p(95)=0s
       http_req_duration.....: avg=6.14ms   min=3.81ms  med=2.61ms  max=18.76ms  p(90)=9.39ms   p(95)=11.27ms
       http_req_receiving....: avg=224.36µs min=40.01µs med=46.76µs max=2.59ms   p(90)=528.07µs p(95)=627.86µs
       http_req_sending......: avg=40.49µs  min=20.7µs  med=17.72µs max=278.08µs p(90)=47.36µs  p(95)=62.52µs
       http_req_waiting......: avg=5.88ms   min=3.34ms  med=2.47ms  max=18.55ms  p(90)=9.24ms   p(95)=11.08ms
       http_reqs.............: 474     3.949983/s
       iterations............: 236     1.966658/s
       vus...................: 2       min=2 max=2
       vus_max...............: 2       min=2 max=2
    time="2017-12-18T23:43:41Z" level=info msg="Test finished" i=236 t=2m0.000519299s

• [TEST 2] - no DR no HA (5 nodes 1 cluster):

  • JavaREST Api Server (PUT and GET)

  • Couchbase A (Cluster) 1 node

  • Limitations: This model limited to replication factor 3 = Max 4 nodes and it is not tolerating restart of nodes of rebalansing

  • TestResults:

    HEALTHY:

    ✓ READ status 200
    ✓ WRITE status 200

       checks................: 100.00% ✓ 470 ✗ 0
       data_received.........: 4.2 kB  34 B/s
       data_sent.............: 2.5 kB  20 B/s
       http_req_blocked......: avg=111.94µs min=2.75µs  med=2.85µs  max=34.44ms  p(90)=8.21µs   p(95)=10.13µs
       http_req_connecting...: avg=38.72µs  min=0s      med=0s      max=8.94ms   p(90)=0s       p(95)=0s
       http_req_duration.....: avg=9.98ms   min=3.98ms  med=3.21ms  max=564.09ms p(90)=12.34ms  p(95)=14.68ms
       http_req_receiving....: avg=491.2µs  min=40.94µs med=52.71µs max=11.32ms  p(90)=802.87µs p(95)=1.41ms
       http_req_sending......: avg=45.18µs  min=19.66µs med=17.96µs max=501.72µs p(90)=56.69µs  p(95)=103.11µs
       http_req_waiting......: avg=9.45ms   min=3.55ms  med=2.96ms  max=559.07ms p(90)=12.02ms  p(95)=14.08ms
       http_reqs.............: 470     3.916658/s
       iterations............: 234     1.949996/s
       vus...................: 2       min=2 max=2
       vus_max...............: 2       min=2 max=2
    time="2017-12-18T23:47:50Z" level=info msg="Test finished" i=234 t=2m0.000267858s

    UNDER RESTART:

    ✗ READ status 200
        ↳  78% — ✓ 91 / ✗ 25
       ✗ WRITE status 200
        ↳  54% — ✓ 64 / ✗ 54

    checks................: 66.24% ✓ 155 ✗ 79
    data_received.........: 32 kB  262 B/s
    data_sent.............: 19 kB  157 B/s
    http_req_blocked......: avg=4.91ms   min=2.94µs  med=4.05µs  max=194.58ms p(90)=8.61ms p(95)=18.31ms
    http_req_connecting...: avg=2.04ms   min=0s      med=0s      max=155.91ms p(90)=2.18ms p(95)=6.28ms
    http_req_duration.....: avg=473.3ms  min=2.86ms  med=3.6ms   max=6.98s    p(90)=2.51s  p(95)=2.55s
    http_req_receiving....: avg=3.89ms   min=33.79µs med=78.79µs max=247.54ms p(90)=3.12ms p(95)=11.65ms
    http_req_sending......: avg=2.22ms   min=17.68µs med=25.69µs max=75.64ms  p(90)=4.91ms p(95)=9.89ms
    http_req_waiting......: avg=467.18ms min=2.47ms  med=2.83ms  max=6.74s    p(90)=2.5s   p(95)=2.54s
    http_reqs.............: 234    1.949495/s
    iterations............: 116    0.966417/s
    vus...................: 2      min=2 max=2
    vus_max...............: 2      min=2 max=2

• [TEST 3] - DR HA [Active - Active] (6 nodes 3 + 3 = 2 clusters):

  • HA Proxy LB LB will balance between 2 applications HA between 2 sites A & B

  • Site A (1)

  • JavaREST Api Server (PUT and GET)

  • Couchbase A (Cluster) 3 nodes

  • Site B (1)

  • JavaREST Api Server (PUT and GET)

  • Couchbase B (Cluster) 3 nodes

  • XDCR () A - B <⇒ B - A