Couchbase server poor ops/sec issue (disk creates very low)


#1

Enviroment:
couchbase -> SG.15(2 node) & CB5.0 CE edition cluster(1 node)
GCP -> 1 LoadBalancer for SG, 1 LoadBalancer for CB

problem:
Our CB server’s “ops/sec” alway shows “200 ops/sec” when i using http post request to create data
about 15k data(using /_bulk_docs)

I think this is very low, right?
documents said: the normal one is 10K ops/sec.

most important: disk new item only around 10/sec
anything i can do better?


#2

@willyou,

are you processing the 15K of data:

  1. all at once via _bulk_docs
  2. chopping up the 15K into smaller chunks and doing them sequential
  3. chopping up the 15K into smaller chunks and doing them in parallel

Is your load balancer routing the traffic to just one Sync Gateway node?


#3

We use 1.

After that
We use nodejs sdk to insert data
Disk creates can reach 150 per second
Is that normal or we can do better?

Load balancer routing traffic to 2 SG nodes

Actually ,we did tried 2.
It’s same with 1.

Fujio couchbase@discoursemail.com於 2017年12月7日 週四,上午12:43寫道:


#4

@willyou,

Sync gateway can do more as long as your docs are not very big 100KB or more.

Couchbase Lite PUSH/WRITES

Here (IMAGE BELOW) is a typical Write operation that Couchbase Lite Dose on a PUSH to Sync Gateway.

CloudApp

So on step 5 the _bulk_doc … it will do 2.

  1. chopping up the 15K into smaller chunks and doing them sequential

but

sync gateway can handle _bulk_doc from multiple Couchbase Lite instance. So it also dose

  1. chopping up the 15K into smaller chunks and doing them in parallel

#5

Thanks a lot!!

I think i’ve found some solutions here.