Currently we are in the process of developing an orchestration layer using Node. this layer will act as a façade to reach out multiple services/datasource to compose client specific response.
Is there any nodejs sizing information available today that we would be able to use as a starting point.
Any kind of help is greatly appreciated.
Can I ask what you mean by sizing information?
May hapijs meets your need.
Probably the two most important things I've used as a guide is that
Node.js is single threaded (a single process cannot benefit from
multiple CPUs) and very low memory footprint.
Anecdotally, in my experience performance testing the Open Academic
Environment [1], an application that does relatively little business
logic and has no local caches, I haven't been able to push it above an
80Mb memory footprint before it becomes bound by IO or CPU.
I would start with an Amazon t1.micro or Joyent xsmall (~512Mb of
memory and 1 CPU) and performance test a single process. Decide how
you need to scale thereafter based on how much load it can sustain
(are you I/O bound? Memory bound?). Remember you also probably want
monitoring tools to run on the machines, configuration management,
etc... Which will impact how much memory and CPU you want on those
VMs, so just test test test under those conditions.
Hope that helps,
PS:
[1] https://github.com/oaeproject/
댓글 없음:
댓글 쓰기