On 13/04/18 11:16, Marc Schlegel wrote:
I would like to know if Akka Http is suitable to implement a small
orchestration layer.
The idea is to send events from the microservice to the orchestration layer
(Akka with Akka Http). I think Akka is a nice fit for orchestration but I
am not sure if Akka Http is the best solution to work with external systems.
Note: the events are implemented by Google Protocolbuffers because my
demo-projects implements the services in several languages (Go, Java, Rust)
and by using protobuf I can generate consistent DTOs. I've first looked
into Alpakka but it seems this has a different focus and currently no
official support for protobuf messages.
I've prototyped something that sounds a little similar to this and it
works well. I have a controller actor running on one node that maintains
a list of pending work. Worker actors are started on other nodes and
connect to the controller actor to request work which they then process.
The worker actors are just thin wrappers around instances of a large,
legacy C++ application which is in effect a batch process. The worker
nodes create a worker actor for each instance of the legacy app, plus
there is a shared localhost-only Akka HTTP server actor to provide a
protobuf endpoint for the C++ app instances on the worker node to talk
to. The C++ app has been modified to use libcurl & protobuf to connect
to the localhost HTTP server to request work details and provide
completion information which is 'proxied' back by the worker actors to
the controller actor on the controller node.
All the external communication is done with protobuf, both between the
akka actors on the master and worker nodes and between the worker actors
and the legacy C++ app, via the localhost HTTP actor. I've used akka
remoting for the cross-node control rather than akka cluster as if the
controller node dies there is nothing further that can be done anyway
and the worker actors just kill any running C++ app instances and exit.
Workers can be added dynamically and will be issued pending work and if
they die the controller will reschedule the work they were doing to
another worker.
This all works nicely and protobuf integrates well into Akka HTTP, with
the addition of a tiny amount of glue code to to the protobuf
marshalling for akka HTTP - see
https://github.com/scalapb/ScalaPB/issues/247 for the details. With that
in place, the routing logic for processing looks similar to this
outline, where JobFinishedMsg is a protobuf message:
path("job_finished") {
post {
entity(as[JobFinishedMsg])(msg => {
log.debug("received message {}", msg.id)
???
}
}
}
--
Alan Burlison
--
--
*****************************************************************************************************
** New discussion forum: https://discuss.akka.io/ replacing akka-user
google-group soon.
** This group will soon be put into read-only mode, and replaced by
discuss.akka.io
** More details: https://akka.io/blog/news/2018/03/13/discuss.akka.io-announced
*****************************************************************************************************
Read the docs: http://akka.io/docs/
Check the FAQ: http://doc.akka.io/docs/akka/current/additional/faq.html
Search the archives: https://groups.google.com/group/akka-user
---
You received this message because you are subscribed to the Google Groups "Akka User List" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To post to this group, send email to [email protected].
Visit this group at https://groups.google.com/group/akka-user.
For more options, visit https://groups.google.com/d/optout.
