[
https://issues.apache.org/jira/browse/PROTON-1910?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16608274#comment-16608274
]
Alan Conway commented on PROTON-1910:
-------------------------------------
Possible fix:
# Re-implement amqp.Message to hold all message data in a Go struct. No C
calls required for any of the message property getters/setters.
# Implement C functions goEncode to copy all the Go data to a C message and
encode it in a single C function, likewise goDecode.
# Keep a single pn_message_t C object associated associated with each Link.
Send/receive is serialized per link, so a single message is sufficient for all
encoding/decoding.
This will avoid multiple message ctor/dtor calls by automatically re-using a C
message (this is a known performance win in plain C code) and will move all
message-related C calls into the proton goroutine. There will be fewer C calls
overall and all will be serialized in a single goroutine which hopefully will
reduce the CGO overhead - which is due mainly to the need to schedule C calls
on a native thread.
> Profiling indicates that cgo becomes a bottleneck during scale testing of
> electron
> ----------------------------------------------------------------------------------
>
> Key: PROTON-1910
> URL: https://issues.apache.org/jira/browse/PROTON-1910
> Project: Qpid Proton
> Issue Type: Bug
> Components: go-binding
> Affects Versions: proton-c-0.24.0
> Reporter: Aaron Smith
> Assignee: Alan Conway
> Priority: Major
>
> While performing scale testing, detailed profiling of Go test clients showed
> that >95% of the execution time can be devoted to the cgo call. The issues
> seems to be related on sends to the NewMessage() call. For receives, the
> bottleneck is both NewMessage() and the call to actually receive the message.
>
>
> This behavior is not unexpected as CGO is a well-known bottleneck. Would it
> be possible to have a NewMessage() call that return multiple messages and a
> recv call that took an "At most" argument. i.e. recv(10) would receive 10 or
> fewer messages that might be waiting in the queue. Also, it would be nice to
> be able to trade latency for throughput in that the callback wasn't triggered
> until N messages were recieved (with timeout)....
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
