Re: Performance over socket?
304 views
Skip to first unread message
crotty.ch...@gmail.com
Dec 20, 2012, 9:40:04 AM12/20/12
to prot...@googlegroups.com
True enough, but the marshalling/unmarshalling performance can be measured.
I guess my question would have been better worded as...
"I have a project that needs to send objects between a Java and a C++ service at a rate of up to several thousand object/second. Would you recommend using protocol buffers in such a scenario?"
On Thursday, December 20, 2012 9:09:31 AM UTC-5, 肖锋 wrote:
I guess my question would have been better worded as...
"I have a project that needs to send objects between a Java and a C++ service at a rate of up to several thousand object/second. Would you recommend using protocol buffers in such a scenario?"
On Thursday, December 20, 2012 9:09:31 AM UTC-5, 肖锋 wrote:
Protobuf is a data marshaling library. How is its performance related to the way you used to transfer the data? I think the two are independent.
Oliver Jowett
Dec 20, 2012, 4:46:13 PM12/20/12
to crotty.ch...@gmail.com, prot...@googlegroups.com
On Thu, Dec 20, 2012 at 2:40 PM, <crotty.ch...@gmail.com> wrote:
> True enough, but the marshalling/unmarshalling performance can be measured.
>
> I guess my question would have been better worded as...
>
> "I have a project that needs to send objects between a Java and a C++
> service at a rate of up to several thousand object/second. Would you
> recommend using protocol buffers in such a scenario?"
We do exactly that, the protobuf cost was never significant when I profiled it.
> True enough, but the marshalling/unmarshalling performance can be measured.
>
> I guess my question would have been better worded as...
>
> "I have a project that needs to send objects between a Java and a C++
> service at a rate of up to several thousand object/second. Would you
> recommend using protocol buffers in such a scenario?"
(It's admittedly hard to see the GC contribution on the Java side)
Oliver
肖锋
Dec 20, 2012, 11:02:38 PM12/20/12
to crotty.ch...@gmail.com, prot...@googlegroups.com
On Thu, Dec 20, 2012 at 10:40 PM, <crotty.ch...@gmail.com> wrote:
True enough, but the marshalling/unmarshalling performance can be measured.
I guess my question would have been better worded as...
"I have a project that needs to send objects between a Java and a C++ service at a rate of up to several thousand object/second. Would you recommend using protocol buffers in such a scenario?"
Well, almost all Google servers are talking protocol buffers to each other and that's what protocol buffers are designed for.
On Thursday, December 20, 2012 9:09:31 AM UTC-5, 肖锋 wrote:Protobuf is a data marshaling library. How is its performance related to the way you used to transfer the data? I think the two are independent.
--To view this discussion on the web visit https://groups.google.com/d/msg/protobuf/-/ppCfsEglYaIJ.
You received this message because you are subscribed to the Google Groups "Protocol Buffers" group.
To post to this group, send email to prot...@googlegroups.com.
To unsubscribe from this group, send email to protobuf+u...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/protobuf?hl=en.
Дмитрий Дьяченко
Dec 20, 2012, 2:43:32 PM12/20/12
to crotty.ch...@gmail.com, prot...@googlegroups.com
in my hardware (Fedora 17, 4 core i5 2.8 GHz) protobuf + libzmq
achieve 10^4 "messages" per second.
"pure" TCP/IP stack w/o any usefull work achieve 5--7 * 10^4 "messages"
its good or poor performance? ;)
Dmitry
2012/12/20 <crotty.ch...@gmail.com>:
achieve 10^4 "messages" per second.
"pure" TCP/IP stack w/o any usefull work achieve 5--7 * 10^4 "messages"
its good or poor performance? ;)
Dmitry
2012/12/20 <crotty.ch...@gmail.com>:
Reply all
Reply to author
Forward
0 new messages