Hi Shiva, this question is impossible to answer generically, the only thing that strikes me is that your CPU utilization shows that the system is not yet maxed out (you might also want to consider testing the load generator whether it can go higher by itself).
If you need specific help for this problem I recommend commercial support, for example ping [email protected]. Regards, Roland 26 feb 2014 kl. 15:15 skrev Shiva Ramagopal <[email protected]>: > Hi, > > I have designed the core of my application using Akka actors which I'm > describing below: > > The application is a request-response system. There are three entities - the > client, the main server and the agent. > > Communication between all the entities are over plain HTTP. For every request > the client sends to the main server, the main server sends a request each to > multiple remote agents, usually between 5 to 10. Each agent responds to the > request within 100 ms with a piece of information which the main server > collates and responds back to the client. > > The main server is on a separate machine and each of the agents are on > different machines. > > In my design the main server is designed using Akka actors. HTTP front-ending > is done using the Simple Framework. Each received HTTP request is converted > to a "DataRequest" message (containing the HTTP response object) and > round-robin dispatched by a RouterActor to 5 instances of MasterActor. A > MasterActor then creates multiple AgentActor instances. Each AgentActor makes > an async HTTP request (using Ning AsyncHttpClient with > NettyAsyncHttpProvider) to the remote agent and responds to the MasterActor > with the received data as a "DataResponse" message. The MasterActor finally > responds back to client with the collated data. > > Peformance testing on a 4-core i7 machine with 16 GB ram gave the following > results: > > Throughput: ~640 tps, peak > CPU usage: ~23% overall (application only, across all cores). There's no > other load on the HW. > Response times, as seen by the client: 105 ms (median), which includes the > 100 ms agent response time. > > My objective now is to increase the throughput to about 4000 tps while > maintaining the same end-to-end response time. > > 1. What's the maximum ballpark throughput that I can expect with the above > design on the same HW? Does aiming for 4000 tps sound reasonable? > > 2. What design changes or tuning can be done to achieve the target of 4000 > tps? > > I would greatly appreciate any pointers or suggestions on how to do this. > > Best regards, > Shiva > > > -- > >>>>>>>>>> Read the docs: http://akka.io/docs/ > >>>>>>>>>> Check the FAQ: http://akka.io/faq/ > >>>>>>>>>> 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 http://groups.google.com/group/akka-user. > For more options, visit https://groups.google.com/groups/opt_out. Dr. Roland Kuhn Akka Tech Lead Typesafe – Reactive apps on the JVM. twitter: @rolandkuhn -- >>>>>>>>>> Read the docs: http://akka.io/docs/ >>>>>>>>>> Check the FAQ: http://akka.io/faq/ >>>>>>>>>> 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 http://groups.google.com/group/akka-user. For more options, visit https://groups.google.com/groups/opt_out.
