FTS middleware doc.
- 1. FTS MW Reliability & Performance 06-23-2009
- 2. 2 Reliability & Performance FTS MW Data Control Connection NIO Bandwidth control Deploy Cluster Performance Object pool Thread pool Split buffers Asynchronous Data filter Data match
- 3. 3 Based on the Jetty Cometd solution • NIO Connectors The connections received by the NIO connectors are always run in non-blocking IO mode and blocking semantics is applied only to the input streams and output streams that wrap the connection when passed to a HTTP request. Where possible and appropriate, advanced NIO features such as gather writes, direct buffers and memory mapped file buffers will be used. • Continuations To deal with the need for HTTP request to have an efficient waiting mechanism, Jetty 6 introduced a Continuation Mechanism, which allows the current threads handler to be suspended and resumed at a later time in response to a timeout or an asynchronous event. • Split Buffers Jetty 6 uses a split buffer architecture and dynamic buffer allocation. An idle connection will have no buffers allocated to it, but once a request received only an small request header buffer is allocated
- 4. 4 The Jetty Cometd Test Report The following table indicates the Jetty cometd just spend 875 threads and 57M memory on 10000 users Formula Web 1.0 Web 2.0 + Comet Jetty Cometd Users u 10000 10000 10000 Requests/Burst b 5 2 2 Burst period (s) p 20 5 5 Request Duration (s) d 0.200 0.150 0.175 Poll Duration (s) D 0 10 10 Request rate (req/s) rr=u*b/20 2500 4000 4000 Poll rate (req/s) pr=u/d 0 1000 1000 Total (req/s) r=rr+pr 2500 5000 5000 Concurrent requests c=rr*d+pr*D 500 10600 10700 Min Threads T=c T=r*d 500 - 10600 - - 875 Stack memory S=64*1024*T 32MB 694MB 57MB
- 5. 5 Data Control Data Source Gather Original data Gather the huge data from different data sources. Data Image The larger dataset will be organized into several data images by applying different Data filter model, such as merge, distinct. The data image will reduce the data size because of data filter. Filter Match Bandwidth Client Data Each client has respective data collection according to respective match action/ criteria.
- 6. 6 Data Control Example All stock exchange prices data, thousands upon thousands records Ten best share Ten worst share merge image give me five best give me stock 001,002,003 price just stock 001 price Pass Original Image User Real Transfer Pass Sorry, your bandwidth can’t transfer so much data, discard some oldest data
- 7. 6 Data Control Example All stock exchange prices data, thousands upon thousands records Ten best share Ten worst share merge image give me five best give me stock 001,002,003 price just stock 001 price Pass Original Image User Real Transfer Pass Sorry, your bandwidth can’t transfer so much data, discard some oldest data