А, ну да. Вкупе со вторым предложением текста - и вовсе каламбур.
Кстати, зашел на сайт SFS, не был у них с полгода. Интересные данные:
Benchmark 1 - Discard message.
Server machine
Intel Core2 Duo, 2.2Ghz
4GB RAM
Linux Ubuntu 9.04
OpenJDK Server VM 1.6.0_0-b12
Round 1
Msg Size: 1024 bytes | Rate: 1 msg/s | Clients: 2000 | Target traffic = 16Mbit/s
CPU %RAM %IN (MB/s)OUT(MB/s)BitSwarm4-5%2.2%2.10Netty6-7%2.1%2.10Mina14-15%2.5%2.10
Round 2
Msg Size: 8192 bytes | Rate: 1 msg/s | Clients: 2000 | Target traffic = 128Mbit/s
CPU %RAM %IN (MB/s)OUT(MB/s)BitSwarm22-23%7%16.30Netty33-36%6.9%16.20Mina80-82%7.2%16.20
Round 3
Msg Size: 16384 bytes | Rate: 1 msg/s | Clients: 4000 | Target traffic = 512Mbit/s
CPU %RAM %IN (MB/s)OUT(MB/s)BitSwarm97%17%650Netty135% (for a while then crashes) N/A~80Mina(crashed before reaching target traffic)N/AN/A0
Benchmark 2 - Echo message.
Server machine
Intel Core2 Duo, 2.2Ghz
4GB RAM
Linux Ubuntu 9.04
OpenJDK Server VM 1.6.0_0-b12
Round 1
Msg Size: 1024 bytes | Rate: 1 msg/s | Clients: 2000 | Target traffic = 16Mbit/s
CPU %RAM %IN (MB/s)OUT(MB/s)BitSwarm11-12%4.2%2.22.1Netty13-14%4%2.22.1Mina23-24%4.9%2.22.1
Round 2
Msg Size: 2048 bytes | Rate: 1 msg/s | Clients: 3000 | Target traffic = 50Mbit/s
CPU %RAM %IN (MB/s)OUT(MB/s)BitSwarm25%5%6.76.5Netty30%5.8%6.76.5Mina65%5.9%6.76.5
Round 3
Msg Size: 2048 bytes | Rate: 1 msg/s | Clients: 6000 | Target traffic = 100Mbit/s
CPU %RAM %IN (MB/s)OUT(MB/s)BitSwarm52%7.2%13.513.1Netty62-64%7.5%13.513.1Mina145-150% (*)8.9%13.313
Benchmark 3 - Broadcast echo message.
Server machine
Intel Core 2 Duo, 2.2Ghz
4GB RAM
Linux Ubuntu 9.04
OpenJDK Server VM 1.6.0_0-b12
BandwidthBitSwarmBitSwarmMinaMinaNettyNettyCPURAMCPURAMCPURAM80 Mbit/s14%1.40%55%N/A50%4.60%320 Mbit/s45%1.30%139%3.50%crashedN/A840 Mbit/s91%2.20%crashedN/AN/AN/A
И не вполне объективная инфа (хотя и приятная):
Client API
Server machine
MBPro dual-core
i7@2.6Ghz, 8GB Ram
Java Runtime: 1.6.0_20
Java HotSpot(TM) 64-Bit Server VM (build 16.3-b01-279, mixed mode)
Client machine
iMac Core2 Duo @3.0Ghz 4GB Ram
Flash Player standalone version 10.0.2.54
Single Message test
The test shows a typical real-time game scenario where lots of messages arrive each second to the client. One message is sent to the client every <interval> amount of milliseconds as specified in the table.
IntervalMsg countBandwidthServer CPUClient CPU10 ms100 msg/s33KB / 38KB3% - 4%6% - 22%5 ms200 msg/s66KB / 78KB4.7% - 7%8% - 40%2 ms500 msg/s167KB / 190KB8% - 14%12% - 90%
1 ms 1000 msg/s 332KB / FAIL13% - 58%13% - FAIL
In red (первая колонка) are the values obtained with SmartFoxServer 2X and related API.
In blue (вторая колонка) are the values obtained with the other game-server and related API.
At 200msg/sec you can already see how the alternative product is eating 40% of the client resources just to deal with the network traffic, while the SFS API have a very light impact (less than 10%).
