在現(xiàn)場施工快一個月，歷經(jīng)前期艱難調(diào)試摸索，目前系統(tǒng)已能正常運行，但是有的模塊作的很低效，我試圖將其改造一下。
現(xiàn)狀：
就拿最基本的一個功能點來說，在地圖頁面上實時顯示地理數(shù)據(jù)（例如：卡點顯示實時過車、實時顯示車輛GPS位置等），以前我的處理方式是采用頁面定時器，間隔一段時間去查詢后臺數(shù)據(jù)庫，過車數(shù)據(jù)則通過前置設(shè)備采集通過Hessian遠(yuǎn)程調(diào)用將過車數(shù)據(jù)插入后臺數(shù)據(jù)庫，其實現(xiàn)方法簡單。但是最大的弊病在于低效和易錯，由于前置采集設(shè)備采集和執(zhí)行Hessian，如果網(wǎng)絡(luò)狀態(tài)不佳，過車數(shù)據(jù)將會丟失，同時存入數(shù)據(jù)庫的時間不可能嚴(yán)格真實，但是我的頁面上定時器是嚴(yán)格按照時間去查詢，容易出現(xiàn)漏查現(xiàn)象，而且頻繁的對數(shù)據(jù)庫進(jìn)行操作，增加了數(shù)據(jù)庫的負(fù)擔(dān)。
改造：
1.引入消息中間件，前置設(shè)備作為消息生產(chǎn)者，將采集到的數(shù)據(jù)裝入消息中間件中
2.前端頁面通過WebSocket和后臺進(jìn)行連接，當(dāng)消息中間件中有消息時立即消費，通過WebSocket傳給前端頁面
改造后結(jié)構(gòu)是這樣

結(jié)構(gòu)圖.jpg

image.png

   @PostConstruct
    public void startConsuming() {
        Configuration config = new Configuration();
        config.setHostname("localhost");
        config.setPort(9093);
        final SocketIOServer server = new SocketIOServer(config);
        server.addEventListener("chatevent", GPSMessage.class, new DataListener<GPSMessage>() {
            @Override
            public void onData(SocketIOClient client, GPSMessage data, AckRequest ackRequest) {
                // broadcast messages to all clients
                server.getBroadcastOperations().sendEvent("chatevent", data);
            }
        });
        server.start();

        listenerExecService = Executors.newFixedThreadPool(LISTENER_NUM_THREADS);
        handlerExecService = Executors.newFixedThreadPool(EXECUTOR_NUM_THREADS);
        consumerConfig = consumerConfigFactory.getConsumerConfig();
        topicCountMap = new HashMap<String, Integer>();
        topicCountMap.put(topic, CONSUMER_NUM_THREADS);
        gson = new Gson();

        handlerTaskFactory = (String gpsText) -> new GPSTask(gpsText, gson, server);

        super.start();
    }

public void start() {
        ConsumerConnector consumer = Consumer.createJavaConsumerConnector(this.getConsumerConfig());
        Map<String, List<KafkaStream<byte[], byte[]>>> consumerMap = consumer.createMessageStreams(this.getTopicCountMap());
        Iterator var3 = consumerMap.entrySet().iterator();

        while(var3.hasNext()) {
            Entry<String, List<KafkaStream<byte[], byte[]>>> entry = (Entry)var3.next();
            List<KafkaStream<byte[], byte[]>> streams = (List)entry.getValue();
            int threadNumber = 0;

            for(Iterator var7 = streams.iterator(); var7.hasNext(); ++threadNumber) {
                KafkaStream<byte[], byte[]> stream = (KafkaStream)var7.next();
                this.getListenerExecService().submit(new ConsumerListenTask(stream, threadNumber, this.getHandlerTaskFactory((String)entry.getKey()), this.getHandlerExecService()));
            }
        }

    }

public class ConsumerListenTask implements Runnable {
    public void run() {
        ConsumerIterator it = this.kafkaStream.iterator();

        while(it.hasNext()) {
            byte[] messageData = (byte[])it.next().message();
            String reply = new String(messageData);
            this.executorPool.submit((Runnable)this.executorTaskFactory.apply(reply));
            System.out.println("Consumed Thread:" + this.threadNumber + ".Consuming User: " + reply);
        }

        System.out.println("Shutting down Thread: " + this.kafkaStream);
    }
}

public class GPSTask implements Runnable {
    @Override
    public void run() {
        try{
            GPSMessage gpsMessage = gson.fromJson(this.gpsText, GPSMessage.class);
            System.out.println("+++++++++++從kafka中取到車牌號"+gpsMessage.getHphm());
            socketIOServer.getBroadcastOperations().sendEvent("chatevent", gpsMessage);
        } catch (Exception e) {
            e.printStackTrace();
        }
    }
}

3.前端頁面利用socket.io監(jiān)聽事件就行

socket.on('chatevent', function(data) {
        console.log(data);
        switch(data.operate){
          case 'create':
            var coordinate = ol.proj.transform([Number(data.lon), Number(data.lat)], 'EPSG:4326', 'EPSG:3857');
            var point = new ol.geom.Point(coordinate);
            var markerFeature = new ol.Feature({
                name: data.hphm,
                type: 'position',
                geometry: point
            });
            vectorSource.addFeature(markerFeature);
            break;
        }

      });

4.通過kafka自帶的messageproducer，注冊topic并模擬過車數(shù)據(jù)，將消息發(fā)往消息中間件，后臺通過監(jiān)聽線程池來及時消費該消息，最后通過netty-socketio廣播到前端頁面

.\bin\windows\kafka-console-producer.bat --broker-list localhost:9092 --topic gpsinfo

目前只做了個Demo，接下來可能會遇到的問題，消息生產(chǎn)和消費不同步，路上過車數(shù)據(jù)將會是短時內(nèi)大量過車，會生產(chǎn)大量的過車消息；而頁面顯示不可能和消息同步，否則一閃而過影響顯示效果，但處理不及時將會造成大量消息積壓，有待進(jìn)一步思考嘗試。