Question

我在3个VM上运行spark 1.6（即1x主站; 2x从站），全部有4个内核和16GB RAM。

我可以看到工作人员在spark-master webUI上注册。

我想从Vertica数据库中检索数据来处理它。由于我没有设法运行复杂的查询，我尝试了虚拟查询来理解。我们认为这是一项简单的任务。

我的代码是：

df = sqlContext.read.format('jdbc').options(url='xxxx', dbtable='xxx', user='xxxx', password='xxxx').load()
four = df.take(4)

输出是（注意：我用@IPSLAVE替换从属VM IP：端口）：

16/03/08 13:50:41 INFO SparkContext: Starting job: take at <stdin>:1
16/03/08 13:50:41 INFO DAGScheduler: Got job 0 (take at <stdin>:1) with 1 output partitions
16/03/08 13:50:41 INFO DAGScheduler: Final stage: ResultStage 0 (take at <stdin>:1)
16/03/08 13:50:41 INFO DAGScheduler: Parents of final stage: List()
16/03/08 13:50:41 INFO DAGScheduler: Missing parents: List()
16/03/08 13:50:41 INFO DAGScheduler: Submitting ResultStage 0 (MapPartitionsRDD[1] at take at <stdin>:1), which has no missing parents
16/03/08 13:50:41 INFO MemoryStore: Block broadcast_0 stored as values in memory (estimated size 5.4 KB, free 5.4 KB)
16/03/08 13:50:41 INFO MemoryStore: Block broadcast_0_piece0 stored as bytes in memory (estimated size 2.6 KB, free 7.9 KB)
16/03/08 13:50:41 INFO BlockManagerInfo: Added broadcast_0_piece0 in memory on @IPSLAVE (size: 2.6 KB, free: 511.5 MB)
16/03/08 13:50:41 INFO SparkContext: Created broadcast 0 from broadcast at DAGScheduler.scala:1006
16/03/08 13:50:41 INFO DAGScheduler: Submitting 1 missing tasks from ResultStage 0 (MapPartitionsRDD[1] at take at <stdin>:1)
16/03/08 13:50:41 INFO TaskSchedulerImpl: Adding task set 0.0 with 1 tasks
16/03/08 13:50:41 INFO TaskSetManager: Starting task 0.0 in stage 0.0 (TID 0, @IPSLAVE, partition 0,PROCESS_LOCAL, 1922 bytes)
16/03/08 13:50:41 INFO BlockManagerInfo: Added broadcast_0_piece0 in memory on @IPSLAVE (size: 2.6 KB, free: 511.5 MB)
16/03/08 15:02:20 INFO TaskSetManager: Finished task 0.0 in stage 0.0 (TID 0) in 4299240 ms on @IPSLAVE (1/1)
16/03/08 15:02:20 INFO TaskSchedulerImpl: Removed TaskSet 0.0, whose tasks have all completed, from pool 
16/03/08 15:02:20 INFO DAGScheduler: ResultStage 0 (take at <stdin>:1) finished in 4299.248 s
16/03/08 15:02:20 INFO DAGScheduler: Job 0 finished: take at <stdin>:1, took 4299.460581 s

正如您所看到的，这需要花费很长时间。我的表实际上非常大（存储大约2.2亿行，每行11个字段），但这样的查询将立即执行使用＆＃34; normal＆＃34; sql（例如pyodbc）。

我想我很想念/错过Spark，你会有这样的想法或建议让它更好用吗？

Answer 1

虽然Spark支持对JDBC的有限谓词下推，但所有其他操作（如限制，组，聚合）都在内部执行。不幸的是，这意味着public class BeaconService extends BaseService implements BootstrapNotifier, BeaconConsumer { private static final int NOTIFICATION = R.string.notify_service_started; private static final String TAG = "BeaconService"; private int size = -1; private RegionBootstrap regionBootstrap; private BackgroundPowerSaver backgroundPowerSaver; private Beacon beacon; @Nullable @Override public IBinder onBind(Intent intent) { return null; } @Override public void onCreate() { super.onCreate(); Log.d(TAG, "onCreate"); regionBootstrap = new RegionBootstrap(this, region); beaconManager.bind(this); backgroundPowerSaver = new BackgroundPowerSaver(getApplicationContext()); } @Override public int onStartCommand(Intent intent, int flags, int startId) { if (intent != null && intent.getAction() != null) { switch (intent.getAction()) { case Constants.ACTION.STARTFOREGROUND_ACTION: startForeground(NOTIFICATION, getNotification()); break; case Constants.ACTION.STOPFOREGROUND_ACTION: Log.d(TAG, "Received stop foreground request"); stopForeground(true); stopSelf(); break; } } return START_STICKY; } @Override public void onDestroy() { beaconManager.unbind(this); regionBootstrap.disable(); Log.d(TAG, "service onDestroy"); } /** * Called when at least one beacon in a Region is visible. * * @param region region */ @Override public void didEnterRegion(Region region) { // TODO: 3/8/16 reload all the resource Log.d(TAG, "didEnterRegion called"); L.object(region); try { beaconManager.startRangingBeaconsInRegion(region); } catch (RemoteException e) { e.printStackTrace(); } } /** * Called when no beacons in a Region are visible. * * @param region region */ @Override public void didExitRegion(Region region) { // TODO: 3/8/16 close all the resource Log.d(TAG, "didExitRegion called"); try { beaconManager.stopRangingBeaconsInRegion(region); } catch (RemoteException e) { e.printStackTrace(); } beaconManager.unbind(this); regionBootstrap.disable(); L.object(region); } /** * Called with a state value of MonitorNotifier.INSIDE when at least one beacon in a Region is visible * * @param region region */ @Override public void didDetermineStateForRegion(int i, Region region) { Log.d(TAG, "switch from seeing/not seeing beacons"); L.object(region); } @Override public void onBeaconServiceConnect() { Log.d(TAG, "onBeaconServiceConnect"); if (null == beaconManager.getRangingNotifier()) { beaconManager.setRangeNotifier(new RangeNotifier() { @Override public void didRangeBeaconsInRegion(Collection<Beacon> beacons, Region region) { Log.d(TAG, "beacons.size():" + beacons.size() + "," + this); if (beacons.size() != 0) { Iterator<Beacon> iterator = beacons.iterator(); if (beacons.size() != size) { saveBeacon(iterator); size = beacons.size(); } } } }); } } /** * Save beacon p-o-j-o to SQLite. */ private void saveBeacon(Iterator<Beacon> iterator) { while (iterator.hasNext()) { beacon = iterator.next(); L.object(beacon); entity.setId(null); entity.setUuid(beacon.getId1().toString()); entity.setMajor(beacon.getId2().toString()); entity.setMinor(beacon.getId3().toString()); entity.setTxpower(beacon.getTxPower()); entity.setTime(Utils.getCurrentTime()); dbHelper.provideNinjaDao().insert(entity); Log.d(TAG, "sql save success"); } } private Notification getNotification() { CharSequence text = getText(R.string.notify_service_started); PendingIntent contentIntent = PendingIntent.getActivity(this, 0, new Intent(this, MainActivity.class), 0); Notification notification = new Notification.Builder(this) .setSmallIcon(R.mipmap.ninja_turtle) .setTicker(text) .setWhen(System.currentTimeMillis()) .setContentTitle(getText(R.string.info_service)) .setContentText(text) .setContentIntent(contentIntent) .build(); return notification; } }将首先获取数据，然后应用take(4)。换句话说，您的数据库将执行（假设没有投影过滤器）等同于：

limit

其余的将由Spark处理。涉及一些优化（特别是Spark迭代地评估分区以获得SELECT * FROM table请求的记录数）但与数据库端优化相比，它仍然是非常低效的过程。

如果您想将LIMIT推送到数据库，则必须使用子查询作为limit参数静态执行此操作：

dbtable

(sqlContext.read.format('jdbc')
    .options(url='xxxx', dbtable='(SELECT * FROM xxx LIMIT 4) tmp', ....))

请注意，子查询中的别名是必需的。

注意：

一旦Data Source API v2准备就绪，将来可能会改进此行为：

SPARK-15689

SPIP: Data Source API V2

一个多小时执行pyspark.sql.DataFrame.take（4）

1 个答案: