This implementation of the AbstractTrafficShapingHandler is for global traffic shaping, that is to say a global limitation of the bandwidth, whatever the number of opened channels.

Note the index used in OutboundBuffer.setUserDefinedWritability(index, boolean) is 2.

The general use should be as follow:

Create your unique GlobalTrafficShapingHandler like:

GlobalTrafficShapingHandler myHandler = new GlobalTrafficShapingHandler(executor);

The executor could be the underlying IO worker pool

pipeline.addLast(myHandler);

Note that this handler has a Pipeline Coverage of "all" which means only one such handler must be created and shared among all channels as the counter must be shared among all channels.

Other arguments can be passed like write or read limitation (in bytes/s where 0 means no limitation) or the check interval (in millisecond) that represents the delay between two computations of the bandwidth and so the call back of the doAccounting method (0 means no accounting at all).

A value of 0 means no accounting for checkInterval. If you need traffic shaping but no such accounting, it is recommended to set a positive value, even if it is high since the precision of the Traffic Shaping depends on the period where the traffic is computed. The highest the interval, the less precise the traffic shaping will be. It is suggested as higher value something close to 5 or 10 minutes.

maxTimeToWait, by default set to 15s, allows to specify an upper bound of time shaping.
In your handler, you should consider to use the channel.isWritable() and channelWritabilityChanged(ctx) to handle writability, or through future.addListener(new GenericFutureListener()) on the future returned by ctx.write().
You shall also consider to have object size in read or write operations relatively adapted to the bandwidth you required: for instance having 10 MB objects for 10KB/s will lead to burst effect, while having 100 KB objects for 1 MB/s should be smoothly handle by this TrafficShaping handler.
Some configuration methods will be taken as best effort, meaning that all already scheduled traffics will not be changed, but only applied to new traffics.
So the expected usage of those methods are to be used not too often, accordingly to the traffic shaping configuration.

Be sure to call release() once this handler is not needed anymore to release all internal resources. This will not shutdown the EventExecutor as it may be shared, so you need to do this by your own.

Nested and Inner Type Summary

Modifier and Type	Class and Description
private static class	GlobalTrafficShapingHandler.PerChannel
private static class	GlobalTrafficShapingHandler.ToSend

Field Summary

Modifier and Type	Field and Description
private final ConcurrentMap<Integer, GlobalTrafficShapingHandler.PerChannel>	channelQueues All queues per channel
pack-priv long	maxGlobalWriteSize Max size in the list before proposing to stop writing new objects from next handlers for all channel (global)
private final AtomicLong	queuesSize Global queues size

Inherited from io.netty.handler.traffic.AbstractTrafficShapingHandler:: CHANNEL_DEFAULT_USER_DEFINED_WRITABILITY_INDEX checkInterval DEFAULT_CHECK_INTERVAL DEFAULT_MAX_SIZE DEFAULT_MAX_TIME GLOBAL_DEFAULT_USER_DEFINED_WRITABILITY_INDEX GLOBALCHANNEL_DEFAULT_USER_DEFINED_WRITABILITY_INDEX maxTime maxWriteDelay maxWriteSize MINIMAL_WAIT READ_SUSPENDED REOPEN_TASK trafficCounter userDefinedWritabilityIndex

Constructor Summary

Access	Constructor and Description
public	GlobalTrafficShapingHandler(ScheduledExecutorService the `ScheduledExecutorService` to use for the `TrafficCounter`. executor, long 0 or a limit in bytes/s writeLimit, long 0 or a limit in bytes/s readLimit, long The delay between two computations of performances for channels or 0 if no stats are to be computed. checkInterval, long The maximum delay to wait in case of traffic excess. maxTime) Create a new instance.
public	GlobalTrafficShapingHandler(ScheduledExecutorService the `ScheduledExecutorService` to use for the `TrafficCounter`. executor, long 0 or a limit in bytes/s writeLimit, long 0 or a limit in bytes/s readLimit, long The delay between two computations of performances for channels or 0 if no stats are to be computed. checkInterval) Create a new instance using default max time as delay allowed value of 15000 ms.
public	GlobalTrafficShapingHandler(ScheduledExecutorService the `ScheduledExecutorService` to use for the `TrafficCounter`. executor, long 0 or a limit in bytes/s writeLimit, long 0 or a limit in bytes/s readLimit) Create a new instance using default Check Interval value of 1000 ms and default max time as delay allowed value of 15000 ms.
public	GlobalTrafficShapingHandler(ScheduledExecutorService the `ScheduledExecutorService` to use for the `TrafficCounter`. executor, long The delay between two computations of performances for channels or 0 if no stats are to be computed. checkInterval) Create a new instance using default max time as delay allowed value of 15000 ms and no limit.
public	GlobalTrafficShapingHandler(EventExecutor the `EventExecutor` to use for the `TrafficCounter`. executor) Create a new instance using default Check Interval value of 1000 ms and default max time as delay allowed value of 15000 ms and no limit.

Method Summary

Modifier and Type	Method and Description
pack-priv long	checkWaitReadTime(final ChannelHandlerContext ctx, long the wait delay computed in ms wait, final long the relative now time in ms now) Overrides io.netty.handler.traffic.AbstractTrafficShapingHandler.checkWaitReadTime. Method overridden in GTSH to take into account specific timer for the channel.
pack-priv void	createGlobalTrafficCounter(ScheduledExecutorService executor) Create the global TrafficCounter.
public long	Returns: the maxGlobalWriteSize default value being 400 MB. getMaxGlobalWriteSize()
private GlobalTrafficShapingHandler.PerChannel	getOrSetPerChannel(ChannelHandlerContext ctx)
public void	handlerAdded(ChannelHandlerContext ctx) Overrides io.netty.channel.ChannelHandlerAdapter.handlerAdded. Implements io.netty.channel.ChannelHandler.handlerAdded. Gets called after the `ChannelHandler` was added to the actual context and it's ready to handle events.
public void	handlerRemoved(ChannelHandlerContext ctx) Overrides io.netty.handler.traffic.AbstractTrafficShapingHandler.handlerRemoved. Implements io.netty.channel.ChannelHandler.handlerRemoved. Gets called after the `ChannelHandler` was removed from the actual context and it doesn't handle events anymore.
pack-priv void	informReadOperation(final ChannelHandlerContext ctx, final long the relative now time in ms now) Overrides io.netty.handler.traffic.AbstractTrafficShapingHandler.informReadOperation. Method overridden in GTSH to take into account specific timer for the channel.
public long	Returns: the global size of the buffers for all queues. queuesSize()
public final void	release() Release all internal resources of this instance.
private void	sendAllValid(final ChannelHandlerContext ctx, final GlobalTrafficShapingHandler.PerChannel perChannel, final long now)
public void	setMaxGlobalWriteSize(long the maximum Global Write Size allowed in the buffer globally for all channels before write suspended is set, default value being 400 MB. maxGlobalWriteSize) Note the change will be taken as best effort, meaning that all already scheduled traffics will not be changed, but only applied to new traffics.
pack-priv void	submitWrite(final ChannelHandlerContext ctx, final Object msg, final long size, final long writedelay, final long now, final ChannelPromise promise) Implements abstract io.netty.handler.traffic.AbstractTrafficShapingHandler.submitWrite.
protected int	userDefinedWritabilityIndex() Overrides io.netty.handler.traffic.AbstractTrafficShapingHandler.userDefinedWritabilityIndex.

Field Detail

channelQueues	back to summary
private final ConcurrentMap<Integer, GlobalTrafficShapingHandler.PerChannel> channelQueues All queues per channel

maxGlobalWriteSize	back to summary
pack-priv long maxGlobalWriteSize Max size in the list before proposing to stop writing new objects from next handlers for all channel (global)

queuesSize	back to summary
private final AtomicLong queuesSize Global queues size