You are viewing an old version of this page. View the current version.
Compare with Current
View Page History
« Previous
Version 33
Next »
Overview
By rebalancing the Apache Ignite cluster, the distribution of primary and backup data copies would be balanced according to applied affinity function on the new set of peer nodes. Imbalanced data increases the likelihood of data loss and can affect peers utilization during data requests. On the other hand, a balanced set of data copies optimizes each peer requests load and each peer disk resources consumption.
Currently, there are two types of the Apache Ignite cluster rebalancing:
- In-memory batch rebalancing;
- Historical (WAL) rebalancing (the native presistence enabled);
Limitations of current approach
Regardless of which rebalance mode is used SYNC
or ASYNC
(defined in CacheRebalanceMode
enum), the Apache Ignite rebalance implementation has a number of limitations caused by a memory-centric desing architecture:
- Although all cache data is sent between peers in batches (
GridDhtPartitionSupplyMessage
used) it still processes entries one by one. Such an process has low impact with a pure in-memory Apache Ignite usage but it leads to additional fsyncs and logging WAL records with the native persistence enabled.
By default, setRebalanceThreadPoolSize
is set to 1
and setRebalanceBatchSize
to 512K
which means that thousands of key-value pairs will be processed single-thread and individually. Such an approach impacts on: - The extra unnecessary changes to keep node data structures up to date. Adding each entry record into
CacheDataStore
will traverse and modify each index tree N-times. It will allocate the space N-times within FreeList
structure and will have to additionally store WAL page delta records with approximate complexity ~ O(N*log(N))
; - Batch with N-entries will produce N-records in WAL which might end up with N fsyncs (assume fsync WAL mode configuration enabled);
- Increased the chance of huge JVM pauses. The more serving objects we produce by applying changes, the more often GC happens and the greater chance of JVM pauses arise;
The rebalancing procedure doesn't utilize the network and storage device throughout to its full extent even with enough meaningful values of setRebalanceThreadPoolSize
. For instance, trying to use a common recommendation of N+1
threads (N
– the number of CPU cores available) to increase rebalance speed will drammatically slowdown computation performance on demander node. This can be easily shown on the graphs below.
CPU utilization (supplier, demaner) |
---|
| |
setRebalanceThreadPoolSize – 9;
setRebalanceBatchSize – 512K; | setRebalanceThreadPoolSize – 1;
setRebalanceBatchSize – 512K; |
Advantages of peer-2-peer balancing
One and the most common case to which the peer-2-peer partition file balancing can by apply – is adding a completely new node or the set of new nodes to the cluster. Such a scenario implies fully relocation of cache partition files of all caches (suppose RendezvousAffinityFunction
used for all of them) to the new node. The partitition file transmitting over proposed low-level network socket communication signified the following fundamental things:
- All data stored in the single partition file will be transmitted within single batch (equal to partition file) much faster and without the serealization\deserialization overhead. To roughly estimate the superiority of partition file transmitting using network sockets the native Linux
scp\rsync
commands can be used. The test environment showed us results – 270 MB/s
over the current 40 MB/s
single-threaded rebalance speed; - The zero-copy file transmission can be used [1]. The contents of a file can be transmitted without copying them through the user space. Internally, it depends on the underlying operating system's support for zero copy. For instance, in UNIX and various flavors of Linux, the Java method
FileChannel.transfertTo()
call is routed to the sendfile()
system call;
Objective
Apache Ignite needs to support peer-2-peer cache partition file transfer using zero-copy algorithm based on extension of communication SPI.
Rebalance process overview
Streaming via CommunicationSpi
Handshake message
The handshake message patched to support new type of connection.
/** */
private static final byte PIPE_DATA_TRANSFER_MASK = 0x01;
/**
* @return If socket will be used to transfer raw files.
*/
public boolean usePipeTransfer() {
return (flags & PIPE_DATA_TRANSFER_MASK) != 0;
}
/**
* @param usePipeTransfer {@code True} if socket should be used to transfer raw files.
*/
public final void usePipeTransfer(boolean usePipeTransfer) {
flags = usePipeTransfer ?
(byte)(flags | PIPE_DATA_TRANSFER_MASK) : (byte)(flags & ~PIPE_DATA_TRANSFER_MASK);
}
Public API
Communication SPI
Communication SPI support new connection type to communicate with peers via sockets.
/**
* @return {@code True} if new type of direct connections supported.
*/
public default boolean pipeConnectionSupported() {
return false;
}
/**
* @param src Source cluster node to initiate connection with.
* @return Channel to listen.
* @throws IgniteSpiException If fails.
*/
public default ReadableByteChannel getRemotePipe(ClusterNode src) throws IgniteSpiException {
throw new UnsupportedOperationException();
}
/**
* @param dest Destination cluster node to communicate with.
* @param out Channel to write data.
* @throws IgniteSpiException If fails.
*/
public default void sendOnPipe(ClusterNode dest, WritableByteChannel out) throws IgniteSpiException {
throw new UnsupportedOperationException();
}
Internal API
Tcp connection listener
Rebalance checkpointing on supplier
Recovery from temporary WAL on demander
Questions
References
- Zero Copy I: User-Mode Perspective – https://www.linuxjournal.com/article/6345
- Example: Efficient data transfer through zero copy – https://www.ibm.com/developerworks/library/j-zerocopy/index.html
- Persistent Store Overview#6.PartitionRecovery