dubbo load balancing is implemented by all subclasses AbstractLoadBalance
RandomLoadBalance random
On a high probability of collision cross-section, but the call volume the greater the more evenly distributed, but also by the weight more evenly probability rights, in favor of the provider dynamically adjust the weights.
- Gets the number of invoker
- Get the first invoker weight, and copied to firstWeight
- Invoker collection cycle, the sum of all the weights thereof, and to copy totalWeight, if the weight is not equal, false is sameWeight
- If the right set of heavy invoker not all equal, then get a random number between 1 and totalWeight, assigned to offset property
- Invoker loop through the collection, and obtain the weights and offset subtraction, when the offset reduced to less than zero, it returns the inovker
- If the weights are equal, then the invoker taken directly set inside a random number in return
@Override
protected <T> Invoker<T> doSelect(List<Invoker<T>> invokers, URL url, Invocation invocation) {
int length = invokers.size(); // Number of invokers
boolean sameWeight = true; // Every invoker has the same weight?
int firstWeight = getWeight(invokers.get(0), invocation);
int totalWeight = firstWeight; // The sum of weights
for (int i = 1; i < length; i++) {
int weight = getWeight(invokers.get(i), invocation);
totalWeight += weight; // Sum
if (sameWeight && weight != firstWeight) {
sameWeight = false;
}
}
if (totalWeight > 0 && !sameWeight) {
// If (not every invoker has the same weight & at least one invoker's weight>0), select randomly based on totalWeight.
int offset = ThreadLocalRandom.current().nextInt(totalWeight);
// Return a invoker based on the random value.
for (int i = 0; i < length; i++) {
offset -= getWeight(invokers.get(i), invocation);
if (offset < 0) {
return invokers.get(i);
}
}
}
// If all invokers have the same weight value or totalWeight=0, return evenly.
return invokers.get(ThreadLocalRandom.current().nextInt(length));
}
RoundRobinLoadBalance poll
Provider exist request slow accumulation of problems, such as: the second machine is very slow, but did not hang, when transferred to a second request on the card in that, over time, all requests are transferred to the card in the second stage.
In the old version, Dubbo will determine a maximum weight and a minimum weight, if the weights are equal, then press modulo, each taking after the value plus one; If the weights are not equal, the order of redistribution according to the weights.
In the new version, this class has been restructured.
- Obtaining a map according to the ServiceKey MethodName example inside manner methodWeightMap + from this example, for the first time. If not, the method proceeds to, placed in the instance of a methodWeightMap in Examples and the acquired map named
- Through all invokers
- To get the current identifyString invoker as a key, to obtain the map weightedRoundRobin instance, if there are no add a map
- If weightedRoundRobin weight and current weight invoker of different weight, indicating that the weight has changed, you need to re-set
- Get current weightedRoundRobin invoker example corresponding the current, plus the current weight and the weight of the invoker
- Set last update time weightedRoundRobin
- 0 maxCurrent beginning is provided, if the current is greater than the current value of weightedRoundRobin the assignment maxCurrent
- After traversing the greatest weight will be right and the invoker invoker of selectedInvoker corresponding weightedRoundRobin assigned to the selectedWRR, also have the right to re-sum totalWeight
- Then selectedWRR in the current property minus totalWeight, and return selectedInvoker
This view is obviously not clear enough, let's give an example:
假定有3台dubbo provider:
10.0.0.1:20884, weight=2
10.0.0.1:20886, weight=3
10.0.0.1:20888, weight=4
totalWeight=9;
那么第一次调用的时候:
10.0.0.1:20884, weight=2 selectedWRR -> current = 2
10.0.0.1:20886, weight=3 selectedWRR -> current = 3
10.0.0.1:20888, weight=4 selectedWRR -> current = 4
selectedInvoker-> 10.0.0.1:20888
调用 selectedWRR.sel(totalWeight);
10.0.0.1:20888, weight=4 selectedWRR -> current = -5
返回10.0.0.1:20888这个实例
那么第二次调用的时候:
10.0.0.1:20884, weight=2 selectedWRR -> current = 4
10.0.0.1:20886, weight=3 selectedWRR -> current = 6
10.0.0.1:20888, weight=4 selectedWRR -> current = -1
selectedInvoker-> 10.0.0.1:20886
调用 selectedWRR.sel(totalWeight);
10.0.0.1:20886 , weight=4 selectedWRR -> current = -3
返回10.0.0.1:20886这个实例
那么第三次调用的时候:
10.0.0.1:20884, weight=2 selectedWRR -> current = 6
10.0.0.1:20886, weight=3 selectedWRR -> current = 0
10.0.0.1:20888, weight=4 selectedWRR -> current = 3
selectedInvoker-> 10.0.0.1:20884
调用 selectedWRR.sel(totalWeight);
10.0.0.1:20884, weight=2 selectedWRR -> current = -3
返回10.0.0.1:20884这个实例
protected <T> Invoker<T> doSelect(List<Invoker<T>> invokers, URL url, Invocation invocation) {
String key = invokers.get(0).getUrl().getServiceKey() + "." + invocation.getMethodName();
ConcurrentMap<String, WeightedRoundRobin> map = methodWeightMap.get(key);
if (map == null) {
methodWeightMap.putIfAbsent(key, new ConcurrentHashMap<String, WeightedRoundRobin>());
map = methodWeightMap.get(key);
}
int totalWeight = 0;
long maxCurrent = Long.MIN_VALUE;
long now = System.currentTimeMillis();
Invoker<T> selectedInvoker = null;
WeightedRoundRobin selectedWRR = null;
for (Invoker<T> invoker : invokers) {
String identifyString = invoker.getUrl().toIdentityString();
WeightedRoundRobin weightedRoundRobin = map.get(identifyString);
int weight = getWeight(invoker, invocation);
if (weight < 0) {
weight = 0;
}
if (weightedRoundRobin == null) {
weightedRoundRobin = new WeightedRoundRobin();
weightedRoundRobin.setWeight(weight);
map.putIfAbsent(identifyString, weightedRoundRobin);
weightedRoundRobin = map.get(identifyString);
}
if (weight != weightedRoundRobin.getWeight()) {
//weight changed
weightedRoundRobin.setWeight(weight);
}
long cur = weightedRoundRobin.increaseCurrent();
weightedRoundRobin.setLastUpdate(now);
if (cur > maxCurrent) {
maxCurrent = cur;
selectedInvoker = invoker;
selectedWRR = weightedRoundRobin;
}
totalWeight += weight;
}
if (!updateLock.get() && invokers.size() != map.size()) {
if (updateLock.compareAndSet(false, true)) {
try {
// copy -> modify -> update reference
ConcurrentMap<String, WeightedRoundRobin> newMap = new ConcurrentHashMap<String, WeightedRoundRobin>();
newMap.putAll(map);
Iterator<Entry<String, WeightedRoundRobin>> it = newMap.entrySet().iterator();
while (it.hasNext()) {
Entry<String, WeightedRoundRobin> item = it.next();
if (now - item.getValue().getLastUpdate() > RECYCLE_PERIOD) {
it.remove();
}
}
methodWeightMap.put(key, newMap);
} finally {
updateLock.set(false);
}
}
}
if (selectedInvoker != null) {
selectedWRR.sel(totalWeight);
return selectedInvoker;
}
// should not happen here
return invokers.get(0);
}
LeastActiveLoadBalance minimum number of active calls
So that less slow provider receives a request, because the greater the difference between before and after the call will be slower provider counts.
- Through all the invoker
- Gets the current number of active invoker, the call is RpcStatus the getStatus method, active filter which will record the number of each method
- Get the current invoker weight
- The first time you come in or the current number of active invoker is smaller than the minimum number of active
- So the leastActive set to the current number of active invoker, set 1, a position leastIndexes array is set to 1 leastCount, record what totalWeight and firstWeight
- If the condition of point 4, it is judged that the number of currently active and the minimum number of active invoker are equal
- If the meet point 6, then the current weight is added to the totalWeight in and put leastIndexes array record at the same minimum number of active index; let's look at whether all of the same weight
- If the invoker invoker only one set of active minimal, less direct return
- If the weights are not equal, the random weight determining which of the heavy weight range of Invoker
- Equally weighted, can directly randomly selected Invoker
最小活跃数算法实现:
假定有3台dubbo provider:
10.0.0.1:20884, weight=2,active=2
10.0.0.1:20886, weight=3,active=4
10.0.0.1:20888, weight=4,active=3
active=2最小,且只有一个2,所以选择10.0.0.1:20884
假定有3台dubbo provider:
10.0.0.1:20884, weight=2,active=2
10.0.0.1:20886, weight=3,active=2
10.0.0.1:20888, weight=4,active=3
active=2最小,且有2个,所以从[10.0.0.1:20884,10.0.0.1:20886 ]中选择;
接下来的算法与随机算法类似:
假设offset=1(即random.nextInt(5)=1)
1-2=-1<0?是,所以选中 10.0.0.1:20884, weight=2
假设offset=4(即random.nextInt(5)=4)
4-2=2<0?否,这时候offset=2, 2-3<0?是,所以选中 10.0.0.1:20886, weight=3
1: public class LeastActiveLoadBalance extends AbstractLoadBalance {
2:
3: public static final String NAME = "leastactive";
4:
5: private final Random random = new Random();
6:
7: @Override
8: protected <T> Invoker<T> doSelect(List<Invoker<T>> invokers, URL url, Invocation invocation) {
9: int length = invokers.size(); // 总个数
10: int leastActive = -1; // 最小的活跃数
11: int leastCount = 0; // 相同最小活跃数的个数
12: int[] leastIndexes = new int[length]; // 相同最小活跃数的下标
13: int totalWeight = 0; // 总权重
14: int firstWeight = 0; // 第一个权重,用于于计算是否相同
15: boolean sameWeight = true; // 是否所有权重相同
16: // 计算获得相同最小活跃数的数组和个数
17: for (int i = 0; i < length; i++) {
18: Invoker<T> invoker = invokers.get(i);
19: int active = RpcStatus.getStatus(invoker.getUrl(), invocation.getMethodName()).getActive(); // 活跃数
20: int weight = invoker.getUrl().getMethodParameter(invocation.getMethodName(), Constants.WEIGHT_KEY, Constants.DEFAULT_WEIGHT); // 权重
21: if (leastActive == -1 || active < leastActive) { // 发现更小的活跃数,重新开始
22: leastActive = active; // 记录最小活跃数
23: leastCount = 1; // 重新统计相同最小活跃数的个数
24: leastIndexes[0] = i; // 重新记录最小活跃数下标
25: totalWeight = weight; // 重新累计总权重
26: firstWeight = weight; // 记录第一个权重
27: sameWeight = true; // 还原权重相同标识
28: } else if (active == leastActive) { // 累计相同最小的活跃数
29: leastIndexes[leastCount++] = i; // 累计相同最小活跃数下标
30: totalWeight += weight; // 累计总权重
31: // 判断所有权重是否一样
32: if (sameWeight && weight != firstWeight) {
33: sameWeight = false;
34: }
35: }
36: }
37: // assert(leastCount > 0)
38: if (leastCount == 1) {
39: // 如果只有一个最小则直接返回
40: return invokers.get(leastIndexes[0]);
41: }
42: if (!sameWeight && totalWeight > 0) {
43: // 如果权重不相同且权重大于0则按总权重数随机
44: int offsetWeight = random.nextInt(totalWeight);
45: // 并确定随机值落在哪个片断上
46: for (int i = 0; i < leastCount; i++) {
47: int leastIndex = leastIndexes[i];
48: offsetWeight -= getWeight(invokers.get(leastIndex), invocation);
49: if (offsetWeight <= 0) {
50: return invokers.get(leastIndex);
51: }
52: }
53: }
54: // 如果权重相同或权重为0则均等随机
55: return invokers.get(leastIndexes[random.nextInt(leastCount)]);
56: }
57:
58: }
ConsistentHashLoadBalance consistency Hash
Request is always sent to the same parameters of the same provider. When a station linked to the provider, the original request sent to the provider, based on the virtual node, in equal shares to other providers, it will not cause dramatic changes.
- Based invokers set, the memory address of the object is calculated based on the hash value defined
- Get ConsistentHashSelector object. If it is empty, or change the definition of a hash value (Description invokers set of changes), created a new ConsistentHashSelector objects
- Call ConsistentHashSelector object's select method
1: public class ConsistentHashLoadBalance extends AbstractLoadBalance {
2:
3: /**
4: * 服务方法与一致性哈希选择器的映射
5: *
6: * KEY:serviceKey + "." + methodName
7: */
8: private final ConcurrentMap<String, ConsistentHashSelector<?>> selectors = new ConcurrentHashMap<String, ConsistentHashSelector<?>>();
9:
10: @SuppressWarnings("unchecked")
11: @Override
12: protected <T> Invoker<T> doSelect(List<Invoker<T>> invokers, URL url, Invocation invocation) {
13: String key = invokers.get(0).getUrl().getServiceKey() + "." + invocation.getMethodName();
14: // 基于 invokers 集合,根据对象内存地址来计算定义哈希值
15: int identityHashCode = System.identityHashCode(invokers);
16: // 获得 ConsistentHashSelector 对象。若为空,或者定义哈希值变更(说明 invokers 集合发生变化),进行创建新的 ConsistentHashSelector 对象
17: ConsistentHashSelector<T> selector = (ConsistentHashSelector<T>) selectors.get(key);
18: if (selector == null || selector.identityHashCode != identityHashCode) {
19: selectors.put(key, new ConsistentHashSelector<T>(invokers, invocation.getMethodName(), identityHashCode));
20: selector = (ConsistentHashSelector<T>) selectors.get(key);
21: }
22: return selector.select(invocation);
23: }
24: }
ConsistentHashSelector consistent hashing selector
ConsistentHashSelector, is ConsistentHashLoadBalance inner class, consistent hashing selector, based Ketama algorithm.
/**
* 虚拟节点与 Invoker 的映射关系
*/
private final TreeMap<Long, Invoker<T>> virtualInvokers;
/**
* 每个Invoker 对应的虚拟节点数
*/
private final int replicaNumber;
/**
* 定义哈希值
*/
private final int identityHashCode;
/**
* 取值参数位置数组
*/
private final int[] argumentIndex;
1: ConsistentHashSelector(List<Invoker<T>> invokers, String methodName, int identityHashCode) {
2: this.virtualInvokers = new TreeMap<Long, Invoker<T>>();
3: // 设置 identityHashCode
4: this.identityHashCode = identityHashCode;
5: URL url = invokers.get(0).getUrl();
6: // 初始化 replicaNumber
7: this.replicaNumber = url.getMethodParameter(methodName, "hash.nodes", 160);
8: // 初始化 argumentIndex
9: String[] index = Constants.COMMA_SPLIT_PATTERN.split(url.getMethodParameter(methodName, "hash.arguments", "0"));
10: argumentIndex = new int[index.length];
11: for (int i = 0; i < index.length; i++) {
12: argumentIndex[i] = Integer.parseInt(index[i]);
13: }
14: // 初始化 virtualInvokers
15: for (Invoker<T> invoker : invokers) {
16: String address = invoker.getUrl().getAddress();
17: // 每四个虚拟结点为一组,为什么这样?下面会说到
18: for (int i = 0; i < replicaNumber / 4; i++) {
19: // 这组虚拟结点得到惟一名称
20: byte[] digest = md5(address + i);
21: // Md5是一个16字节长度的数组,将16字节的数组每四个字节一组,分别对应一个虚拟结点,这就是为什么上面把虚拟结点四个划分一组的原因
22: for (int h = 0; h < 4; h++) {
23: // 对于每四个字节,组成一个long值数值,做为这个虚拟节点的在环中的惟一key
24: long m = hash(digest, h);
25: virtualInvokers.put(m, invoker);
26: }
27: }
28: }
29: }
public Invoker<T> select(Invocation invocation) {
// 基于方法参数,获得 KEY
String key = toKey(invocation.getArguments());
// 计算 MD5 值
byte[] digest = md5(key);
// 计算 KEY 值
return selectForKey(hash(digest, 0));
}
private String toKey(Object[] args) {
StringBuilder buf = new StringBuilder();
for (int i : argumentIndex) {
if (i >= 0 && i < args.length) {
buf.append(args[i]);
}
}
return buf.toString();
}
private Invoker<T> selectForKey(long hash) {
// 得到大于当前 key 的那个子 Map ,然后从中取出第一个 key ,就是大于且离它最近的那个 key
Map.Entry<Long, Invoker<T>> entry = virtualInvokers.tailMap(hash, true).firstEntry();
// 不存在,则取 virtualInvokers 第一个
if (entry == null) {
entry = virtualInvokers.firstEntry();
}
// 存在,则返回
return entry.getValue();
}