Hazelcast is an open source, transactional, distributed caching solution for Java. It is released under Apache license.
Hazelcast is actually a little more than a distributed cache; it is distributed implementation of queue, topic, map, multimap, lock, executor service for Java.
JBoss Cache and EHCache are decent clustered cache solutions. But none of them have all the features and a nice admin console like Tangosol Oracle Coherence which is indeed the most advanced product in this category. For the price, there is a "Buy Now" button on the right of the Oracle Coherence web page.
I have been investigating open source distributed caches and have focused on EHCahe and JBoss Pojo Cache.
They are both decent products, however the deal breaker is that my use case requires me to distribute an object graph and maintain shared references on all my nodes. EHCache cannot handle this. It uses standard java serialisation, and as such, shared references are lost.
e.g. if I have manager/employee object hierarchy, i want a single 'manager' object on all my nodes, with references to their employees. With EHCache, I dont get this, as the java serialiser copies all dependent objects. With JBoss Pojo Cache, my object graph is preserved.
The downside is that JBoss needs to instrument your classes (the primitives that make up your object graph are stored in a tree structure behind the scenes). This can cause some issues when running in an OSGi environment.
JBoss uses JGroups as its communication stack, you can choose between multicast or TCP/IP to manage communication between the nodes.
I have been watching for a good alternative also. The closest overall in features I have found are Terracotta and Redis but it would be helpful to know what aspects of Coherence are important to your project.
发布评论
评论（10）
Hazelcast 是一个面向 Java 的开源、事务性、分布式缓存解决方案。它是在 Apache 许可证下发布的。
Hazelcast 实际上不仅仅是一个分布式缓存；它是Java的队列、主题、映射、多重映射、锁、执行器服务的分布式实现。
Hazelcast is an open source, transactional, distributed caching solution for Java. It is released under Apache license.
Hazelcast is actually a little more than a distributed cache; it is distributed implementation of queue, topic, map, multimap, lock, executor service for Java.
Infinispan 看起来很有前途。
Infinispan is looking promising.
EhCache 提供了不错的复制缓存，但与 Coherence 提供的功能集相去甚远。
EhCache provides a decent replicated cache but has nowhere near the feature set Coherence provides.
JBoss 缓存 和 EHCache 是不错的集群缓存解决方案。但它们都没有像
TangosolOracle Coherence 这样的所有功能和漂亮的管理控制台，这确实是该类别中最先进的产品。对于价格，Oracle Coherence 右侧有一个“立即购买”按钮 网页。JBoss Cache and EHCache are decent clustered cache solutions. But none of them have all the features and a nice admin console like
TangosolOracle Coherence which is indeed the most advanced product in this category. For the price, there is a "Buy Now" button on the right of the Oracle Coherence web page.Memcached 也是一个很好的开源替代方案，被许多知名企业所使用。高流量/容量的应用程序。
Memcached is also an great open source alternative that is used by a lot of high-profile & high-traffic/volume apps.
我一直在研究开源分布式缓存，重点关注 EHCahe 和 JBoss Pojo Cache。
它们都是不错的产品，但最大的问题是我的用例要求我分发对象图并在所有节点上维护共享引用。 EHCache 无法处理这个问题。它使用标准的 java 序列化，因此共享引用会丢失。
例如，如果我有经理/员工对象层次结构，我希望所有节点上有一个“经理”对象，并引用其员工。
使用 EHCache，我无法理解这一点，因为 java 序列化程序会复制所有依赖对象。使用 JBoss Pojo Cache，我的对象图得以保留。
缺点是 JBoss 需要检测您的类（构成对象图的原语存储在幕后的树结构中）。在 OSGi 环境中运行时，这可能会导致一些问题。
JBoss 使用 JGroups 作为其通信堆栈，您可以选择多播或 TCP/IP 来管理节点之间的通信。
希望这有帮助。
I have been investigating open source distributed caches and have focused on EHCahe and JBoss Pojo Cache.
They are both decent products, however the deal breaker is that my use case requires me to distribute an object graph and maintain shared references on all my nodes. EHCache cannot handle this. It uses standard java serialisation, and as such, shared references are lost.
e.g. if I have manager/employee object hierarchy, i want a single 'manager' object on all my nodes, with references to their employees.
With EHCache, I dont get this, as the java serialiser copies all dependent objects. With JBoss Pojo Cache, my object graph is preserved.
The downside is that JBoss needs to instrument your classes (the primitives that make up your object graph are stored in a tree structure behind the scenes). This can cause some issues when running in an OSGi environment.
JBoss uses JGroups as its communication stack, you can choose between multicast or TCP/IP to manage communication between the nodes.
Hope this helps.
Project Voldemort 是 Coherence 的绝佳（免费）替代方案。
Project Voldemort is great (free) alternative to coherence.
我也一直在寻找一个好的替代方案。我发现的总体功能最接近的是 Terracotta 和 Redis 但了解 Coherence 的哪些方面对您的项目很重要会很有帮助。
I have been watching for a good alternative also. The closest overall in features I have found are Terracotta and Redis but it would be helpful to know what aspects of Coherence are important to your project.
看起来很有前途，因为 Google 在自己的产品中使用了它们
Looks promising as Google used them in their own products
检查“Berkeley DB”，它是开源的，工作原理与 Coherence 相同
Check for "Berkeley DB" which is a open source and works same as Coherence