There is an application server (WildFly), it is synchronized by transaction number, i.e. That one transaction was not processed simultaneously in two flows.

There is a need to make a cluster with a second server, respectively, the question arises - how to make synchronization between servers? Those. for example, several requests came at the same time to different servers for a single transaction.

What are the most popular ready-made solutions? You can make your decision, but do not want to reinvent the wheel. DB / Lok file / Terracota not to offer.

  • there remains a client-server part to do. The server itself will send IDs to clients and tell them what to do. and the client back to the server is the result - Senior Pomidor
  • one
    Well, what does it mean not to offer if you yourself request a distributed lock? - etki
  • 2
    Huge selection of popular ready-made solutions: Hazelcast, Redis, GridGain (Apache Ignite), Zookeeper, Apache Helix. Virtually any in-memory storage. - Nofate ♦
  • one
    You might like Helix - a framework specifically created for building distributed applications. But he uses Zookeeper, and that needs three nodes for work. Redis is easy to start out of the box. - Nofate ♦
  • one
    @Nofate Redis is a very bad choice, it is extremely unreliable, it should not be used for something more serious than a wooden cache. Zookeper / Hazelcast is much better. - etki

0