Elasticsearch is an excellent massive data search engine in the industry, with the advantages of distributed architecture, friendly interface, easy to use and easy to use, and has become the preferred solution for the above scenarios. At first, Weimeng used its own ES cluster. With the rapid growth of business, the number of clusters and the scale of business data have increased dramatically, and online services are facing more and more challenges:
1. Low delivery efficiency
To deliver a cluster, you need to purchase hardware, deploy software, and then parse or LB (load balancing). This part of work is highly repetitive and worthless.
2. Complex software and hardware configuration and changes
Software configuration, such as index management, requires self writing scripts to manage the life cycle of indexes; Hardware configuration, such as cluster expansion and contraction, involves a wide range of aspects, requires long verification process, requires manual attendance, and the efficiency of fault handling is low. For example, there was a fault because the network between the self built machines was not good, which led to a continuous fault. After several times of troubleshooting, it was found that the fault was caused by the network.
3. Data security is not guaranteed
Once hacker attacks lead to data leakage, the loss cannot be calculated.
4. High availability
The lack of a unified data backup platform will lead to disaster recovery hazards if the server goes down, and there is no multi active architecture in the same city to provide more secure system support.
5. Lack of professional technical support
R&D and O&M personnel have limited time and energy to dig deep and learn ES, and it is difficult to solve problems quickly without professional guidance. It takes time and effort to turn to search, and is not friendly to troubleshooting.