Miscellaneous about solrcloud

Miscellaneous about solrcloud

When I read SOLR in action, I saw the introduction to SOLR deficiency: "One final limitation of SOLR worth mentioning is its elastic Scalability: the ability to automatically add and remove servers and redistribute content to handle load. while SOLR scales well known SS servers, it doesn' t yet elastically scale by itself in a fully auto-matic way. so I couldn't help thinking about solrcloud deployment problems in the company.

1. solrcloud expansion problems, SOLR has obvious disadvantages for cluster expansion, and cannot achieve dynamic expansion and data load balancing. My company sells servers. For example, if I sold a server with a capacity of 0.5 billion data, the customer found that the data volume has exceeded 0.5 billion, then it will purchase another one and form a cluster with the previous one. In this case, the expansion problem is involved. How can we change from a single server to a cluster, and how can we balance one 0.5 billion data to each 0.25 billion data, how to ensure that the server still performs Online indexing and query operations during resizing. This has plagued me for half a year and has never been an effective strategy. This is a short board of solrcloud, but it is also a must for every SOLR application. It is said that elasticsearch has done a great job in this area. Next, you have to learn about elasticsearch to obtain the classics. I hope to find the solution soon.

2. solrcloud depends on zookeeper. In my disaster recovery and stress tests on solrcloud, there may be solrcloud crashes, or the shard either fails the recovery or is always in the recovery, preliminary estimates are related to the root zookeeper communication (of course, this is related to our use of zookeeper. Our servers run SOLR and zookeeper at the same time, and there is no way for us to sell servers, the Disaster Tolerance Performance of solrcloud is not as good as he said. More than 10 clusters need to be fully stabilized recently.

3. the book says that the query performance ratio of the cluster documentary server is as follows. Most of the cases are true, but the performance of aggregation overhead still affects the query of the cluster, for example, in extreme cases, paging + sorting queries. (Query speed on N + 1 indexes) = aggregation overhead + (query speed on N indexes)/(n + 1)

Miscellaneous about solrcloud