October 2018
Beginner to intermediate
348 pages
10h
English
Another table setting that can help to tune performance is bloom_filter_fp_chance. Remember from earlier chapters that the bloom filter is a probabilistic data structure. Its job in the Cassandra read path is to provide guidance during query time. Specifically, the bloom filter is capable of indicating whether or not the requested data might be present in a particular SSTable file, or that it definitely is not. The false-positive (fp) chance is the main tunable property of the bloom filter. It defaults to a value of 0.01 (1%), unless the table is using Leveled compaction, and then it starts at 0.1 (10%).
Systems with SSDs can handle a higher value (of a false positive), as disk I/O is less of a bottleneck. ...