Efficient compression and quick decoding of sorted integer lists are fundamental in search engine internals, particularly in systems like Elasticsearch, where Lucene indices split data into segments. These segments use document identifiers (doc IDs) as indexes, which are integral to inverted indexes that map terms to postings lists. Utilizing techniques like delta-encoding, Lucene compresses these postings lists by encoding blocks of doc IDs, a method known as Frame Of Reference. Meanwhile, filter caching, which speeds up the execution of frequent filters, benefits from various encoding techniques. While integer arrays are simple but memory-heavy, bitmaps are efficient for dense sets but less so for sparse ones. Roaring bitmaps, which combine the advantages of arrays and bitmaps, dynamically choose between them based on memory efficiency, making them appealing for filter caching despite not always being the fastest. The analysis reveals that while roaring bitmaps are not consistently superior, they offer a balanced trade-off between performance and memory usage, suggesting their utility in scenarios requiring efficient caching, especially when compared to traditional disk-based postings lists.