Current IBM Z processor family introduced a new technique for compressing data. It provides optimized on-chip compression and uses Huffman encoding to achieve better data compression efficiency than the prior fixed length algorithm. The IBM SAP on Z Performance Team conducted a study to see if there are any space saving benefits or significant system performance impact from the improved compression algorithm. This study used banking services from SAP, with a database of 100 million accounts. This is comparable to the number of accounts held by some of the largest banks in the world. The database was compressed using Huffman encoding and the space usage was compared to the same database compressed with fixed length encoding instead. Both the SAP Banking Account Settlement and Day Posting workloads were run to quantify the database growth, and any impact to the system performance with Huffman compression compared to fixed length compression. Using both a batch and an OLTP workload gave a more complete picture of potential performance impacts since these workloads have different characteristics. Both these workloads are good representations of customer workloads.

https://mydigitalpublication.com/publication/?i=690514&ver=html5&p=38