To avoid these disadvantages, the desire for HTAP systems that combine analytical and transaction processing in one system arose. One may naturally ask why this desire did not arise much earlier. There are essentially two reasons for this: Firstly, there are many more business processes in which the result of analyses is required as immediate feedback in order to be able to intervene in a controlling manner. In addition, the value of data often decreases over time. Secondly, there have been technological advances such as in-memory processing of data that make it easier to implement HTAP systems.
Nevertheless, these newer HTAP approaches also have to solve problems as mentioned above if one wants to operate OLTP and OLAP together in belize telegram screening one system: These include different optimal storage formats (row-oriented storage for OLTP systems versus column-oriented storage for OLAP systems) and different optimal auxiliary structures such as indexes, which are necessary or recommended in one case, superfluous or harmful in another.
One approach, especially to address the problem of outdated data in analytical queries, is to continue to have two copies for transaction and analytical processing, but at the same time increase the update rate. This can be done by significantly increasing the loading frequency of the data warehouse, for example by running the ETL processes every 5 minutes instead of once a day. This can be improved even further by not loading the data into the data warehouse using load commands, but by bringing the data from the transaction-processing source systems to the data warehouse using (log-based) asynchronous replication.