[3/3] What Goes Around Comes Around... And Around... (Data column)

Continuing story (1 and 2) article 2024 It is worth finishing about different approaches to system architecture, and then tell about the authors’ farewell comments, which partially repeat the warnings of the authors. 2005 When the first article was published,What Goes Around Comes Around"

Architecture 5. Hardware accelerators This approach involves the use of specialized equipment. (For example, GPU or FPGA) to speed up the execution of requests. The approach looks workable only for cloud vendors because of the cost of developing custom iron and writing custom software. In addition, this usually does not give a multiple increase in efficiency. 6. Blockchain-based databases Blockchain promised guarantees of data immutability and transparency of transactions. But such systems remain niche due to performance limitations and the difficulty of integrating with traditional applications.

Findings relating to database architectures The authors note that many of these architectures have either occupied niche markets or are gradually being integrated into relational databases. Column systems are now the standard for analytics. Cloud databases dominate due to the ease of management. NewSQL systems become a bridge between traditional relational bases and scalable NoSQL approaches. These changes demonstrate the DBMS adaptation to modern application requirements and hardware innovations.

Farewell comments **Never underestimate the importance of good marketing to bad products.**Oracle in the 1980s, MySQL in the 2000s, and MongoDB in the 2010s:) The latest. 2 I saw an example with my own eyes. Beware of DBMSs from large vendors outside the DBMS market. Big tech companies often write their databases in-house and then let them go. (open-source). Some big tech companies make cool products like this: Apache Hive, Presto, Apache Cassandra, RocksDB or Apache Kafka, Apache Pinot, VoldemortThe others don't. The authors suggest that the promotion system within such companies encourages the creation of new technologies internally, rather than the use of ready-made tools. So there's a bunch of freaks from teams who don't have any experience in building databases. Don’t ignore the user’s first impression with your database. (out-of-box experience). It is important that the product is convenient to use a beginner. n (Otherwise, they will go spinning data using Python notebooks.) **Developers still need to sometimes make database queries directly.**Despite the efforts of the creators of ORM:) The impact of AI/ML on databases will be significant.. Most likely, natural language queries will not replace SQL for OLTP query types, but for OLAP, this may be a matter of the near future. Within corporations, data is used to make decisions, but LLMs will not be easy to help with this. There is a problem with explaining their results to people, as well as the amount of data that is required for learning. And this data is not so easy to send for generation in the crowd system. Separately, the authors note the direction of research on the optimization of the DBMSs themselves using ML/AI. And even if the results are obtained, they will not eliminate the need for high-quality system engineering.

In conclusion, the authors threaten to 20 Years of writing an article

We caution developers to learn from history. In other words, stand on the shoulders of those who came before and not on their toes. One of us will likely still be alive and out on bail in two decades, and thus fully expects to write a follow-up to this paper in 2044. #Architecture #Software #DistributedSystems