What are the options and when should you use them?

The Emerging Database Landscape

Big Data is the buzzword of the day, but it pains our ears to hear it without any quantifier. Is 100 gigabytes Big Data? Is 100 terabytes big data? It’s all in the eye of the beholder, and it behooves the beholder to understand which options may best suit their storage and retrieval needs.

The database world is massive, and many low-cost options exist not only to save money but also to boost overall performance. Unfortunately, it’s difficult to understand which database to choose in your specific use case. In today’s world, we have three complementary database types to power your data warehouse: row-oriented, column-oriented, and NoSQL.

As the standard-bearer for traditional databases, row-oriented databases perform best in transaction environments such as inventory systems or CRM applications, delivering good performance when there are many updates and single-row inserts. Within the architecture, these databases store each column of one row consecutively on disk. As you insert a new row, update an existing row or delete an old row from the dataset, the row-oriented database can simply point to the beginning of that row on disk. From there, it acts similarly to an array: it can quickly modify/delete as it goes to one spot on disk for that entire row.

Row-oriented databases are robust, proven technologies to capture, input, retrieve, update and delete single records in a database. As data size moderately increases, additional overhead begins to bog the system down. Administrators typically turn to tuning or indexing their environment to squeeze out as much performance as possible – unfortunately, there comes a point at which administrative overhead (a.k.a. ‚the DBA tax‘), along with sluggishness of inserts and queries, tends to cause major technical hurdles.


Unsere Redaktion empfiehlt:

Relevante Beiträge

Meinungen zu diesem Beitrag

- Gib Deinen Standort ein -
- or -