5 Approaches to Scalable Storage Solutions

Thinking about the scalable storage solutions market, I’ve ended up dividing it in 5 categories:

1. NoSQL databases

   • they abandoned all relational model concepts, building everything from scratch
   • most of the solutions were born to solve practical problems
   • they make a set of trade offs to achieve scalability

2. New age RDBMS implementations and appliances

   • they preserve almost completely the relational concepts, but are building everything from scratch
   • sometimes they are required to make trade offs to achieve scalability
   • they try to preserve existing APIs or connectivity models

3. Classical RDBMS

   • these well know solutions are either moderately scalable or not so cost effectively scalable

4. Scalability layers on top of Classical RDBMS

   • they reuse an existing RDBMS or engine and create different layers on top in an attempt to achieve scalability

5. Repurposed/Repositioned solutions

   • the products in this category have started in a different space, usually caching or grid computing
   • benefiting from Moore’s law, they usually tout the “RAM is the new disk slogan”
   • most of the time they use (memory) replication to provide durability
   • some have added disk persistency as additional layer of the original product
   • having all data in memory, they are usually able to mimic various access patterns (including SQL or key-value APIs)

Any other categories I’ve left out?

While not dismissing any of these, I’m more interested in solutions in categories 1 and 2 above.

Original title and link: 5 Approaches to Scalable Storage Solutions (NoSQL databases © myNoSQL)