3

The various NoSQL database systems all make their own specific choices and tradeoffs, making them more or less suitable for different use-cases. Fortunately, there are many good comparisons (also here on Stackoverflow) that explain and contrast the ideas, strengths and weaknesses of all major players.

But I cannot find any independent information about what Oracle NoSQL Database is, and where it fits in the NoSQL landscape. There is not even a Wikipedia page.

So, what is it, and how does it work?

Thilo
  • 257,207
  • 101
  • 511
  • 656

3 Answers3

2

Here is what I have found so far:

  • It is a key-value store based on the (potentially modified) BerkeleyDB engine

  • It manages a cluster of nodes across which the data is sharded and/or replicated

  • It has the concept of major keys and minor keys. You can get atomic commits for the minor keys belonging to the same major key.

  • The values associated to the keys are arbitrary binary data.

  • Data is distributed across the storage using a hash function. As a result, you can apparently only iterate in on unordered fashion, and not do range searches (you can iterate minor keys of a given major key in order).

  • You can configure durability by specifying how many replicas you want to have the data you are saving committed to

  • It uses single-master replication, so you have a master with the guaranteed latest version of the data, and read-only slaves which may be slightly behind

  • It is written in Java

  • There is a commercial and an open-source community edition (the latter is currently pending release while a license review is going on)

Thilo
  • 257,207
  • 101
  • 511
  • 656
2

This is the most information I have seen about it, from a presentation at the High Performance Transaction Systems Conference:
http://hpts.ws/agenda.html
http://hpts.ws/sessions/cwl-hpts-for-website.pdf

From slide 9:

What the Programmer Sees

  • Simple data model – key-value pair (major/minor key paradigm)
  • Simple operations – CRUD, RMW (CAS), iteration
  • Conflict resolution not required
  • ACID transactions for records within a major key, single API call
  • Unordered scan of all data (non-transactional)
  • Ordered iteration across sub keys within a key
  • Consistency (read) and durability (write) spec’d per operation
Zanson
  • 3,991
  • 25
  • 31
0

/. picking up on the issue right now, see First Look: Oracle NoSQL Database

after reading some of the comments I'm lead to believe Oracle just had to offer something in that category as well.

Update after reading Zansons pdf:

It's basicly built upon Berkeley DB and adds dynamic partitioning (“sharding”) ... "No single point of failure" sounds very impressive.

Does that mean it competes with Enterprise Edition partitioning and RAC? As a free (OSS) product?? Are they serious?

HAL 9000
  • 3,877
  • 1
  • 23
  • 29