I lately look at this article nosql-versus-rdbms and I'm not sure an excessive amount of about nosql and that i did not utilize it during my projects, and so i possess some questions:

What's the primary feature that nosql has over RDBMSs?

If you feel it is best than RDBMSs : Where and just how can one find out about it (books - lessons)?

I wish to be considered a DBA. What this career is going to be after moving to nosql?


The initial title of the technology before people began calling it "NoSQL" would be a distributed key/value store. This can be a much more descriptive title, and that i initially remember searching in internet marketing on and on "hey, awesome, I'll wager which will finish up being very helpful to numerous people." The word has since broadened to basically include "something that is not a relational database", truly, when many people discuss NoSQL, they're speaking about key/value stores.

Since the word NoSQL was created, it has been getting recommended like a silver bullet. I am thinking about items like Cassandra and follow their progress, but they're still immature technologies, and also to claim that they're "changing" SQL or RDBMSes generally (or that they'll soon) is specious reasoning at the best, otherwise an out-and-out lie.

Items and technologies fitting underneath the NoSQL umbrella are targeted toward the next problem domain:

  • You intend to deploy a sizable-scale, high-concurrency database (100s of GB, 1000's of customers)
  • Which does not need Acidity guarantees
  • Or associations or constraints
  • Stores a reasonably narrow group of data (the same as 5-10 tables in SQL)
  • Is going to be running on commodity hardware (i.e. Amazon . com EC2)
  • Must be implemented on the really low budget and "scaly out."

This really describes a great deal of internet sites today. Google and Twitter fit very nicely in to these needs. Will it really matter if your couple of tweets are lost or postponed? However, these specs affect nearly % of economic systems, that is such a high quantity of us focus on developing. Most companies have completely different needs:

  • Medium-to-large-scale databases (10-100 GB) with fairly low concurrency (100s of customers for the most part)
  • Acidity (particularly the A and C - Atomicity and Consistency) is really a hard requirement
  • Information is highly correlated (hierarchies, master-detail, histories)
  • Needs to store a large range of data - 100s or 1000's of tables aren't uncommon inside a stabilized schema (more for denormalization tables, data warehouses, etc.)
  • Operate on high-finish hardware
  • Plenty of capital available (in case your business has countless clients you'll be able to most likely find $25k approximately laying behind the couch).

High-finish SQL databases (SQL Server, Oracle, Teradata, Vertica, etc.) are equipped for vertical scaling, that like standing on machines with numerous of memory, fast I/O through SANs and SSDs, and also the periodic horizontal scaling through clustering (') and partitioning (HC).

"NoSQL" is frequently in comparison more positively to "SQL" in performance terms. But fully maxed-out, a higher-finish SQL database server or cluster will scale almost infinitely. That's the way they were supposed to have been used. Watch out for dubious benchmarks evaluating poorly-stabilized, poorly-indexed SQL databases running mysql on entry-level servers (or worse, cloud servers like Amazon . com EC2) to similarly-used NoSQL databases. Apples and oranges. If you use SQL, you shouldn't be scared with that hype.

SQL is not going anywhere. DBAs aren't any more prone to vanish consequently of NoSQL than PHP developers were consequently of Java and XML.

NoSQL is not going anywhere either, since the development community has properly recognized that RDBMSes aren't always the perfect means to fix every problem.

So, like a developer your debt it to you to ultimately a minimum of learn what NoSQL is, what items it describes (Cassandra, BigTable, Voldemort, db4o, etc.), and just how to construct and code against an easy database produced using these. Try not to start throwing out all of your SQL databases yet or convinced that your job will probably be made obsolete - that's hype, not reality.