NoSpinningDiscs: Why your NoSQL solution thirsts for SSDs

Organizations are always looking for opportunities to tune their architecture. To achieve massive scalability, many organizations have turned to various NoSQL solutions. But to blend performance with that scalability, organizations need to start incorporating solid state drives as well.

The enterprise databases of the future may be a lot quieter than those in use today. The whirring hard disk drives may fall silent as they are replaced by silent solid state drives. Dan McCreary, coauthor of Making Sense of NoSQL, say the migration to a solid state drive (SSD) is a big deal. However, not every organization is ready to make the switch. "There are a lot of existing databases that are only optimized for spinning media." What are they missing out on?

Benefits for enterprise DB use

There are a lot of business opportunities for organizations that are rewriting things from scratch to take advantage of solid state drives.

Dan McCreary, coauthor of Making Sense of NoSQL

SSDs have a lot to recommend them. Because solid state drives don't rely on any mechanical parts, the speed limitation associated with spinning disks simply isn't an issue. It's possible to get extremely fast reads with solid state—almost as fast as RAM. It's not uncommon to speed up queries as much as three orders of magnitude compared to a hard disk drive (HDD).

Along with this ultra-fast search and retrieval system comes the benefit of being able to index at will and access the index data rapidly. Indexing is no longer a process saved for special occasions or high-value use cases. According to McCreary, "It's changing the entire way organizations design and build databases."

Why SSD works best with NoSQL

Non-relational databases tend to be the easiest DBs to transition to SSD. Organizations that can break large sets of data into small pieces that look like documents are most likely to enjoy success with the solid state approach to storage. They can hash and cache data, especially large quantities of archived data, and still get amazingly fast retrieval times. The concept of keys, which can easily be kept in solid state memory, makes NoSQL the ideal candidate for migration to SSDs.

Older, relational databases don't lend themselves to a hash and cache approach. It's simply not realistic to do a check sum on a table of a million documents. As Dan says, "It's going to take a complete architectural redesign to make these DBs ready for solid state." That's not necessarily a bad thing for companies that are ready to make a serious change. "There are a lot of business opportunities for organizations that are rewriting things from scratch to take advantage of solid state drives."

It's not all or nothing

For some businesses, gaining access to the advantages of solid state is just one more reason to explore the benefits of a non-relational database. However, the thought of losing a substantial investment in existing hard drives might cause hesitation. Fortunately, as with the SQL vs. NoSQL choice, not all data has to be handled the same way. It may be possible to use HDD and SSD together to maximize storage and increase query speed based on the type of data being stored and how it is being used in the enterprise. This potential should be fully explored as part of the migration decision.

Have you been able to integrate SSDs into your architecture? Let us know your experience.

Dig Deeper on Java performance tuning

Start the conversation

Send me notifications when other members comment.

Please create a username to comment.