Improving Performance of Database Services with Multi-Dimensional Scaling

Scaling databases has changed over the years. Companies have begun to view scaling out as a better option than scaling up.

In a recent DBTA webcast, Anil Kumar, senior product management with Couchbase, tackled the topic of multi-dimensional scaling (MDS) within NoSQL database clusters. This method of scaling enables organizations to isolate specific database services of their database and scale those accordingly.

NoSQL databases contain three core workloads: core data processing, indexing, and querying.  “All of these workloads require a varying amount of resources from CPU to RAM, there are numerous methods to optimize latency and throughput for each,” stated Kumar.

 The scalability model today is a homogeneous scaling model with each node getting a slice of the workload. The issue with this process though is that workloads compete and interfere with each other. “Multi-dimensional scalability is the architecture that enables independent scaling of data, query, and indexing workloads,” stated Kumar.

The MDS allows for the different workloads to be isolated on separate nodes. Minimizing indexing and query overhead on core KV operations allows for best computational capacity per service. With the MDS all of the different services don’t have to be scaled together. Users can scale whichever individual workload they deem necessary.  

Couchbase servers with MDS enhance the scalability and performance of the database’s three core workloads, said Kumar, who finished the webinar with an in-depth demonstration of multi-dimensional scaling.  

To access a replay of the webinar, go here