Database vendors Couchbase Inc. and VoltDB Inc. are addressing big data application needs with new approaches to...
boosting database scalability. For example, a new release from Couchbase allows different types of database operations to be processed independently on a per-node basis. Additional support for Hadoop ecosystem streaming data tools is also on the docket for both companies.
NoSQL database vendor Couchbase this week detailed plans to add a "multidimensional scaling" feature as part of its upcoming Couchbase Server 4.0 release. The company said the new option will enable Couchbase users to run querying, indexing and data read-write services separately from one another on different server nodes in clustered systems.
Such optimized provisioning can improve the performance of common database actions, according to Couchbase CEO Bob Wiederhold. This isn't a multidimensional database of the kind that came to prominence in the era of online analytical processing (OLAP), he said. Instead, it's an approach that lets database developers tune their architectures for different applications.
"Some applications need to do quick reads and writes in the database -- others may need to build indexes, while others need to support queries," Wiederhold explained. "Building queries is very CPU-intensive, while building indexes is very disk-intensive. So, if you run them all on the same server, you can't tune for each operation."
The assignment of hardware resources can be configured on the fly while applications are running and modified as processing requirements change, said Wiederhold, who added that the new option is backward-compatible with existing Couchbase implementations. Couchbase Server 4.0 is scheduled for release this summer; other planned features include a SQL implementation that was codename N1QL (pronounced "nickel") and will be called SQL for Documents.
More flexible database tuning is something Couchbase is pushing in an effort to differentiate itself from rivals: With release 3.0 of the database last year, it also began to support a tunable memory process for handling data on disk or in-memory, depending on application needs. Meanwhile, the 4.0 announcement comes on the heels of Couchbase's move last month to increase support for Hadoop-related technologies such as the Storm streaming data engine and Kafka publish-and-subscribe messaging system.
When NoSQL databases rose up during the great leap to large-scale, Web-based applications, they were built simply with the main purpose to be fast. Going forward, NoSQL vendors are likely to add other traits in an effort to enhance their offerings for wider use -- with Couchbase's focus on database scalability serving as an example.
VoltDB 5.0 has Hadoop hooks, Kafka connections
In-memory database vendor VoltDB is also looking to expand the utility of its NewSQL software with improved support for applications that connect to Hadoop and related Apache software. To that end, VoltDB Version 5.0, released earlier this year, adds an export connector to the Hadoop Distributed File System, plus a Hadoop bulk data import option and Apache Kafka export and import links.
Interfacing with Hadoop applications is increasingly important among VoltDB customers looking to handle vast amounts of Web and mobile data, said Peter Vescuso, the vendor's chief marketing officer.
"The data flowing into an organization today represents a different type of application segment," Vescuso said -- one that VoltDB refers to as fast data or data in motion. While noting the value of Hadoop ecosystem components for handling such data, he suggested that the VoltDB server software can offer development teams a palette for simpler application creation, as well as faster performance and ACID compliance.
Recently announced VoltDB user applications reinforce the fast data theme. They include Myfox, a home security systems provider that employs VoltDB for real-time analytics on security data; Snapcard, which uses the software as part of its digital-currency payment processing systems; and AsiaInfo, which includes VoltDB in its Veris C3 analytics offerings for mobile marketing uses.
As development teams string together separate open source components, they may opt to include offerings like VoltDB in which data handling is tightly integrated from the ingestion to the storage stages, Vescuso said. That has become more feasible with VoltDB's latest update, which is positioned to play amid diverse tools and varied streams of information.
Learn some NoSQL basics
Check out our guide to new database technologies
See where NewSQL can trump NoSQL