Recent data management product announcements by Hewlett-Packard, SAP and NuoDB add, respectively, increased support...
for SQL on Hadoop analytics processes, enhanced cloud deployment capabilities and the ability to run hybrid applications combining operational and analytical processing.
HP puts a SQL stamp on Hadoop via Vertica
Last week, HP's big data business group joined a rash of recent Apache Hadoop activity, releasing HP Vertica for SQL on Hadoop, which uses the company's Vertica technology to bring SQL analytics to Hadoop distributions from Hortonworks, MapR and Cloudera.
The release lets users keep their data in Hadoop but query it via a built-in HP Vertica Enterprise analytical engine running existing SQL code, according to Steve Sarsfield, product marketing manager for the big data unit HP is looking to tap into new revenue streams by providing SQL analytics capabilities to customers running Hadoop clusters, Sarsfield added.
"We're seeing adoption in the marketplace by people that are using Hadoop as a way to store data in a data lake," he said. "The reality is they're storing the data in Hadoop and using commodity servers."
That's a shift from a time when the data warehouse was the only game in town. IT shops want to reduce the cost of data storage, and data warehouse players like HP, Teradata, Actian and others are clearly taking notice.
Sarsfield also pointed to a shift in HP's pricing on this offering. HP Vertica for SQL on Hadoop is priced per node, he said. That's in keeping with the current trend in pricing for NoSQL and Hadoop technologies.
Multi-tenancy moves into HANA house
The SAP HANA in-memory computing platform was updated last month. Known as SAP HANA SPS 09, the latest version adds features to support multi-tenancy in cloud applications and dynamic-tier assignments for more cost-effective memory use.
At the heart of SAP's multi-tenancy scheme are database containers -- ones using an operating system-level virtualization method that doesn't employ software emulation. The approach means stronger separation and isolation of data in cases where multiple users share a database system, according to Mike Eacrett, head of HANA product management at SAP.
The dynamic-tier assignments capability is an interesting step for HANA, which primarily has been built around the notion that transactions and analytical queries should be processed in semiconductor memory. HANA users gain more flexibility with the new tiered approach, according to SAP.
"With 9.0, you can run 'hot' data in-memory. But you can pick which data you want to run 'warm' on-disk," Eacrett said. Though the media may vary, it's important that the data resides conceptually in a single data store.
The updated HANA platform, which is available now, also offers new data streaming and Hadoop options. A "smart streaming" feature lets users ingest data of immediate interest into memory while placing long-lived data into Hadoop for historical analysis. One new Hadoop option allows the development of user-defined functions that specifically enable HANA to call Hadoop MapReduce jobs directly in Hadoop.
NewSQL DB tuned up for mixed transactions, analytics
It's not too surprising to see analytical capabilities among the traits being added to distributed NewSQL databases -- products that seek to combine aspects of established SQL-based technologies with those of newer NoSQL database types.
After all, the relational databases that power transaction processing applications are often used for analytics, too. Analyst group Gartner Inc. has dubbed this phenomenon "hybrid transactional/analytical" processing.
"Traditional relational databases and most NoSQL databases were designed for the operations space, for what we used to call online transaction processing," said Seth Proctor, CTO at NewDB vendor NuoDB. "They've grown over the years to do analytics as well. The challenge is typically doing those on the same data set in the same database."
That is one of the factors behind design enhancements in NuoDB's latest version of its namesake distributed database. NuoDB Swifts Release 2.1 can dedicate specific transactional portions of the database to carry out operational jobs while dedicating other parts for running analytical queries. The overall architecture is fashioned so that operational and analytical workloads can run comfortably on the same database, Proctor said.
The 2.1 release also includes improvement to browser-based management tools, a REST API for metrics, and improvements and extensions to SQL indexing. It is available in a free community edition and a free-trial enterprise edition.
Find out why Adobe opted for SAP Data Services
Learn about cloud data management in NuoDB
Discover the details on running HP Vertica analytics in healthcare