Although Hadoop is of great benefit, the open source market is still not stable

Some financial analysts questioned Teradata's ability to continue its growth in the IT market. While this segment of customers does put Hadoop on the agenda, customers often choose to give up only if data warehouse vendors with decades of experience find it difficult.

"What's more, it's more important to get rid of the misunderstandings than the realities," Scott Gnau, president of Teradata Labs, pointed out in a recent interview - the focus of this interview was to explore the historical perspective of open source Hadoop for Teradata Software, equipment and related prices brought about what kind of impact. Due to management, security and performance issues, he said the current proposal to make use of Hadoop as an effective substitute for the Teradata system was "quite stupid."

In all fairness, Gnau thinks Hadoop and Teradata should complement each other. "The combined effect of our two sides is more prominent than any one-sided effect," he explained.

At least for now, such a conclusion is indeed more fair in its position. On Monday, Teradata unveiled a series of new products and new features. One of its features, called QueryGrid, allows users to write SQL queries in a Teradata environment and use it to automatically analyze data in Hadoop or Teradata Aster systems. The company also said it will support the semi-structured JSON file format, which means Teradata users will be able to extend their deployment to sensors, GPS tracking devices, web logs and even other popular machine-generated data sources.

This means that data types that are historically in need of being stored and analyzed by Hadoop (mainly NoSQL databases such as MongoDB for JSON) are now being analyzed by the Teradata program, and Hadoop continues to handle only the storage requirements can. In Teradata's view, the larger the size of data retained by enterprise users-including Hadoop-the more likely it is that valuable information exists. More importantly, the more likely an enterprise user will want to analyze it using a mature mission critical mission system like Teradata.

"Hammers can also screw screws into a wall," Gnau said of an example when it comes to building a SQL query engine for Hadoop, "but this is obviously not the best solution."

In fact, most current Hadoop vendors may agree. As a vendor who works closely with Teradata, Hortonworks, a start-up of Hadoop, never considered allowing its own solutions to act as a data platform to help users save and process their big data. Hortonworks products are designed with the intent to interface with more sophisticated analytics software - and of course their collaboration partners include Teradata.

"We will not force customers to make the tough choice of 'having to give up one solution and move on to another and hopefully everything will be fine,'" said Rob Bearden, CEO of Hortonworks, in a recent interview.

As another Hadoop provider, MapR is working on its open source SQL-on-Hadoop tools, but they also provide support for the HP Vertica Analytic Database in software. Even Cloudera, one of the most aggressive Hadoop vendors in the camp, claiming it will do a devastating blow to enterprise data warehouses such as Teradata, is still heavily dependent on its software partners - including those in data warehouses Solution vendors, which do not hurt their business value.

Being does not mean growth

However, this does not mean that Teradata or any other similar business (in fact most of Teradata's original competitors were acquired by other technology giants such as IBM, Hewlett-Packard, and EMC) could continue to weather this big data technology storm To ensure that their revenues intact - the opposite is true, and even in the face of the scale of its major customers crazy expansion orders, the negative effects still exist. The advent of Hadoop means that customers are able to hold more data in a relatively cheaper solution without having to pay for the ideal transfer rate per TB. The evolution from SQL to Hadoop means that customers can migrate part of their analytical workloads to lower-cost systems.

Intel expects Hadoop to eventually become the dominant workload running on its server processor. If this assertion turns out to be true, then Hadoop solutions that cover every aspect of the enterprise will force CIOs to create more pressing market demands for vendors who are able to deliver products. For now, EMC-VMware has partnered to create Pivotal Inc., apparently a re-bet, hoping customers will be able to buy their Hadoop solutions, databases, and data warehousing products from a single source. Currently, they are still adjusting their pricing policies to further translate their dreams into reality.

As a next-generation in-memory processing framework that runs on top of the Hadoop file system, the widespread concern sparked will further advance Hadoop. Spark includes a sub-project called Shark, which is designed to implement interactive in-memory SQL queries. Even business intelligence startups such as ClearStory Data and Platfora - both of which are sure to take a share of the market from Teradata - are involved, one for Spark and one for support.

Beyond Hadoop, database start-ups such as MemSQL and Citus Data are also struggling to create high-speed, distributed, relatively low-cost, SQL-enabled analytics databases that are also optimized for different formats and across multiple different storage systems Data analysis.

Teradata is probably the world's largest and best-performing supplier of pure analytics solutions, but the so-called "royalty has no eternity." One might wish to review IBM's history: While the mainframe is still being sold and serviced, blue It is clear that the Giants now no longer hold their future on such businesses. The technology giant even completely abandoned the server market. As the open source community (and the start-up that builds its own software solution) continues to proliferate, Teradata's ruling territory continues to suffer from nibbling; the company may have to make a surprising move to regain its dominance, Otherwise, its future prospects will be as precarious as IBM's mainframe business.