Analyst: Oracle or will force opponents to push large data bundle systems

Earlier this month, Oracle began shipping large data machines (Oracle http://www.aliyun.com/zixun/aggregation/13527.html ">BIG Data Appliance"), analysts said. This will force major competitors such as IBM, HP and SAP to come up with Hadoop products that tightly bundle hardware, software and other tools.

On the day of shipment, Oracle announces that its new product will run Cloudera's Apache Hadoop implementation and management software.

Tightly bundled large data machines include Sun server hardware (18 linux-based x86 Sun servers configured with 216 processor cores, 864GB RAM and 648TB of original hard disk space) and Oracle NoSQL database, open source distributed r statistics software, and Oracle Java hotspot virtual machine copy.

Large data machines provide 40Gb InfiniBand access between multiple nodes, implementing Hadoop deployments, many of which are connected to nodes over Ethernet. Oracle Data Warehouse VP of product Management George Lumpkin points out that multiple racks can be configured as a single cluster.

Lumpkin added that Oracle is also starting to ship large data connectors this month, which is used for big data Connector and several other Oracle products, including Oracle database 11g, exadata data warehouses, Data exchange between online transaction processing equipment and new exalytics business intelligence machines.

Last fall, Oracle published large data machines and NoSQL databases at the OpenWorld user Conference.

David Menninger, an analyst at Ventana, said the product is more likely to encourage IT managers to keep a close eye on the bundled systems that provide "one-stop services" to the company's data requirements.

Oracle systems are designed to manage and analyze datasets such as telemetry data, click Stream data, or other log data sets that are often too large or unsuitable to be stored in a database, Menninger said.

In fact, if IT managers accept bundled Hadoop systems, the mainstream technology providers that have put a lot of effort into big data development will have to emulate it, says James Kobielus, an analyst at Forrester Research. He added that [multi-bundle systems will] be very useful to the market and provide customers with more choices of available products.

So far, most enterprise-class Hadoop deployments have been done by internal IT engineers, Kobielus says, and companies have to buy hardware and software licenses and then integrate.

Kobielus said the growing popularity of the open source Apache Hadoop Framework will further increase the choice of large packets for IT managers. Oracle's decision to adopt Cloudera Hadoop deployment will be beneficial for both enterprise users and Oracle itself. When Oracle first announced a large data machine, it was expected to run on Oracle's own Hadoop.

Menninger added that the choice of a already accepted Hadoop deployment meant that Oracle did not have to struggle to create another deployment, and if Oracle and its own deployment were created, it would expose a potential weakness that could be captured by competitors.

Original link: ZDNet

(Responsible editor: admin)