By Tom Haunert
I believe the words “commodity system hardware” should describe something very simple. Being a commodity, system hardware should be easy to fit to a system design, source, assemble, configure, run, and manage. Based only on the words “commodity system hardware,” that’s what I think the experience should be.
I also believe that as a commodity, commodity system hardware should deliver optimal or near-optimal performance to the business quickly and easily, and that performance should improve in a linear fashion for every additional gigabyte or gigahertz or other measure of improvement in that new hardware. After all, commodity hardware is sold based on clearly enumerated specifications, and an increase in hardware capability should map simply and easily to business application performance improvements. Right?
But as businesses struggle to integrate commodity system hardware into their data centers now, the reality of “commodity” is not simple. The reality of sourcing, assembling, configuring, running, managing, and optimizing a collection of best-of-breed or any less-than-best mix of different commodity system hardware is complexity.
Get Real
In his “The Real-Time Enterprise” keynote at Oracle OpenWorld on October 1, 2014, Oracle Executive Vice President of Systems John Fowler opened with a reference to a quote from Oracle Executive Chairman and Chief Technology Officer Larry Ellison: “…the cardinal sin of the computing industry is the creation of complexity.”
Fowler talked about the complex competitive challenges that businesses face today, and offered a statement of support for Oracle customers. “We want to make it simpler to tackle complex problems,” said Fowler.
The road to simple solutions to complex business problems does not go through what Fowler called generic technologies from a range of vendors. “In order to get at truly simplifying large-scale business problems, you have to both improve the underlying technologies to have better than generic underlying technologies, but also aggregate these things,” said Fowler.
Underlying hardware technologies, and technologies that make revolutionary use of underlying hardware technologies, are a critical part of the Oracle strategy for simplifying large-scale business problems.
Get Specific
In Fowler’s keynote and in other sessions at Oracle OpenWorld, Oracle executives described key Oracle hardware technologies and technologies that use hardware in revolutionary ways, including Oracle Exadata, Oracle Big Data SQL, Oracle FS1, Oracle Exalytics In-Memory Machine, Oracle Database In-Memory, Oracle’s Zero Data Loss Recovery Appliance, and SPARC M7.
The original engineered system, Oracle Exadata, pioneered the idea of “smart storage”—storage server hardware enabled and optimized for database queries by smart storage software. Oracle Exadata’s smart storage supports the concept of smart scans, which moves significant volumes of query processing from system servers to storage servers and improves Oracle Database query processing time by orders of magnitude. The recently announced Oracle Big Data SQL also uses the technology pioneered in Oracle Exadata smart scans to move query processing to storage servers, delivering fast and integrated big data query processing across Oracle Database, Hadoop, and NoSQL data.
Oracle Exadata also blazed a trail with the addition of a different type of physical storage: fast PCI-based flash storage. Oracle’s commitment to flash is evident in the increased flash storage in each new generation of Oracle Exadata. That commitment is also clear in the new Oracle FS1 Series flash storage system, announced September 29, 2014. More than a flash storage array, Oracle FS1 Series is coengineered with Oracle servers, operating systems, applications, and databases for maximum operational efficiency.
As the price of system memory, DRAM, has fallen, the demand for faster answers from faster, in-memory computing has increased. The Oracle Exalytics In-Memory Machine was introduced in 2012 to deliver faster answers for business intelligence and enterprise performance management. In 2014, Oracle introduced Oracle Database In-Memory, which loads a database table column-store format into system memory for faster Oracle Database 12c queries and transactions—much, much faster queries and much faster transactions. And announced September 29, 2014, the updated Oracle Exalytics In-Memory Machine X4-4 added support for Oracle Database In-Memory.
Most general backup and recovery systems look at storage locations and files and not the type of content. On September 29, 2014, Oracle announced its Zero Data Loss Recovery Appliance, the world’s first and only engineered system designed specifically for Oracle Database protection.
Currently under development, the SPARC M7 processor is a CPU, not a system or an engineered system. But it is much more than a collection of the fastest SPARC cores to date. The Software in Silicon technology in the SPARC M7 processor delivers revolutionary hardware features designed to provide better performance and reliability for software. Application data integrity stops memory corruption caused by coding errors and some attempted security exploits. Acceleration units offload and process some database (SQL) instructions, speeding up in-memory database queries, and freeing the SPARC M7 cores for other tasks. Real-time decompression of compressed data will enable large in-memory databases to run on smaller machines, without compromising performance. The combination of the silicon and the software will enable the SPARC M7 to scan up to 170 billion database rows per second. Yes, that’s a lot of rows.
Get Converged
Delivering powerful, high-volume, high-speed, business solutions from a collection of generic, commodity infrastructure hardware is not simple or easy. A combination of engineered solutions—prebuilt integrations of hardware and software as well as software solutions purpose-built and optimized to make better use of hardware—is the key to simple and powerful business solutions.
Tom Haunert is editor of Oracle Magazine.
Source: iStockphoto
