Light-Speed Performance

Continued

In addition, the Oracle Exadata Storage Servers can be combined into a grid to serve one application or many. “Each one of these servers can deliver a gigabit per second of data bandwidth,” says Oracle’s Mendelsohn. “So when you create a grid of 10 servers, you can achieve 10 gigabytes per second of data bandwidth to the database servers, or 100 if you have a hundred servers. It’s a completely scalable architecture that can keep pace with data growth. As customers create bigger and bigger data warehouses, they can be assured of bigger and bigger data bandwidth.”

They can also be assured that time to query isn’t going to be cumbersome. With large data warehouses, time to query can become longer and longer as tables become bigger and bigger. Not so with Oracle Exadata. “We can scale out so that as a data warehouse doubles in size, we double the number of storage servers, and that doubles the amount of data bandwidth and keeps the query time constant,” Mendelsohn says. “Because we have Oracle’s software running on the storage servers, we can do something you can’t do today with conventional storage.”

For example, when a customer sends a SQL query to the database server requesting rows for a specific customer, the Oracle Exadata Storage Server will send back only the rows that meet that request and thereby greatly reduce the network traffic. “It’s a pretty radical shift from conventional storage servers,” says Mendelsohn. “Oracle Exadata Storage Servers can form massively parallel grids to do massively parallel query processing. It’s really a revolutionary change.”

It may also be a big change for organizations used to spending time and energy to build out their own database and storage solutions. “If you buy storage and install your software on it, even if it’s a configuration that’s supported by Oracle and the storage vendor, there are always situations where it’s a little ambiguous as to whether the problem you’re having is due to software or storage,” says IDC’s Olofson. “With the Oracle Exadata storage system, you obviously won’t have that kind of issue.”

Data Warehouse in a Box

Oracle’s hardware vision doesn’t stop at the Oracle Exadata Storage Server. Working with HP, Oracle has also developed the HP Oracle Database Machine, a complete package of software, servers, and storage designed for large, multiterabyte data warehouses. The HP Oracle Database Machine uses Oracle Exadata Storage Servers as one of its key components.

“The primary benefit of the HP Oracle Database Machine is that it’s got the Oracle Exadata Storage Server built into it, so it’s a database server and storage server all in one cabinet,” says Olofson. The HP Oracle Database Machine is also a hardware building-block component—it contains 14 of the Oracle Exadata Storage Servers, 8 database servers, and an InfiniBand switch for connecting the database servers to the storage servers. The result is an out-of-the box data warehouse experience that can scale as business needs require.

“If someone wants to build a new data warehouse, they can just order the HP Oracle Database Machine and it’ll come pre-set-up and installed and configured to run a data warehouse really efficiently right out of the box,” says Mendelsohn. “You don’t have to figure out how to design it, what types of storage and servers to select, how to configure it, or anything else. Instead, you get a really nice, simple, out-of-the-box experience.”

“Like any appliance offering, it simplifies the configuration and setup as well as the support relationship,” adds Olofson. “The result is that a system administrator won’t have to research the best way to configure the combination of server and storage for an Oracle data warehouse. It’s already done, and it’s fully supported, so you don’t have to worry about the finger-pointing between vendors that might occur if you build your own solution.”

Optimized Data Analysis

Really testing out the Oracle Exadata solution requires a data warehouse environment with millions or billions of records—a data warehouse that’s really large and requires a lot of analysis. That’s why LGR Telecommunications was invited to beta-test Oracle Exadata.

Using Oracle databases and data warehousing technologies, LGR has developed a powerful call detail record (CDR) solution used by companies such as AT&T, Telstra, Vodafone, and MTN for extraction, transformation, loading, analysis, and reporting of all the data coming from telecommunications networks. The mountains of telephone and networking equipment owned by a phone company can typically generate billions of records a day. Mining that amount of data for valuable business insights is no small matter. “Given the highly competitive business environment, every subscriber is gold. The best way to determine how the subscribers are using the network is from the information in the CDR,” notes Salmon.