Follow Datanami:
September 28, 2021

Exadata X9M Scans 1 TB Per Second, Oracle Says

Oracle rolled out its new X9M version of its Exadata Database Machine today, and as you might expect, it’s gotten faster. In fact, the company says it’s 87% faster than the previous generation for analytics workloads, and can now process 1 TB of data per second for SQL analytic scans.

“We’ve crossed a boundary now,” Juan Loaiza, Oracle’s executive vice president of mission critical database technology said today during a presentation. “It’s 1 TB per second worth of analytics scan, which is a lot of data. One terabyte is a gigantic amount of data to process in one second.”

The ability to scan 1 TB of data per second with a SQL engine will undoubtedly benefit a number of Exadata customers, which includes 87% of the Global 100 list of the world’s largest companies, Loaiza says.

But the Exadata isn’t a one-trick pony. In addition to analytics (or OLAP) workloads, the machines can also run OLTP (online transactional processing) workloads, such as those for managing a bank, running an e-commerce operation, or managing orders in a factory. Plenty of Oracle software offerings, from Fusion apps and E-Business Suite to Siebel and Netsuite, run on Exadata, Loaiza said.

On the OLTP front, the combination of technologies in the core system, such as remote direct memory access (RDMA), persistent memory (P-MEM), and RDMA over Converged Ethernet (RoCE) is enabling Oracle to cram an ever greater number of transactions into a given second.

“In our latest Exadata generation, we’re making OLTP for example 70% faster than our previous X8M generation. That’s huge,” Loaiza said. “We’re now at 27 million I/Os per second–real database I/Os per second.”

It’s doing all that, while maintaining latencies for OLTP transactions at under 20 microsecond. “Which is extraordinary,” Loaiza added.

Oracle isn’t raising prices on the X9M, which means that customers will get more bang for their Exadata buck. Specifically, Oracle says OLTP costs are 42% lower, while analytics workloads are up to 47% lower than previous generation machines.

“In our new generation, we’ve added 28% more storage capacity, 33% more compute capacity, 30% higher memory capacity,” Loaiza said. “So our Exadata has become significantly more cost effective with the X9M generation, so it’s a big benefit for our customers.”

Thousands of companies around the world are running Exadata, and most of them are on-prem, according to Oracle’s figures. For the past several years, the company has offered customer the option to run Exadata in the public cloud. They can also run Exadata in a cloud-like manner, while keeping the hardware on site for latency purposes.

Oracle also touted speedups for the Exadata Cloud@Customer X9M mode. The company says scan rates are up to 80% faster, and I/Os per second is 87% higher (that number should be familiar now).

Related Items:

Oracle Touts an Exadata Backlog

Persistent Memory Can Change the Way Enterprises Navigate Advanced Analytics

Oracle Rejiggers Exadata for Emerging In-Memory Workloads

 

Datanami