Oracle Launches Fifth Generation Exadata Database Machine
December 13th, 2013 By: John Rath
Oracle (ORCL) launched its fifth generation database machine, with increases in performance and capacity, new software capabilities to optimize OLTP, and a new focus on Database as a Service (DBaaS) and Data Warehousing. Oracle Exadata Database Machine X4 has enhanced hardware, and is fully compatible and interoperable with previously released Oracle Exadata Database Machines, enabling customers with existing machines to easily expand with the newest system.
The new X4 offeres improves end-to-end performance, with up to 2.66 million read IOPs per full rack, and up to 1.96 million write IOPs to flash. It has a 50 percent increase in database compute performance on X4-2 systems using two 12-core Intel Xeon Processors E5-2697 v2. It can also push close to a 100 percent increase in InfiniBand network throughput and the addition of Network Resource Management to ensure ultra-low response times for latency critical database operations.
With up to 44 TB per full rack the Exadata Database Machine X4 has a 100 percent increase in physical PCI Flash capacity, and up to another 100 percent increase in logical flash cache capacity to 88 TB per full rack. Its Exadata Flash Cache Compression transparently compresses database data into flash using hardware acceleration to compress and decompress data with zero performance overhead at millions of I/Os per second. Using memory expansion kits, the X4-2 can have up to 4 TB of memory per full rack, and over 200 TB of disk storage capacity. It features a 33 percent increase in high capacity disk storage capacity to 672 TB per full rack.
“Enterprise data center managers are often burdened with the complexity and cost associated with managing data in siloed environments,” said Carl Olofson, research vice president at IDC. “Oracle Exadata Database Machine X4 aims to provide database administrators a single, centralized and automated environment to manage databases. The new system’s technology upgrade should deliver significant performance improvements that address the demand to implement Database as a Service, offering the scalability, control and availability to support a full range of database workloads.”
With the increase in flash capacity, the X4 is sufficient to hold the vast majority of OLTP databases entirely in flash memory. Random I/O rates that are critical for OLTP applications have been improved close to 100 percent to 2.66 million 8K database reads and 1.96 million writes, even with full flash compression enabled. Performance of Data Warehousing workloads is accelerated by new Flash Caching algorithms that focus on table and partition scan workloads that are common in Data Warehouses. Tables that are larger than flash are now automatically partially cached in flash and read concurrently from both flash and disk to speed throughput.