IBM has re-engineered its PureSystems line of x86 servers to improve memory performance.
For memory-hungry applications, the new X6 server family has been designed to deliver three times the scalable memory of servers, IBM claimed.
The new x86-based systems support cloud and analytics.
Erich Baier, global vice-president for development of the technology, said: “With X6, we have focused on three areas: speed, agile and resiliency. The X6 supports more cores. We also worked explicitly with Flash memory providers to minimise memory write latency by integrating the system with flash memory channels, which enables data to be committed to memory faster.”
IBM said these improvements mean the server now supports up to 12.8TB of ultrafast flash storage close to the processor, increasing application performance by providing the lowest system write latency available – essential for analytics applications.
More datacentre articles
IBM said it has also addressed long-term cost, by making the X6 easier to upgrade. Baier said: “Traditionally, with rack-mounted servers you would buy the whole infrastructure upfront. When you made a significant change you then needed to replace the whole server.”
He said the X6 was based on a modular, scalable design capable of supporting multiple generations of Intel server CPUs.
IBM estimated that this would reduce acquisition costs by up to 28%.
Along with the server, IBM announced new solutions for its X6 architecture for analytics, database and cloud deployment, including IBM System x Solution for DB2 with BLU Acceleration on X6 for accelerating analytics.
Baier said the fast memory bandwidth makes the X6 suitable for in-memory and columnar databases. The configuration of the server also enables up to 12TB of data to be located close to the processor, without going to a SAN. With 120 cores and four threads per core, Baier said the X6 was also able to provide the massive number of cores and memory mean desktop virtualisation.