On Monday, October 12, IBM announced their first Power8-based enterprise class systems and yet still did not deliver the full realization of its 12-core Power8 processor, previewed by IBM in August 2013. Potential customers should also really consider IBM's lack of deep-level optimization, as compared to Oracle and SPARC.
By opening Power to partner co-development, IBM has lost much of its ability to retain control of deep-level optimizations. Consider that IBM agreed to pay Globalfoundries Inc. $1.5 billion to take an unprofitable chip-manufacturing unit off its hands. It now makes much more sense as to why IBM is investing in partner add-on capabilities through CAPI (Coherent Accelerator Processor Interface), such as chips from NVIDIA, instead of investing in its own software in silicon enhancements like Oracle. And while Oracle was busy acquiring Sun, Pillar, Xsigo and more, IBM was divesting its networking, hard disk, x86 server business and most recently their chip manufacturing unit.
From a product feature standpoint, consider that CAPI lacks the true software in silicon (as Oracle is doing with SPARC/Solaris) approach and
co-engineered Hardware + Software solutions (such as Oracle ECI Solution for SPARC),
from a workload optimization and total business solution standpoint.
Customers increasingly state that they want and need optimized solutions
for their data centers to reduce complexity and simplify data center
management, as well as resource utilization. They need this to drive
their time to value; Oracle delivers. With CAPI, who is going to test,
support, and optimize for key customer workload environments – IBM? NVIDIA?
Why Invest with Oracle?
What do you think? Are these game changing innovations from IBM?