IBM STG (the hardware group) will tell you infrastructure matters for big data solutions. This is correct. The capabilities of servers and storage, networks, and clouds will all definitely have a significant impact on a number of characteristics of an analytics environment, including but not limited to performance, scalability, reliability, and cost.
So when customers and prospects are looking to build an enterprise-grade solution, they need to think beyond the development of functionality of applications and supporting data platforms (not necessarily just traditional structured databases) and engage all the various lines of IT operations and infrastructure to select the optimal hardware.
The challenge for IBM and its clientele is that the plethora of options can cause so much confusion. Off the top of my head, IBM is positioning the following products as big data solutions (deep breath here!) :
InfoSphere Streams (real-time analytics)
InfoSphere BigInsights (Hadoop)
InfoSphere Information Server with Data Explorer, Global Name Management, Anonymous Resolution, & Identity Insight
Master Data Management (MDM)
Watson (cognitive) with Engagement Advisor, Explorer, & Discovery Advisor
PureData with Netezza (data warehouse)
DB2 with Blu Acceleration (in-memory database)
Business Analytics (BI)
SPSS (predictive) with Statistics, Analytic Server, Analytic Catalyst, Analytic Modeler
Informix (IoT sensor data, real-time analytics, with NoSQL capabilities)
Cognos Performance Management
POWER systems with Power8
System x (going to Lenovo any day now)
System z (mainframe)
While this is an incredibly diverse list, likely most customers will only be familiar with a couple of these offerings. So there is a great opportunity to up-sell and cross-sell for IBM field teams, but also a conversation that will undoubtably have to revert to first principles, ("What are you actually trying to accomplish anyway?") before positioning any single or multi-product solution for big data and analytics. IBM does have both reference architectures and Global Services (systems integration) to help find the right combination for any particular workload or goal.
That said, IBM is also seeing some powerful opportunities to combine these technologies to achieve further impressive results. For one example, DB2 with Blu Acceleration can do in-memory columnar database processing leveraging the POWER8 processors and actionable compression for much improved performance. Another example is using Elastic Storage to dynamically tier the large volumes of data across different pools, including Flash, for the right mix of resources to balance speeds and costs. Or consider Hadoop on the mainframe, not a typical deployment model, but an interesting one. Many more such possibilities are being explored.
If a customer wants to play "big data blackout bingo," they'll find IBM has covered most of the squares on the board. But it may take some effort to design the optimal mix of hardware and software to meet the business goals and IT operational requirements. The good news is IBM still has a lot of potential to build tighter connections between all these products, not just sizing appropriately, but doing some really clever integrations for more value. I look forward to seeing them execute on a tighter story.