Content Copyright © 2018 Bloor. All Rights Reserved.
Also posted on: IT Infrastructure
Recently I spent two days with IBM at their Industry Analyst event in London. My brother joined IBM in the mid sixties working as a systems’ engineer on the early mainframe reservation system for BOAC (later merged with BEA to become British Airways). My formative technology years were with Burroughs, where every day was a battle against a huge array of IBM hardware, from 370 architecture mainframes through System 36/38 mini computers down to humble IBM PCs. Imagine my surprise then, when for two days there was not a single mention of mainstream IBM computers.
I say mainstream, because they did showcase IBM Q, their nascent Quantum computer; absolutely fascinating and ground-breaking with huge potential sometime in the future, but hardly something your average CIO will have at the top of their shopping list today.
IBM do have a very good cloud story covering public, hybrid and private cloud models. But it was curious that they made no mention of their IaaS capabilities to the point where another analyst had to ask if IBM actually still focused on selling pure IaaS contracts. The answer was yes, but no detail was provided on the service.
Much was made of their AI programs which, in my opinion are genuinely industry leading. Other colleagues of mine will no doubt analyse and comment on this in more detail. My area of focus is IT Infrastructure and I know that AI, machine learning and large-scale data science and data analytics requires quite specific High-Performance Computing (HPC) configurations to handle the processing and data handling requirements for complex algorithms and huge amounts of data. There was a small breakout session on IBM’s POWER AI… in other words AI solutions running on IBM Power hardware. But the focus was on some specific use cases, with no presentation of why Power might be better than other architectures for AI solutions.
After the event, I caught up with David Chancellor-Maddison, the presenter at the POWER AI breakout, and European Sales Leader for Cognitive and AI in IBM’s System Group. He gave me a brilliantly concise and powerful explanation of why POWER systems are so good at AI. Basically, it comes down to 2 key things, the way IBM interconnects CPUs to GPUs as well as GPUs to GPUs and, coherent memory that allows very large models to work across all CPUs and GPUs without being split.
Suffice it to say, in a short blog like this, the hardware design of POWER systems gives users the ability to run much larger models, far more quickly and with less potential errors than on other standard server systems. In my opinion, the combination of IBM’s AI and data analytics software, their deep industry understanding, their systems integration expertise AND the unique design and capabilities of POWER architecture is what makes POWER AI truly differentiated solution. I’m sure I will be having a look at this in more detail soon.
Coming back to Cloud, I was struck by how different IBM is compared to, say, OVH. OVH are all bits and bytes, feeds and speeds; a real focus on infrastructure design and performance. IBM are all about use cases, platforms, software and security. Given that a 2017 Cloud Spectator survey of the top 10 Cloud providers in Europe placed OVH top and IBM bottom in IaaS price performance this is perhaps not surprising. However, this is not the whole story. Cloud native businesses and applications may be grabbing the headlines and could well be an existential threat to many existing business models. However, according to IBM, 80% of the data in the world belongs to, or resides in, existing organisations. Turning data into insight will be critical in enabling businesses to become truly mutable and survive the challenges they are facing. IBM have the broad set of skills, knowledge and experience to help Enterprise CIOs re-architect IT to achieve that goal. Along the way, IBM needs to remember that the genuine strength of their underpinning IT infrastructure offerings will often provide key differentiators which will protect their own future. The M in IBM remains a critical element in their offer to their customers. They would do well not to forget it.