Tuesday, December 12, 2006
Thoughts on the 451 Group Conferences
Yesterday, I attended the Enterprise Computing Strategy Summit put on by the guys at the The 451 Group and have to admit these guys put on high quality information dense discussions, something that other analyst firms should aspire to...
Unlike the large analyst firms which spend a lot of time doing hand waving and repeat after me phrases such as IT needs to better align with the business, you need a good ROI and sage advice that IT should always find a business sponsor, the 451 Group conferences are none of this. If you enjoy bare metal, candid discussions not only with analysts who get it but with their customers who tend to be early adopters you need to find your way to their next event.
The primary theme of this conference was on the notion of grid. The participants all acknowledged the industry term but none of them actually use this within their own enterprises.
The analogy of IBM did grids in 1942 was pervasive and if an enterprise wants a grid strategy all they have to do is ask a mainframer. Seriously, enterprises need to proceed with caution, not because grid technology isn't mature, but folks within the enterprises thinking on grid technology aren't mature and the first time it fails, there will be many mainframe dinosaurs coming out the closest pontificating, see I told you so and the enterprise won't get another shot at doing something cool for the next ten years or at least till the next change in IT management.
Another key consideration in terms of driving grid performance is the usage of caching technology. I listened to an intriguing discussion with Cameron Purdy of Tangosol who talks about why in many enterprise applications CPU utilization is low. The popular theory is over-provisioning which has led to the proliferation of virtualization and consolidation initiatives. His answer though had more engineering thinking behind it and talked about the fact that a vast majority of CPUs remain idle because they are waiting for I/O which is always the slowest thing in terms of processing. Caching helps remove I/O bottlenecks which in turn will keep your CPUs busy which in turn may help enterprises actually consume all that wonderful stuff int the data center. Maybe the problem is that industry analysts need to start talking about how to correctly design an enterprise application.
Anyway, the next several blog entries will cover many of the insightful conversations I had while attending this event. The problem is that I actually have tons of notes that I want to share and may need to resort to censoring myself. Stay tuned...
| | View blog reactionsUnlike the large analyst firms which spend a lot of time doing hand waving and repeat after me phrases such as IT needs to better align with the business, you need a good ROI and sage advice that IT should always find a business sponsor, the 451 Group conferences are none of this. If you enjoy bare metal, candid discussions not only with analysts who get it but with their customers who tend to be early adopters you need to find your way to their next event.
The primary theme of this conference was on the notion of grid. The participants all acknowledged the industry term but none of them actually use this within their own enterprises.
The analogy of IBM did grids in 1942 was pervasive and if an enterprise wants a grid strategy all they have to do is ask a mainframer. Seriously, enterprises need to proceed with caution, not because grid technology isn't mature, but folks within the enterprises thinking on grid technology aren't mature and the first time it fails, there will be many mainframe dinosaurs coming out the closest pontificating, see I told you so and the enterprise won't get another shot at doing something cool for the next ten years or at least till the next change in IT management.
Another key consideration in terms of driving grid performance is the usage of caching technology. I listened to an intriguing discussion with Cameron Purdy of Tangosol who talks about why in many enterprise applications CPU utilization is low. The popular theory is over-provisioning which has led to the proliferation of virtualization and consolidation initiatives. His answer though had more engineering thinking behind it and talked about the fact that a vast majority of CPUs remain idle because they are waiting for I/O which is always the slowest thing in terms of processing. Caching helps remove I/O bottlenecks which in turn will keep your CPUs busy which in turn may help enterprises actually consume all that wonderful stuff int the data center. Maybe the problem is that industry analysts need to start talking about how to correctly design an enterprise application.
Anyway, the next several blog entries will cover many of the insightful conversations I had while attending this event. The problem is that I actually have tons of notes that I want to share and may need to resort to censoring myself. Stay tuned...
