* Readers share their concerns about the 'godbox' approach to convergence
We received several reader responses to last week’s discussion about “enterprise on a blade” as a trend in convergence. Today we’ll address readers’ concerns with reliability, ASIC design demands, and enterprise infrastructure supplier approaches to the ‘godbox.’ For the purposes of discussion, we’ll assume that enterprise on a blade – in which discrete boxes such as routers, switches and PBXs will be replaced by services on a blade server – will happen, although as we pointed out last week, we also think the enterprise-on-a-blade will co-exist with hosted services, software-as-a-service and cloud computing.
First, on the issue of reliability. One reader commented: “What’s going to happen when the blade chassis dies? That’s a lot of eggs in one basket.” We agree. However, for mission critical applications IT managers are well advised to have a back-up to all critical failure points based on the cost-benefit of duplicate systems. For example, at Current Analysis the IT manager has uninterrupted power supply (UPS) available for critical systems so office workers and IT staff can gracefully save work and back up systems when power fails. A remote collocation back up is also available for every mission critical system, application and content so even if the main internal site goes down, work can continue from remote locations.
In a second point on reliability, another reader observed: “Hardware reliability hasn’t been as big an issue as incomplete failover. When one processor thinks the other is down (but it isn’t really), and both end up fighting for control is when the larger issues seem to arise.” Another valid point and a problem that can also be corrected with proper network and software testing in advance of deployments. We recall a catastrophic failure of the entire AT&T Signaling System 7 (SS7) in January 1990 that resulted from a “fight” between network controllers causing the entire AT&T voice network to shut down for nine hours! But AT&T (and others) learned from the problem, and network planners now can test to scale in advance of new hardware and software releases, mitigating – -but probably not entirely eliminating the potential for disaster.
We’ll return next time with more reader comments.
