Tuesday, January 18, 2011

Independent management tools

Take a long hard look at an independent infrastructure management solution and make the break from platform-specific tools. Products from companies like SolarWinds and APTARE can help monitor and manage the independent components in the environment, from disk arrays to networks to backup infrastructure, from a single interface. Several factors point to the value in this approach:

1. an increasing interdependence of heterogeneous components, especially within virtual server environments
   

2. more requirements to repurpose gear, given flat budgets, no more ‘sweeping the floor’ with new systems. Consolidation also means making different pieces work together, not just combining systems
   

3. more interest in charge back as a way for IT to spread out costs and how it must include resource usage beyond just disk capacity
   

4. the popularity of appliances and ‘commodity hardware’ which means more vendors to deal with
   

Single digit disk utilization?

An SRM vendor said something in a briefing several months ago that still blows me away. They said storage utilization is hovering around the single digit mark in most large enterprises, based on their experiences. Even if it’s twice that, disk capacity utilization of 20% is still abysmal. The reason is that everyone’s trying to improve application performance (reduce latency) but most organizations don’t really know where the bottlenecks in their infrastructures are. Since they rely on their disk vendors to provide ‘guidance’ in these situations, the solution ends up being to add more disk (go figure). A better answer may be a platform independent monitoring solution from Virtual Instruments that provides real time data from network traffic (fibre channel) and can identify the real sources of these bottlenecks - which usually don’t turn out to be the disk arrays.

Power management

Most companies plan for data center expansion and new construction using the same information that the facilities department uses, the power ratings on the back of each server. These numbers are typically the maximum power the unit could draw. In reality, server power consumption varies widely as CPU load increases, and more importantly, most servers sit idle for the majority of the time when power consumption is very low.  The result is an over-provisioning of power and distribution infrastructure in most data centers and a lack of optimization of existing resources. There are tools available like those from Viridity that bring a software approach to power management, similar to the way other IT resources are managed. They can predict power usage by server model at various CPU loading levels from idle to 100% and give IT the information it needs to intelligently build out IT infrastructure and effectively manage resources.

Virtualized environments

Virtualization is somewhat of a perfect storm for infrastructure problems. It’s applicable to most environments, it abstracts physical resources, it’s complex and it’s easy to implement. In order to optimize the resources of storage, networks, servers, applications, etc, IT needs independent data in real time. Again, platform-specific tools are helpful but not always the best answer. Tools like Xangati that capture real time data directly from each resource and present it in a ‘cross-silo’ perspective, can be invaluable at untangling a resource issue in the virtual environment. These products can even record real time data feeds and replay them like a TiVo for faster problem resolution.

Eric Slack, Senior Analyst