logo
A "darkened" or "dark" data center has been designed to eliminate the need for human intervention so that system admins and managers can deal with mission critical development and process improvement projects vs. spending all their time fighting fires. These "lights out" data center initiatives have allowed managers to achieve the following:
- Lower TCO - By introducing tools that allow for remote management of IT platforms, IT managers can centralize systems administrators into primary locations. By having SAs work out of a few locations managing multiple remote sites, IT teams will reduce the amount of travel expenses and drive up the productivity of their staff. IT staff will be able to deliver "more with the same" and maximize their working potential. IT managers will also have better control of their asset management enabling them to reduce waste, reduce server sprawl and make smarter buying decisions.
- Faster Time to Server Recovery - By having the tools in place to remotely troubleshoot computer systems and applications, subject matter experts are empowered with a much broader toolset. No longer are experts dependent on hands and eyes at a keyboard (physically) in front of a server or a "crash" cart hundreds of miles away. With the introduction of better remote management tools, these experts can be introduced more frequently and at a higher volume during high severity fire fighting.
- Faster Time to Server Deployment - Many organizations blame the long deployment time of a new server on the coordination of activities between multiple parties. For example, to setup a new server, the facilities team has to provision power, space, and cooling; the network team has to run fiber and allocate ports on switches and routers; the SAN team has to do the same on their side; and the OS provisioning team has to build the server image. Every time a new server is added this process needs to be recreated over and over again. If these steps could be consolidated to only occur once and then all new server add ons could be done remotely, one can reduce time-to-deployment from a matter of weeks to hours.
- Reduction in Error and Downtime - Most IT downtime is the result of human error or change controls gone bad. By introducing a management software layer with unbreakable processes to control these changes, downtimes can be reduced. Imagine changing 100+ network cables from one switch to another during a change control. Instead of having to unplug and plug each one at a time, one can run a few scriptable software commands to get the job done. This is very achievable with the right tools in place.
- Remove physical components and replace with software. We talk about this a lot at Egenera. The less hands-based robotic functions required in your data center, the better. If these tasks can be replaced by software commands, those commands can be initiated from anywhere.
- Invest in compute hardware that is modularly field replaceable. Screwdrivers should not be needed to fix any component of the server. If a blade fails just pull the blade out and insert a new blade - no re-cabling or re-configuration needed if a new part needs to be swapped in.
- Introduce virtualization tools like hypervisors to reduce the number of physical servers you need, such as VMware, Xen, SWSoft etc.
- Introduce SAN remote management/provisioning tools to replace local disks needing tape back-up. Replace with offsite disk to disk mirrroring.
- Introduce management software that allows system administrators to better monitor and manage the software commands mentioned in step 1 as well as manage the virtualization complexity mentioned in step 3. Egenera PAN Manager software is a good option. PAN Manager and a few other tools (like Bladelogic and OpsWare) have the ability to up your remote system admin span to 80 blades. For example, Egenera has a very large military client that uses PAN Manager to manage 100+ BladeFrame systems at 100+ data centers from one NOC location by one NOC team.
