Remote server management is fraught with difficulties caused by the server management tools used and the idiosyncrasies of data center facilities. As data centers proliferate, data center managers rely on remote server management technologies to extend their reach -- controlling servers and systems to a very granular level without ever setting foot in the building. Even modern, lights-out server management tools like
Remote servers are generally accessed over dedicated management hardware or keyboard, video display and mouse (KVM) over IP. Both approaches pose potential limitations. Use them in tandem for detailed insights and convenient control.
Hardware-based management platforms such as Dell's integrated DRAC or the HP iLO Management Engine offer power management, media access and a remote server console accessible through Web browsers or a command-line interface. The downside of hardware-based management is its hardware requirements -- a separate card or module that is usually compatible only with specific server models. This is fine for homogeneous data centers, but complicates data center management using a heterogeneous mix of servers.
Traditional KVM switches relied on local wiring to switch user I/O devices to servers, which isn't suitable for remote server management. Data centers can use KVM-over-IP technology to encapsulate and exchange server control signals over the existing Ethernet network. A technician hundreds of miles from the data center can see and control specific servers as if they were standing in front of it. Any server model with KVM-over-IP support, which is most server baseboard management controllers today, can be controlled directly. Unfortunately, KVM-over-IP does not provide any specific management functions like brand-specific cards do, forgoing insights into physical parts of the server. For example, KVM-over-IP does not report on processor temperature or fan speed.
Lights-out server management tool limitations
Don't expect remote management software tools to show you every detail about every server, at least not clearly or intuitively. And even when the tool reveals a problem, that doesn't mean it can actually fix the problem.
The remote server management GUI might lack features and functionality that you expect, making the system cumbersome or hard to read. For example, remote management tools often forgo support for high-resolution video displays, since video traffic can tax the network. The display often suffers as a result.
Always perform management testing and proof-of-principal projects before rolling out remote management platforms to distant data centers. IT professionals must be thoroughly trained on the management platform and understand its limitations. IT staff will need on-site, contracted personnel available to work directly on problems and provide hands-on assistance.
Remote server power
You can only manage a remote server when it's actually on, so make sure power and backup power at the remote site are up to the task. A fleet of uninterruptable power supply (UPS) systems is standard for temporary ride-through power when local utility power is disrupted. This battery power only lasts a few minutes -- just enough time to execute controlled server shutdowns.
Some servers simply cannot shut down. For zero-downtime application availability, consider alternative power sources that can be switched on quickly while UPS systems run: diesel-powered generators, local cogeneration facilities such as solar or wind farms or an array of methane-powered fuel cells. You could opt for a completely redundant utility feed and provider, though this is impractical for most businesses.
Remote server management tools cannot help -- yet -- when the power fails due to a fault in the switching panel or open circuit breakers. Hire a remote technician to investigate and troubleshoot problems on-site when necessary.
You can only manage a remote server via network access, which requires reliable Internet connectivity through your local service provider, regional backbones and the remote service provider. Any disruption to network communication will prevent remote server management.
How to choose remote server management tools
How to manage remote servers well
Redundant Internet connections are common and practical for remote data centers. True redundancy occurs when separate carriers use separate lines. Use a carrier with T1 or T3 connectivity, for example, and rely on a business-grade cable provider for backup connectivity. Any redundant Internet providers must include a redundant cable; many organizations contract with redundant Internet providers that use the same physical wires.
You could deploy a dial-up Internet connection for emergencies, but remote server management over dial-up is a challenge for even the most experienced administrators.
Consider employing a remote technician to maintain the internal network at the remote site. A technician can locate a failed router that is causing connectivity problems, internal network adapter or switch port problems. Remote management tools cannot fix these problems.
This was first published in September 2013