Control Tower, RLX’s application for managing its server blades and blade chassis, has long been one of the best server administration tools available. With the newest release, 6G, it gains new capabilities for detecting problems with an RLX server system and for using policies to automatically remediate those problems.
Although Control Tower 6G is a worthy upgrade, the software remains narrower in scope than data centre management tools such as IBM Tivoli, Hewlett-Packard OpenView, or Computer Associates Unicenter. On the other hand, it’s only a fraction of the price.
I tested Control Tower 6G using an RLX server system provided by the company, consisting of a 600ex chassis and five ServerBlade 2800i servers. Two were set up with Windows, and three with Red Hat Linux.
The Control Tower 6G software itself resides on a separate 1U server, branded by RLX but manufactured by SuperMicro. I looked at a similar hardware combo last May; this review focuses only on Control Tower 6G.
There’s much to admire about Control Tower 6G. Because it’s designed to work only with a narrow set of hardware devices, the management can go deep, providing on-screen reports with rich graphics, intuitive menus, and clean navigation structures. The browser-based software uses digitised photos and animation to display real-time status reports, and it can e-mail to alert administrators of malfunctions.
Where the software continues to fall short is in its capability of managing a broader array of hardware and software, even those from RLX. The release I looked at in May was designed to manage the company’s blades as well as its SAN hardware. Although the SAN switches and disk arrays carried the RLX logo, they were a hodgepodge of devices assembled from other manufacturers, and Control Tower XT couldn’t manage them - so much for an integrated solution. Control Tower 6G hasn’t evolved further in that direction; it remains geared specifically for managing the servers.
But as a dedicated server admin tool, the software remains solid, particularly for automated provisioning. Control Tower provides an excellent point-and-click way to capture a desired configuration for a server blade and then to deploy it based on a particular event or schedule. This can be used either to swiftly bring up a new server after one fails or to change a server’s role. For example, a server might be deployed as part of a Windows-based employee portal cluster during business hours but then be wiped clean and added to a scientific-computing Linux cluster at night and during weekends.
You can set policies that allow you to allocate resources according to workload and other parameters. So for instance, if one application needs extra horsepower, more servers can be allocated from other jobs, or from a spare pool if you have such a thing, and applied to the needy task.
Conversely, if workloads are low, you can shut down servers that are surplus to requirements. If, for instance, you have an application using 10 servers that are 20 per cent utilised, in an ideal world you could instead allocate just two servers to the application, and shut down the other eight to save power. Of course in practice, you'd be more likely to shut down just five or six...
Three improvements in 6G stand out: dynamic polling, workload monitoring, and “task chain” response automation. Those features are implemented as extra-price add-on modules.
Previous versions of Control Tower were passive: server blades or other hardware components notified it when something went wrong using SNMP. The weakness there is that if the monitored device wasn’t capable of sending an alert -- perhaps it died completely or the SNMP service failed -- Control Tower wouldn’t know what happened.
Control Tower 6G now can dynamically poll servers and other devices to determine their health, temperatures, CPU utilisation, network traffic, and other key metrics. This means that failures can be detected more quickly and more accurately. Also, it’s much easier to modify how Control Tower 6G interprets a specific condition via polling than it is to change the thresholds where individual devices are programmed to trigger SNMP alerts.
Control Tower now allows administrators to customise alert levels for monitoring specific groups of servers; one set of monitoring parameters might be applied to an e-commerce Web server cluster, while another is applied to a pair of DNS servers.
Using polling techniques, RLX’s Workload Inspector can now go down to the OS level to monitor the status of Windows or Linux services, as well as the system’s TCP/IP stack. This allows alerts or automated actions when a server’s I/O is maxed out or a service fails. But Control Tower still doesn’t provide the probing depth offered by enterprise platforms such as Tivoli or Unicenter, which can probe specific applications.
Another step forward is the 6G’s capability of defining policies with multi-step responses to events. These task chains can be customised to monitor specific server groups and can be triggered based on specific alert conditions, such as a state change of a monitored device or parameter. The company provides several pre-programmed task chains, including those that capture detailed diagnostics, restore a server to a known state, reboot it, and then place it in operation.
Although the task chains are a definite improvement in Control Tower’s automation, they’re still a work in progress. For example, there’s no way to build in “if-then-else” conditions -- the chain consists of a linear progression of tasks, nothing more. There’s no error handling within the chain itself; if a step fails, execution of the task chain stops.
Finally, the scope of a task chain is restricted to the component that triggered its execution. For example, if Control Tower has found a high temperature situation on a server, a task chain can’t bring a hot spare server online and then bring down the problem server for maintenance when the spare has been confirmed to be working properly. That level of programmability, RLX told me, will have to come later.
As the software takes its first steps toward becoming a more general data centre tool, those capabilities are still beyond its reach. Despite those limitations, Control Tower 6G excels at its primary charge: the daily administration, monitoring, and provisioning of server blades and blade hardware. For that task, it’s still the best tool around.
Option prices: Control Tower 6G and licence for as many as 250 servers; Provisioning Manager, $199 (£108) per server; Workload Inspector, $299 (£162) per server; Automation Policy Manager, $399 (£217) per server
The best blade-server management tool gets better, with active probes, the capability of monitoring a server’s TCP/IP stack, and multi-step responses to scheduled events or error conditions. Although Control Tower 6G continues to automate server administration, it remains limited to working with the company’s own server, and its programmability doesn’t approach that of a general data centre management suite.