Old foreign teacher, you can maximize the normal running time of the server

Issue alert to tools

Beddoe uses the Uptime software company's normal running time software, which he thinks is important because they can be used when the server condition exceeds a threshold value, for example, if the memory is overloaded or the CPU usage is too high, you can issue a warning.

Although most tools have built-in alert functions, Beddoe emphasizes that you should still find a product that can configure alert trigger conditions, for example, a product that can send an email or text message once the preset threshold value is exceeded. "You need meaningful warning information so that you can take necessary measures to correct this situation ."

Walter Beddoe, vice president of IT and logistics at Six Telekurs, said, "We have never experienced major downtime events that affect the interests of our customers over the past 17 years ."

Jerry Gregg, Operation Manager of Carfax, an automotive service company, said it is important to understand that the normal running time calculated by many performance measurement tools is only an approximate value. "Such a value can only be used for reference at best ."

Gregg observed that the values obtained by some preliminary normal running time measurement tools may actually cause misunderstanding because these tools cannot be correctly classified into the following two situations: A one-hour server downtime occurs during sleep on Sunday morning and a 10-minute system failure when key business processes are running on Thursday afternoon. This is why you should purchase measurement tools that provide full-time, event-based analysis capabilities.

To make normal running time analysis more meaningful, Gregg decided to choose a measurement tool that shows the impact of server crashes on key business services. Gregg uses BMC's ProactiveNet performance management software, which can directly associate server downtime with sales transactions and other types of business-oriented data. "We can use US dollars instead of just time to quantify server downtime events ."

The information generated by the software helps him determine whether a down event threatens the bottom line of the enterprise's profit and loss, defends against the budget for purchasing new servers, better network devices, or other reliability enhancement technologies and services. "Without such information, you can only make cost-benefit decisions without knowing the operating costs," Gregg said.

Do not let hackers "steal" the normal running time

Security also plays an important role in ensuring the normal running time of the server. If the server suffers a malicious software attack, or the network path is insecure, server downtime is not surprising. "We need to start from physical security-that is, the building of a data center-to ensure physical security first," Beddoe said.

Second, it is important to establish access rules for servers and enforce them. At the same time, it is important to enforce security programs, anti-virus programs, and firewalls to train law-abiding administrators. "All these elements play the same important role in server security and improving normal operation time," Beddoe said ."

John Luludis, who supervises server operation for IT Consulting and customer software developer Superior technology solutions, said IT is important to go beyond basic security practices to maximize the normal running time of servers. Luludis strongly advocates regular independent security audits. "The network I supervise must carry out penetration tests on a regular basis. The reason for doing so is to make my network as secure as possible, and it is best to be safe from the outside ."

Protect your data

Although Howard of Princeton Radiology strongly believes in regular server maintenance, he also pointed out that managers and employees cannot avoid a certain number of failures. To prevent any data loss caused by Server failure, Howard recommends developing a data protection plan and integrating it into an enterprise's comprehensive business continuity strategy.

Princeton uses an out-of-site storage solution from Compellent technology to replicate all the data that has been stored. "Even if there is a disaster recovery data center, we actually have to run some servers out of the main facilities, so we need two-way backup of data ."

Raoul Gabiam, IT operations and engineering design manager of the University of Washington, believes that lifecycle management is an internal part of the normal server running time planning.

Gabiam of the University of Washington is relying on the Load Balancing Technology built into the network infrastructure to prevent sudden server downtime. "If a server crashes or an application does not respond, the network traffic will be redirected to other servers, and the same server can handle this workload ."

Unlike Howard of Princeton, Gabiam is optimistic about clusters and uses the Novell cluster service to provide additional redundancy layers. If a node in the cluster fails or requires downtime maintenance, the cluster application or a service component running on the node can be seamlessly migrated to another node in the cluster.

This migration process can be configured as manual failover or automatic failover. "In general, when the hardware or software becomes invalid, the application should be automatically invalidated for backup to the next alternative node," Gabiam said, however, administrators can manually migrate applications to another node when a specific node requires maintenance tasks.