The Internet never shuts down, and your directory service shouldn't have to either. We've designed our UnboundID server software to be as robust and admin-friendly as possible so that most administration can be done with the server online. But even if you have to take an instance offline (e.g., to update the server software, or in the course of managing the underlying system), it doesn't need to have any effect on the overall availability of the service. The UnboundID Directory Proxy Server proactively monitors the health of all backend servers and can automatically de-prioritize or stop using a server if a problem is detected, immediately notifying administrators of the problem (using extensible mechanisms like email, SMS, SNMP traps, etc.), and automatically re-attempting any operations that failed as a result of the problem on a different system so that clients never notice anything wrong. Some of this robustness can also be achieved by taking advantage of advanced availability features in the UnboundID LDAP SDK for Java, so you can even insulate your applications from outages involving the Directory Proxy Server itself.
Even without the Directory Proxy Server, the UnboundID Directory Server provides a number of advanced monitoring capabilities that you can use to ensure the service stays up and running in tip-top shape. The server automatically tracks and exposes a broad variety of metrics, from information about the underlying system (CPU utilization, disk space availability, memory usage) to the health of the instance (worker thread busyness, database cache utilization, operation response times) and the overall topology (replication backlog, replication latency). You can define thresholds for these items so that the server will automatically notify administrators if any of the values go outside acceptable bounds, and in some cases the server can take further action to prevent a mounting problem (e.g., from running out of disk space or falling too far behind in replication) from adversely impacting clients. Of course, if you do have the Directory Proxy Server in place, then it will take advantage of this monitoring information to decide how best to route requests and ensure the best overall client experience.
Much of this monitor information can be exposed via SNMP for integration with standard enterprise management systems, and we also provide an API for accessing and interpreting the monitor data in real time so that you can provide even greater integration with whatever monitoring system you might happen to use. On top of that, the optional UnboundID Metrics Engine software offers graphical real-time and historical views of a broad range of performance and availability metrics, for a single server or the entire topology.
The UnboundID suite of software is designed to help recognize and alert about issues before they become critical, but it's also fully capable of reacting to completely unexpected problems and insulating clients from their effects while ensuring that administrators are aware of them so that they can be resolved quickly. We also provide a wide range of diagnostic tools to help you discover the source of a problem or collect analysis to pass on to UnboundID support personnel if required. And in the absolute worst-case scenario, our simple, scriptable setup and high-speed replica initialization process can allow you to bring up a whole new instance as quickly as possible. In Part 4 of this series, I’ll explain how EASY it is to migrate to the UnboundID directory service.
Don't forget to join us for our new LIVE WEBINAR: DSEE is Dead--Now What? on Nov. 5th at 1pm CT/2pm ET.