Alert! Alert!
14 February 2006 — DanielWith modern telecommunications, it’s now relatively easy to set up systems to call their operators for help when error conditions occur, no matter if those operators are in the office or elsewhere. In eVision’s case, if issues occur with our servers, the hardware guys know about it pretty much instantaneously.
Likewise with the applications, we’ve got a number of alerts set up that will advise if there are issues. For a system like the Message Exchange (MXC), error conditions might include things like:
- Network problems, including being unable to see specified remote servers
- Scheduled messages not arriving or being sent as expected — in one client’s case, it’s critical for their business that they receive their daily stock count. It’s dependent on an upstream system which occasionally has problems, so it’s something we need flagged ASAP to get it resolved.
- Unscheduled messages not arriving, when we generally expect them to turn up regularly — for instance, we expect many sales orders each hour during business hours on weekdays, but few on weekends. If half an hour goes by during business hours with none arriving, there’s probably a problem.
- Messages causing unexpected errors
- Warnings of the approaching lack of database or disk space — obviously this is a condition that needs to be avoided.
- System health checks, such as ensuring the database is running okay, and that the various system processes are all doing their thing.
(Note, alarms above are just examples. Alerts actually used in the MXC system may vary between versions.)
Obviously any system should be reliable as possible, and able to handle most error conditions automatically, and the MXC does so. But for some issues you need a human to go troubleshooting, and when your business is relying on it, letting those humans know quickly is the key to minimising the impact on your customers.
