問題文
A logistics group keeps a warehouse copy current with a subscriber. The architect is setting both the lag threshold that raises an alert and the response time expected of the on-call engineer, and knows the platform keeps the messages for a bounded period after which they are no longer available. How should the two figures be chosen?
選択肢
- By alerting on the age of the oldest unprocessed message once a day, which gives the on-call engineer one daily figure to judge the size of the backlog against.
- By working backwards from that period, so that the alert, the engineer's response and the recovery itself all complete inside it with margin to spare.
- By setting the lag threshold at the point where the business first notices stale data and letting the response time follow the group's standard for all systems, then extending the retention of the messages if a recovery is ever found to have run out of time.
- By alerting when the subscriber disconnects, which is the only event that matters here, since a subscriber that stays connected is by definition keeping up with the stream.