Flexibility required in SLM/SLO handling of NULL values

Idea created by Chandresh_Patel Employee on Jan 29, 2018
    New
    Score5
    • kesv
    • Ryan Dunn
    • danst04
    • GuanHua1378
    • Chandresh_Patel

    Currently we sometimes receive the odd NULL value from certain probes during normal operation. While a series of NULL values would be a cause for concern, we can live with the odd one. Currently SLM's handling of null values is an all or nothing situation i.e. ignore all null values or don't ignore them at all (treating them as up or down). The problem is that we do not have control over when the null values could happen, so not possible to timeslice the slo's to look at particular times of the day. Obtaining a series of null values is something that we would not want to ignore.

    What we would like is to allow some flexibility in how we handle 1 null against a series of nulls. There is probably more that needs to be considered but some flexibility around the handling would be a good.