Wrong display of bottleneck: - throttling

[dweide]

Hello,

some of my backup jobs show the bottleneck to be throttling.
But this must be wrong since I have not set any throttling rules.
Moreover the statistic displays the target to be the real bottleneck, see below

09.01.2015 01:10:08 :: Busy: Source 24% > Proxy 9% > Network 10% > Target 99%
09.01.2015 01:10:08 :: Primary bottleneck: Throttling

Other jobs show realistic bottlenecks although the same repository and proxy is used.

Any idea about this?

Thanks!

[Vitaliy S.]

Did you have any rules configured before by any chance? By saying "realistic bottlenecks" do you mean that you have target reported for other jobs?

[dweide]

Did you have any rules configured before by any chance

- I did test storage latency control, but disabling it did not change the behaviour.
- The repositories have load control enabled (with both max concurrent tasks and data rate Limit set).
But again, some other jobs use the same repository without showing throttling as bottleneck.
- Apart from that I am not aware of any global network traffic rules ever beeing set.

By saying "realistic bottlenecks" do you mean that you have target reported for other jobs?

I simply mean that throttling seems unrealistic when the Job statistics display >99% Target and the next lines states throttling - or do I simply misinterpret the meaning of throttling?
Most other jows writing to the same repository Show source as the bottleneck, one has Network.
Within the recent history I could only find a target bottleneck with a job writing to a realy slow external USB drive (but I used to see more target bottlenecks in the past before improving the target raid).

Just to make sure: I should only ever get throttling as bottleneck when I have a rule defined globally, right?

[foggy]

- The repositories have load control enabled (with both max concurrent tasks and data rate Limit set).

As far as I can recall, ingestion rate settings are also considered as repository throttling (meaning that the speed of writing to repository is throttled) and can result in Throttling being reported as bottleneck.

But again, some other jobs use the same repository without showing throttling as bottleneck.

That means that some other stages perform slower during their execution (probably, they use some other slower production storage, some other traffic route, or the repository itself is not loaded much with other tasks at that time).

[Gostev]

Foggy is 100% spot on.