Database Reference
In-Depth Information
Now in the section depicted in the following screenshot one can get into by drilling deeper
into the topology details. This can be achieved on the Storm UI by clicking on any of the
topology names. This section holds the details about the components of the topology in-
cluding the level of bolts, spouts, and details about them, as shown in the following
screenshot:
The preceding screenshot has details ranging from the number of executors or tasks alloc-
ated to each component, to the number of tuples emitted by the bolts or spouts and the
number of tuples transferred to the next component in the Directed Acyclic Graph
( DAG ).
Other notable details one should observe on the topology detail page are as follows:
Capacity of bolts in the last 10 minutes: This should be well below 1.
Execute latency is time in milliseconds: This determines how long it would take
to execute a tuple through this component. If this value is too high, then we would
probably want to break the execution into two or more bolts to utilize parallelism
and have better efficiency.
Executed : This stores the number of tuples executed successfully by this com-
ponent.
Process latency : This value displays the average total time taken to execute a
tuple by the component. This value should be analyzed with the execute latency.
These are practical cases that may happen:
Execute latency and Process latency are both low (that's the best pos-
sible case)
Execute latency is low but process latency is very high (that means actu-
al execution time is lower in comparison to the total execution time and
increasing parallelism might help achieve efficiency)
◦ Both Execute latency and Process latency are high (again, increasing
parallelism might help)
Search WWH ::




Custom Search