Infrastructure Metrics Streaming
Created: | Updated:
z/IRIS transformed RMF SMF records into metrics and streams them to a target metrics database which could be a time-series database or an enterprise observability.
Supported RMF SMF records
SMF Type | Record description |
|---|---|
70 | Processor Activity |
Measurements
Tags,labels and fields provide indexing, filtering and grouping of metrics.
CPU load
zos.rmf.cpu.mvs.utilization
Utilization is computed as a percentage of online time:
percentage_lpar_busy_time = (online_time - (wait_time + parked_time) /
(online_time - parked_time) x 100
Name | Metric Type | IBM Source |
|---|---|---|
- | timestamp | SMF70TME + SMF70DTE |
sysplex_name | tag | SMF70XNM |
host.name | tag | SMF70SNM |
cpu.type | tag | SMF70TYP |
gauge | field | The percentage of the online time that the logical processor was busy. |
zos.rmf.cpu.mvs.lpar.utilization
LPAR utilization is computed as a percentage of dispatched logical core time.
When LOADxx PROCVIEW CORE is in effect:
percentage_lpar_busy_time = (multithreading_core_time) / online_time x 100
Otherwise:
percentage_lpar_busy_time = (online_time - wait_time) / online_time x 100
Name | Metric Type | IBM Source |
|---|---|---|
- | timestamp | SMF70TME + SMF70DTE |
sysplex_name | tag | SMF70XNM |
host.name | tag | SMF70SNM |
cpu.type | tag | SMF70TYP |
gauge | field | The percentage of the online time that the logical core was dispatched by LPAR. |
In-Ready Work Unit Queue
$work represents the sum of the IBM SMF source fields SMF70U00,…,SMF70U15
zos.rmf.in_ready_work_queue.contention
Name | Metric Type | IBM Source |
|---|---|---|
- | timestamp | SMF70TME + SMF70DTE |
sysplex_name | tag | SMF70XNM |
host.name | tag | SMF70SNM |
gauge | field |
|
zos.rmf.in_ready_work_queue.units
Per unique value in cpu.waiting_work_unit
Name | Metric Type | IBM Source |
|---|---|---|
- | timestamp | SMF70TME + SMF70DTE |
sysplex_name | tag | SMF70XNM |
host.name | tag | SMF70SNM |
cpu.waiting_work_unit | tag | SMF70U00,…,SMF70U15 |
gauge | field | 0 → 100 * cpu.waiting_work_unit/ $work ⋮ 40 → 100 * cpu.waiting_work_unit/ $work ⋮ 150 plus → 100 * cpu.waiting_work_unit / $work |