Hi @KAI DING

Yes,  I am aware of the separate pages for Locking....   ​That's fine....  
but having more and "call-out attention" information in the main summary page is way better,  because it's there where everything starts.  You detect something that is not expected..  then you can drill down..   But that's exectly my point about locking information in the summary page..   the level of information got reduced from 3 metrics  (lockwaits,  deadlocks,  Locks-Escalations) to just one  (lock waits). 

Here is a DSM main dashboard sample 

As you can see,   in the top-right corner, we have a button that allows us to select the metrics we "want" to see. 
Each metric selected is plotted in a graphic.   so,  here,  the result is that we have 3 isolated graphics for locking information. 
I don't mind having these 3 metrics in a single graphic,  with 3 lines..   that's good as well, and has the advantage of occupying less space in the dashboard. 

My original post was an attempt to enhance the information even further..  Besides having the 3 metrics,,  also have a better understanding of distribution on lock-waits..  are they important or not ?  

Regards

It tells me something, but hard to say if I had lock contention or not.  
It tell me that on average I have 20 lock-waits/min.. and a peak of 130 lock-waits.   
Ok.  But what kind of locks are they? Does the peak, really represents a lock contention , yes or no? Hard to say from it. 

Those activities that waited for locks,  they waited too much , compared with the activity total time ? Or they were just regular fast and despicable lock waits that last just few ms that happens normally in a OLTP system ? 

On DSM,  you had ALARMs for lock-waits,  where you can set threshold (warning and critical level) for activities that waited on locks for more than a specified % threshold.   

Now,  wouldn't be nice if we can have all of them together, in a single graphic ? 

For all lock-waits detected in a monitor interval..   
 - How much of them waited less than 10% of the activity time?   (low impact) 
 - How much waited between 10 and 40% ? (medium impact) 
 - and how many were highly impacted,  above 40% ? 

What about a graph that could show us all of that..  ? Something like this..  using lines for the 3 basic metrics. and stacked bars for the lock wait distribution 

or,  a more similar with the time-Spent graphic,  with stacked Areas 

Personally, I like more this last one. 

with this kind of information,  we have much more information now..  We can see for example, that we had two peaks of lock-waits..  but just one of them really represented a real contention.  The second one,,  was just workload increase related, as all of them just waited no more than 10%. 

Off course,  the range of 10% and 40% here are just examples.. 

The determination of such values to set the ranges (low,  medium and high) should be detected automatically by statistical ABC or PBA analysis,  and/or allow the user to set these thresholds. 

Regards 

Hi @KAI DING

Yes,  I am aware of the separate pages for Locking....   ​That's fine....  
but having more and "call-out attention" information in the main summary page is way better,  because it's there where everything starts.  You detect something that is not expected..  then you can drill down..   But that's exectly my point about locking information in the summary page..   the level of information got reduced from 3 metrics  (lockwaits,  deadlocks,  Locks-Escalations) to just one  (lock waits). 

Here is a DSM main dashboard sample 

As you can see,   in the top-right corner, we have a button that allows us to select the metrics we "want" to see. 
Each metric selected is plotted in a graphic.   so,  here,  the result is that we have 3 isolated graphics for locking information. 
I don't mind having these 3 metrics in a single graphic,  with 3 lines..   that's good as well, and has the advantage of occupying less space in the dashboard. 

My original post was an attempt to enhance the information even further..  Besides having the 3 metrics,,  also have a better understanding of distribution on lock-waits..  are they important or not ?  

Regards

It tells me something, but hard to say if I had lock contention or not.  
It tell me that on average I have 20 lock-waits/min.. and a peak of 130 lock-waits.   
Ok.  But what kind of locks are they? Does the peak, really represents a lock contention , yes or no? Hard to say from it. 

Those activities that waited for locks,  they waited too much , compared with the activity total time ? Or they were just regular fast and despicable lock waits that last just few ms that happens normally in a OLTP system ? 

On DSM,  you had ALARMs for lock-waits,  where you can set threshold (warning and critical level) for activities that waited on locks for more than a specified % threshold.   

Now,  wouldn't be nice if we can have all of them together, in a single graphic ? 

For all lock-waits detected in a monitor interval..   
 - How much of them waited less than 10% of the activity time?   (low impact) 
 - How much waited between 10 and 40% ? (medium impact) 
 - and how many were highly impacted,  above 40% ? 

What about a graph that could show us all of that..  ? Something like this..  using lines for the 3 basic metrics. and stacked bars for the lock wait distribution 

or,  a more similar with the time-Spent graphic,  with stacked Areas 

Personally, I like more this last one. 

with this kind of information,  we have much more information now..  We can see for example, that we had two peaks of lock-waits..  but just one of them really represented a real contention.  The second one,,  was just workload increase related, as all of them just waited no more than 10%. 

Off course,  the range of 10% and 40% here are just examples.. 

The determination of such values to set the ranges (low,  medium and high) should be detected automatically by statistical ABC or PBA analysis,  and/or allow the user to set these thresholds. 

Regards