Hahnming Lee
My feedback
-
13 votesHahnming Lee
supported this idea
·
-
6 votesHahnming Lee
supported this idea
·
-
9 votes
-
7 votesHahnming Lee
commented
Unfortunately, as was mentioned in the other UV request, this would require most of our AF server client products to start being version aware. Since we're still cataloging use cases, this is likely not something that will happen in the near future. The complexity in dividing the query into subqueries that could have different time ranges depending on the version and how far back it goes could make this both expensive and difficult to troubleshoot.
If versioning becomes more of a focus in future versions (which is no guarantee), the other clients could revisit this, but for now, you would likely need to investigate custom coding to be able to accomplish this.
-
8 votesHahnming Lee
commented
Import/export never takes versions into account. It sounds like you have a use case for using versions, and it would be good to know what it is. Almost none of our modern client products actually take versioning into account when displaying data, so our understanding of why people need/want it is somewhat limited.
This is one of those features that is often argued as a bug by some and a feature by others. Some customers have said that they use import/export to "delete" old versions, while others want it preserved.
-
13 votesHahnming Lee
commented
In response to Felipe Rejes, "Hey Hahn, I am not sure why but the desc..."
I think we're looking for the specific use case of the delivery channel. What does it do? What is the limitation right now?Hahnming Lee
commented
This seems too broad. How would you be using it, and why don't the current ones satisfy the customer's need? The current ones certainly can't do everything, but what is this specific use case for why the customer needs something custom.
-
18 votesHahnming Lee
commented
Where? Are you just looking for something that alerts you when two analyses are writing to the same place, regardless of where it is in the database (or even server)? Is this a common scenario for the customer, or do they do this unintentionally during configuration? If it's unintentional, what is the common path to getting into that scenario?
You can get a higher level look at your analysis via PI Builder and pulling in attributes and analysis. This will give you all the PI points data reference string configurations associated with attributes. While this won't exactly map it together, you can look for duplicate config string (or tag references) across attributes.
-
14 votesHahnming Lee
commented
This seems like a recipe for a lot of data confusion. Like if it's event-triggered with no data, you would still want a default value written? Wouldn't that effectively be just replacing the error value like Calc Failed with a #? All users would think of it as the sam, and if the value is valid, how would you know?
It seems like the underlying idea would be to be able to better account for error handling? Not sure what form that would take, but something that would likely happen before a data write.
-
3 votesHahnming Lee
commented
Is this only for PI Points? It's not valid for custom DRs and Table Lookups.
-
5 votesHahnming Lee
commented
What is overfilling an archive? % of space taken in an Archive as compared to total PI count? Since fixed archives are time based, I would think it's natural for some tags to have more data than other tags.
-
8 votesHahnming Lee
commented
Can you elaborate why? Is this about getting to some other desired behavior (breaking out of the calculation since there is no result or trying to "skip" certain parts of the analysis)? Is this being used with Analysis DR? OR a custom DR?
It's unclear to me to immediately see the use case outside of what is being described above. If I am not writing to PI, and ignoring a custom DR, I don't see a reason I would skip an output.
-
5 votesHahnming Lee
commented
This seems a bit broad and dependent on configuration. Since clients are the ones asking for info, are you saying that regardless of if a client is requesting info, the AF server should maintain some sort of connection with an external data source and periodically check the health of the connection? Would this be a Performance Counter that is tracked by table connection?
What does it mean to look for stale attributes within an element? Since the attribute is configured with a data reference, this seems to be a separate request to track the staleness of the underlying data. Even if timestamped and conceding the fact that they are all different data references, what is defined as stale in this scenario? Is this a user-defined mechanism, and is it at the attribute level? If so, you could use an Analysis to figure out the previous timestamp and do comparisons.
-
3 votesHahnming Lee
commented
So it sounds like there are two enhancements. One is related to informing people when a backfill has occurred at all, and then providing context of which EFs were backfilled.
I have another question about the other one. If they have late arriving data that has missing information, I would likely not alert on that until I have confidence in the data. Is the customer looking to use the Notification as an alert on whether or not data is missing? How do they verify correctness or that the data is missing in the first place?
Hahnming Lee
commented
Are we alerting that the event frame itself was backfilled at some point and to include the listed info if the EF was backfilled? Or what Steve said, that a backfill has occurred?
Why is it important to know if this is backfilled? How would that affect user response on if it was generated real time?
-
5 votesHahnming Lee
commented
Is this a troubleshooting mechanism for Notifications? Have you looked into the Notification history (2012 or earlier) or the Event Frame annotations (2016 R2 or later)? It wouldn't seem productive to provide a list of failed Notification sends (if it indeed failed) as if it can record the failure, it should have just sent. So I'm curious if this is a deficiency in the product functionality (in which case a TS case might be best to address it) or a deficiency in the product feature set.
As for manually sending the Notification, I'm also curious what this would be. Are you saying that the Event Frame is already created, but you want to send out a Notification based on it at any given time? Is this to address the failure of sending or just as a manual resend based on an administrator's preferences?
-
7 votesHahnming Lee
supported this idea
·
-
4 votesHahnming Lee
commented
We have a YouTube playlist that looks like it covers almost all of these requests, I believe. Have you checked it out?
https://www.youtube.com/playlist?list=PLMcG1Hs2Jbcs43wQ2qO2dxW72PqGhZjKW
-
16 votesHahnming Lee
commented
It sounds like this may actually be a request about the creation of the Event Frame rather than the sending of the Notification. Or is it both? Also, if historical EFs are getting created, why is the preference to receive an e-mail instead of just using a client tool to look up the history?
It's unclear what this request actually is. What specific part of PI Analysis needs to be read/write? What specific part of Notifications need to be read/write?
I can see why PI AF HA Collective can be seen as a preferable solution to a SQL Server with AlwaysOn, especially if there are current implementations out there, but why is it preferable to enhance Collectives.