I have PI 2012 with AF installed on a single server.  We use EventFrames for production tracking (creating say 100 events a day, each with 25 attributes...but we will be expanding that considerably as we add additional event frame based products/solutions).  I also have EFGEN running on that same server.  

The hardware is a Dell PowerEdge R520 with an E5-2470 processor (2.3GHz).  It has 8GB of RAM (which currently shows 768MB available at the moment).

Currently we only have about 1500 tags or so.  My plan is to expand that to say 5000 over the next year or two.

My questions are:

1.  Does this seem like a bad idea to have it all on the one server?

2.  Assuming I wanted to change that now...are we talking major surgery to setup a 2nd server with SQL Server on it and move AF to it...then move EFGEN off to an interface node?

Thanks.

Hello!

I am trying to build a simple element relative trend in ProcessBook. I perform the routine steps of selecting an AF Database, AF Element, and an AF Attribute. When I click [OK] I get a blank trend and the element attribute "E.Temperature" says "Unable to cast object of type 'PISDK.PISDKClass' to type 'PISDK.PISDKClass'". I have built plenty of element relative displays without a problem and am not sure why this is happening now. BTW the result is the same regardless of the AF Database, AF Element, or AF Attribute used to build the trend. Any feedback is welcome.

Hi,

I hope somebody would/could help me with this.

I want to find from a stringtag the lasted entry that is equal to a part of a string (CIP_MAN) during a start- and end time.

So fare I filter the tag with the following filter expression.

Instr ('NLOS:IBATCH:BE8913:BETA14:StateChange:GEIB', "CIP_MAN")

I had allready the following that did the trick but it has only an end time and it is looking to the first value back in time from this endtime but I wanted to stop the search at the start time. Because else it could be looking trough years of data.

PINCompFilDat('Tag';End Time;-1;

Instr ('NLOS:IBATCH:BE8913:BETA14:StateChange:GEIB', "CIP_MAN")

;0;0;"PISERVER";"auto")

I hope it is clear and that somebody could point me to the right direction.

Hi guys!

I came across something really interesting last week and I'm hoping someone can shed some light on my observation and guesses.

Background

We had a few heartbeat tags archiving heartbeat readings from our interfaces every 5 seconds for the past 1 year.

Those tags were incorrectly setup to archive as we use PE tags to monitor our interfaces by interrogating the previous event's timestamp of the heartbeat tags that we use, and it's the results of the PE tags we want to archive.

When I came across this, I started deleting all the useless (to us) heartbeat readings from the heartbeat tags.

That was about 19 million events over the span of one year, and performed the cleanup using the Archive Editor in SMT.

Now, compression on those tags was off, and out of curiosity I checked the size of the archive before the deletion and after the deletion to get a sense of the degree of impact the storage those readings had on our archives.

Much to my surprise, the archive sizes didn't change at all after the deletion was complete.

Reading though some KB articles on PI archives this morning, I decided to re-process two of our archives that contained data from the past 1 year, and thus would've contained the events from the heartbeat tags that were cleaned up.

Once I reprocessed the archives, their size was signficantly smaller (one was smaller by 100MB).

We do not delete tags, nor do we allow event deletions on non-admin tags as it's against our system policy (for audit purposes), thus I'm inclined to believe the only reason these archives shrunk is due to the deletion of the 19 million heartbeat events archived (fragmented of course across different archives based on the event timestamps).

Questions

1) If events are deleted via the Archive Editor plugin in PI SMT, are they are simply marked in the archive(s) as say "deleted", but are not really deleted (ie. removed from the archive(s))?

2) The only way tag readings/events are actually removed from archives is by reprocessing the archive(s) after having the events deleted in the Archive Editor plugin?

3) If events are removed using the piconfig tool, is the behavior the same as deleting events using the Archive Editor plugin? For example, using the commands:

  @table piarc
  @mode edit, t
  @istr tag, time
  @modi mode = remove

    or does piconfig remove the events from the archive(s) without the need to reprocess?

Thanks a lot guys!

Cheers,

Mario

Hi

We would like to validate data comming into PI beyond the quality stamps it gets from the OPC interface.

Lets say we have a tag which a range of 0 to 100 and we want to check if the values suddenly has erratic behavior like jumping from 10 to 90 in a timespan not possible in the real world. When that happes I want to be able to mark the value as "bad" or questionable.

I realise I can check this from ace or Abacus, but can I also write to the point marking it bad?

/Lars

Hi all,

I've tried to find the answer of my question in many places (this forum, tech support website, etc.) but couldn't find it.

I'm using PI Server 2012 set up as a PI Collective, and I came across a problem when try to write data to PI Collective using PI buffering Subsystem.

I've installed PI AF Client 2012 SP2 on my client machine, and in my code I'm using methods from OSIsoft.AFSDK.dll to save data to PI Collective.

I've read that new AF SDK doesn't support buffering...("The .NET 4 AF SDK does not support buffering of PI Server data writes nor direct fanning of data writes to PI Server Collectives.") and that Buffering of PI Server is scheduled for a later AF SDK release.

I've also read that users who need to write data to PI Servers should use .NET 3.5 AF SDK, with buffering enabled. Where can I found .NET 3.5 version of AF SDK?

Is there any other workaround to solve my problem? I must use AF SDK. With PI SDK, everything works fine.

Tanks

Hello folks,

I need a little help with pirecalc.exe.

I have a PE named PE-CALC-TEST with this expression:

IF TagVal('PE-CALC-TEST', '*') = 0 AND TagAvg('MONITORED-TAG', '*-3m', '*') > 34 THEN 1 ELSE IF TagVal('PE-CALC-TEST', '*') = 1 AND TagAvg('MONITORED-TAG', '*-3m', '*') < 34 THEN 0 ELSE NoOutput()

The goal of this equation is change himself based on MONITORED-TAG. And seems work fine to me.

But.. I need to recalculate my last month for this condition and in my mind use the
 

pirecalc.exe /execute=PE-CALC-TEST,"01-jan-4","01-feb-14"

would  solve my problem but the archive values does not update.

It is a problem with my PE or with the way i tryed use the pirecalc?

Thanks a lot.

Hi,

Is there any way to disable sort option in Coresight tables?

I'm trying to emulate a bidimensional table in Coresight like this:

We would like to disable the sort option to keep the initial relation with other table rows

Thank you in advance

I have a list of batches that I want to search for in PI BatchView.  Is there a way to search for all of the batches at one time?  I am using the add-in in Excel 2010.  I have tried selecting the list in Batch ID field, but it generates no results.

Thank you.

A recent question on vCampus from a Batch Pharma Customer:

"As long as not all client tools do support event frames (RtReports) I'm struggling with the question what's the best way to move forward with new implementations.  We do need the batch events because of rtReports for our  deviation reporting purposes, but besides that we also want to start using the CoreSight & Mirabo Livepoint product and I prefer to go for the new Event Frame features.   it would be nice if there was a kind of BE to EF sync interface (or vice versa).  Maybe there's already an easy solution out there, but at this moment I'm not aware of it ...    What do other customers do?  How do they solve this?"

=====

Here is my response ... I wanted to make sure batch customers see where we are on this topic.

First, I know from talking to our batch customers that just having EFGen or our Batch Interfaces or the upcoming PI Server Batch DB to EF migration is not enough to enable you to move to PI Event Frames.  It needs to be the ‘whole package’, whatever this is for your specific situation.  For the majority of existing pharmaceutical PI Batch customers, the ‘whole package of EF support’ is Interfaces + Server Database Migration + PI DataLink + Overlay Trend (BatchView replacement) + RtReports.  This formula is essentially is ‘functional equivalence’ for what most have ‘out of the box’ with PI Batch today.  I can tell you that we’ve made significant progress on the PI Event Frames roadmap but we have not yet crossed the threshold where I believe the majority of batch customers can move to PI Event Frames.  As such, our standard recommendation for batch customers is to go with PI Batch today.  For non-batch customers (like other industries or other non-batch event use cases in Chem/Pharma) the recommendation is to start with PI Event Frames.  The recommendation is different because different companies are coming from different starting points.  So for batch customers, the gaps we have today are mainly the Overlay Trend and RtReports. 

Overlay Trend (PI BatchView replacement) – First, we do not plan on adding an overlay trend into PI ProcessBook.  There are several reasons for this but the main one is that we don’t believe this would be the fastest route to delivering this functionality to our customers do to the aging code base of PI ProcessBook.  We also want to invest our engineering resources into our visualization innovation platform which is PI Coresight and put our customers on a path to the future.  So our plan is to add overlay trend functionality into PI Coresight.  Your next question is ‘when?’  For this question, I don’t have a definitive answer today but we are working on this plan and this is the next highest priority for visualization of EF.  In terms of delivering this functionality, there is a lot of other related work to making this happen in terms of setting our customers up for the future (such as transitioning PI Coresight to HTML5 from Silverlight).  So while I don’t have a date to give right now, it’s likely to not be in 2014.  In the meantime, Axel is right that functionality like the Overlay Trend and other types of EF visualization can be supplemented by 3^rd Party partner products such as Mirabo Livepoint. 

RtReports – Our RtReports development team is working on revamping our printing mechanisms to not rely on Word which should resolve many of the issues our customers have with printing.  This release is expected as early as the end of March or as late as sometime in May.  We are in testing now.  After this release, we will turn our attention to adding AF and EF support in RtReports as we know this is a pre-requisite for existing RtReports customers to be able to move to PI Event Frames.  Again, there is no date we can put on the roadmap at this time (but this will happen in the months after the upcoming release, RtReports 3.4).  While I don’t have a date to give right now, it’s likely to not be in 2014.  Alternatives to RtReports on PI Event Frames would be to use our PI System Access products (PI OLEDB Enterprise) to write SQL queries against EF leveraging a reporting solution such as SSRS or Crystal Reports.  But this approach is obviously not the same as configuring a report in product like RtReports, but I did want to mention it as an option, for completeness. 

Batch & EF Sync – Regarding Axel’s suggested Batch to EF sync interface that would essentially enable running PI Batch and EF in parallel … this is not an approach we are taking with our PI Server 2014 Migration functionality which will be a 1-time migration of the old Batch Data to EF and then PI Batch will be marked to read-only.  There are many reasons why we are not doing a live synchronization approach but the main one is that it significantly increases the technical complexity of the implementation and would delay our release to a time period that isn’t desirable for customers either.  So it’s a lot more work and it doesn’t get our existing batch customers to PI Event Frames any sooner than our current plan.  We also don’t recommend that customers run dual instances of the Batch/EF Interfaces in parallel in production (one writing to PI Batch and another writing to PI Event Frames) because this approach will likely increase the migration complexity significantly when it becomes available in PI Server 2014.  Our design will not account for this. 

Summary – In summary, based on where we are currently with our released and upcoming products supporting PI Event Frames, our recommendation is that PI Batch customers continue to use PI Batch today.  We do not have a complete end-to-end solution for batch customers at this time.  This is the focus of Wave 3 of our PI Event Frames roadmap (we are currently finishing Wave 2).  We don’t have official dates for the overlay trend or RtReports today, but it is not likely to be delivered in 2014.  When we have dates available, I will put them on our official Product Roadmap. 

Also, below is a checklist I keep on this subject.  It is also important to note that customers that have a lot of custom PI Batch applications (or third party product integration) is really the focus of Wave 3. 

I understand that this is not the best news to hear but I do want to be honest and upfront with our batch customers and let you know everything that I know as of today.  We’ve built a lot of great EF functionality up in the PI System over the past several years and we are getting closer but we still have some final steps to take so that a majority of our batch customers can move to PI Event Frames.  I’ll continue to update everyone as more information becomes available. In the meantime, most existing batch customers should continue to use PI Batch which end-to-end still provides more value than PI Event Frames for batch customers at this time. 

Thanks,

Todd Brown

Product Manager, PI Batch, PI Event Frames, RtReports

Background:

First, for a recent summary of the status of PI Event Frames for batch customers, refer to the following post:

http://community.osisoft.com/index.php?/topic/1899-batch-customer-roadmap/

We've been receiving more and more questions of late from batch customers asking if it's possible to run PI Batch Interfaces in parallel with PI Event Frame [Batch] Interfaces so that data is stored in both the PI Batch Database and PI Event Frames database redundantly. 

The thought with this approach is that it would potentially enable batch customers to continue to use batch clients they need (like PI BatchView overlay trend and RtReports) but still enable them to take advantage PI Event Frames advantages such as flexibility, scalability, and new client support in PI Coresight and PI DataLink.

In terms of historical batch data, customers have asked if they can just run the EF Interface instances in history recovery mode to backfill that data into PI Event Frames and then they wouldn't need to run the PI Server 2014 migration functionality when released. 

As a product manager, when I hear these customer questions, I see them as clear evidence that batch customers are very excited about moving to PI Event Frames as soon as possible and are even willing to consider redundantly storing data in two places to get there sooner (typically these approaches are greatly frowned upon from an IT and Quality perspective).  At the same time, it also points out the frustration that some of our batch customers feel with the progress we've made on the PI Event Frames roadmap that they are considering unproven sub-par architectures that result in redundant data and redundant implementation work and could potentially result in problems down the line just to get to use PI Event Frames sooner.  This tells me that we need to continue to work to deliver EF support for an Overlay Trend and RtReports as soon as possible (along with all the other work we already have in progress). 

So, does OSIsoft recommend running PI Batch and PI Event Frames interfaces in parallel?

The bottom line is that OSIsoft does not recommend running PI Batch interfaces and PI Event Frames interfaces in parallel in production (this includes the suggested approach where customers don’t plan on using PI Server 2014 migration functionality).  Why?

There are several reasons why:

1. It is generally not a best practice to store the same data into two identical databases within the same system.  These architectures also typically require pre-approval from quality departments. 
2. Our PI Server 2014 migration strategy will not take into account customers running interfaces in parallel.  The result could be duplicate EF records which would require extensive manual work to clean up. 
3. Since these batch execution system interfaces write to PI tags, it's very likely that a customer may configure both instances to write to the same PI tag and cause duplicate values. 
4. This architecture of PI Batch and PI Event Frames running in parallel potentially keeps batch customers running PI Batch indefinitely whereas we believe it's in batch customers best interest to have a strategy to move them fully forward to PI Event Frames at a time in the future. 
5. OSIsoft has not designed nor tested our batch and EF interfaces to run in parallel so the behaviors in various situations are unknown.  So using the interfaces in this configuration is outside of the intended design. 
6. In regards to our migration strategy from PI Batch to PI Event Frames, this architecture and design is outside of the design assumptions we're making in our migration strategy and as such is a higher likelihood to fail and cause problems for customers in the future.

For these reasons, OSIsoft does not recommend that batch customers run PI Batch and PI EF Interfaces in parallel for the same batch data.  At this time, our standard recommendation is that that most batch customers should continue to use PI Batch over PI Event Frames because it still provides the most complete functionality for this use case at this time. 

The biggest remaining gaps we have to enable most customers to move to PI Event Frames are the Overlay Trend functionality (PI BatchView replacement) and support for AF/EF in RtReports.  For RtReports, this work will begin following the upcoming printing release (RtReports 3.4).  For the overlay trend, we are planning this functionality for a future version of PI Coresight (details TBD) and working it into a plan with some of the other technology changes that need to happen with PI Coresight at the same time (such as Silverlight to HTML5 transition). 

While existing batch customers will likely be disappointed by this news, we believe it is in our batch customer's best interest not to take this unproven approach.

Sincerely,

Todd Brown

Product Manager, PI Batch, PI Event Frames, & RtReports

We are using PI Datalink (Excel) to pull some information from a digital state tag. 

Today we were asked to do something unique; pull the state number rather than the actual digital state condition.

Example of Digital States:

0 = Stopped

1 = Running

2 = Waiting

Question: Is there a way to change the equation to return the state numbers (0,1,or 2), rather than the state condition (Stopped, Running, or Waiting)?

Example:

=PIArcVal("0395.VP.CASEPACKER.STATUS","2/20/2014  7:00:00 AM",0,,"auto")

Thanks,

Brian

Hi guys.

I need some help.

I'm building a process book PDI file and i need to know the functions that can be used accord to "PI calculations Datasets".

Hope can get some help. 

I already know this ones:

1. TagValue

2. TagAvg

3. TagTot

I need the "Time Filtered" expression.

Ty.

In excel I've a column of timestamps, I want to find the fraction of the hour following each timestamp that the value of a tag is above a certain value. I don't think this should be too tricky and having spent a lot of time looking at this I feel that a TimeGE expression is probably the way to go but I just can't make it work, can anyone help? Perhaps explain like I'm 5.

Thanks

Hi everybody,

I bought on Thursday Virtual Learning Lab Environment - One Month Pass, but can't lounch any session til now. Checked connection, and everything is working as it should from my side.

When I'm trying to lounch the environment, I can see only communicate:

Session Start Time:    Object reference not set to an instance of an object.

And then, after closing the session, what's the only possibility:

Description: An application error occurred on the server. The current custom error settings for this application prevent the details of the application error from being viewed remotely (for security reasons). It could, however, be viewed by browsers running on the local server machine.

Details: To enable the details of this specific error message to be viewable on remote machines, please create a <customErrors> tag within a "web.config" configuration file located in the root directory of the current web application. This <customErrors> tag should then have its "mode" attribute set to "Off".

I know that it may take til 48 hours to wait for active session, but it's already 4th day.

Is it sth wrong with settings from my side?

I tried to connect withouth firewall as well, with same effect.

How can I get the access to platform?
 

Hello,

I´m trying to do a live value plot of the sum of 8 signals in Processbook, problem is 5 are stored in one Server and 3 in another one. I´m able to create and plot 2 PI Calculations datasets: one for the 5 tags in server1 and another for the 3 tags in server2 but I´m unable to add all 8.

Using excell and ODBC I´m able to add all 8 but only for historical data, not a live value trend like I want.

Does anyone know how to do this?

Regards

Rafael

Hi everyone,

I've just read the latest OPC interface release note and have come across the trend advise function. I think I follow what it does but I just need clarification on 1 issue; does the PI server record the previous value as well as the new value so that PI data retrieved by Datalink will tie-up with what is shown on the trend?

Example; a plant data value changes from 1 to 10 (1, 10, 1, 10, 1, 10, etc) every 30 seconds. Normally this would give a sawtooth profile, but with trend advise I would get a stepped profile. So far so good. Normally (excluding the effects of exception and compression) the PI archive would store the original data profile (1, 10, 1, 10, 1, 10, etc) , but does the trend advise add extra data events to the PI archive to allow the PI trends to show the stepped profile, i.e. 1, 10, 10, 1, 1, 10, 10, 1, etc, when looking back at a trend over a number of hours/days/months, or is it just a live trend update function, that if you refresh the trend then you would get a sawtooth profile again?

Does my question make sense??

Hello!

I'm new to PI and have been experimenting with PEs to determine local min/max's.  

I have tag which operates mostly consistently but occationally drops down (sometimes sharply, sometimes not) before going back to normal.  

In 'calculated data' there is a range function. Effectively I want to know the anytime the range is greater than some number x. And then I want to know the min and the max for the time frame determined by that range.

So the issue I'm having is that I need PI to tell me when (start and end time) one of these cycles is happening.

I have tried using limits (abs((PrevVal('TAG','*')-NextVal('TAG','*')) > 200) but this doesn't necessarily define a range in fact, I get multiple instances where this is true for one cycle. So it cuts the digging down but it is still not exactly the output that I want.

Many thanks for your responses!

K.

Hello everyone.

I installed PI UFL (Universal File And Stream Loader) in the PI Server, and when i want to create a new interface, i have this problem.

I've tried to create it in the PI SMT as well.

I hope get some help.

Thank you.