35 Replies Latest reply on Oct 21, 2011 5:59 PM by wpurrer

    Coresight for PI - Server   "PI - Server Version Easy"

    wpurrer

      Hello vcampus!

       

      Sorry the last couple of weeks I was a little bit absent from vcampus. But i really like the "coresight" product, i hope in a couple of years 80% of my visualization needs come true with the product...

       

      With visualization moving in the right direction ;-) i see my next "mission" to give ideas move the PI - Server forward.
      What would be a pi server where the "management" is so easy like the visualization on PI - coresight.

       

      Positive Formulated: I just like add tags and data to the pi server.
      Negative Formulated: I don't like to take care about reprocessing archives when i like to add  new tags and data from the past.

      Pos: I like to let my keyusers add tags and formulars without they risk the security and stabilty of my server.
      Neg: I don't like to have to "maintenance" the tags by myself, because if i give the users  the right to create tags they can do everything on the server.

       

      Pos: I like to reveive the data fast
      Neg: I don't like to take care of speed and compression of the data in the archives and  right tags ..( raw tags, daily avg, month avg,....)
      Nice article by the way: http://www.nt.ntnu.no/users/skoge/prost/proceedings/escape16-pse2006/Part%20B/Volume%2021B/N52257-Topic3/Topic%203-%20Poster/1398.pdf

       

      Pos: I like that the pi server itselfs sends me a message when something is wrong.
      Neg: I don't like to create a couple of Notifications rules, scripts to monitor the server or spend money for expensive EA - NOC.

       

      Pos: i like to give the users the full control about their calculation and it should be possible to easy to use.
      Neg: I don't like to have x tools with different limitiations for solving the same problem (Processbook calculations, PE, ACE,...)

       


      Whats your vision of the future pi server ? Whats osisoft statement ?
      If needed I can give for every point a couple of usecases ...

      Regards from Austria
      Wolfgang

        • Re: Coresight for PI - Server   "PI - Server Version Easy"

          Wolfgang, on your point about creating tags...I fully agree.  I mentioned in my post on future data in the PI server (vcampus.osisoft.com/.../2015.aspx) that the server should take care of these type of things.  Especially in a collective where you shouldn't care about keeping a collective in sync, it should take care of itself - data and configuration.  

           

          When you create a tag, the server should have a maintenance subsystem to take care of reprocessing archives 'online' and be intelligent enough to only add the new tags to those archives that are writeable - if someone makes an old archive writeable then the subsystem adds the relevant tags etc

           

          To add to your list, I would like to see Interfaces completely moved to AF with templates for monitoring installed as part of an Interface installation.

           

          Oh, and SSB

            • Re: Coresight for PI - Server   "PI - Server Version Easy"

              By the way, how is your mission going to spend less than one hour a month on maintenance & administration?

              • Re: Coresight for PI - Server   "PI - Server Version Easy"
                aabrodsky

                Rhys @ RJK Solutions

                Oh, and SSB

                 

                Rhys, you won't let it go away

                 

                If we ever get SSB released, we all know who to thank.

                  • Re: Coresight for PI - Server   "PI - Server Version Easy"
                    MichaelvdV@Atos

                    Interesting post Wolfgang, thank you for that.

                     

                    I like your idea of having a 'centralized' place where you can manage different kinds of calculations (PE, ACE, maybe StreamInsight, AF, etc). I'm not sure how this has to be done, but it's a very interesting idea! Maybe when Configured Analytics will be available, a lot of the analysis and calculation needs can be provided from a single place. This is just a thought though.

                     

                     

                     

                    edit: spelling

                    • Re: Coresight for PI - Server   "PI - Server Version Easy"

                      Alex Brodskiy

                      Rhys, you won't let it go away

                       

                      If we ever get SSB released, we all know who to thank.

                       

                       

                      Haha, yeah I do go on about it too much don't I...

                       

                      To be honest, part of me would be satisfied if client side buffering could buffer & fan to multiple collectives from a single node and if the PI server had better failover - for example, if the primary goes down clients failover to the secondary but when the primary comes back up clients remain on the secondary; the primary right now doesn't pull back (or even notify) the clients that were connected before that it is available again.  
                      I'm in the middle of a project where I am just putting in a PI collective but there is so much extra effort to ensure all interfaces and custom applications writing data fan to all members, which my vision of SSB would solve.  Once everything is set up it works fine but once the primary is down and clients connect to the secondary , PItoPI interfaces etc are going the wrong way...anyway, rant over.

                       

                       

                        • Re: Coresight for PI - Server   "PI - Server Version Easy"
                          mhalhead

                          Michael @ OSIsoft

                          Maybe when Configured Analytics will be available, a lot of the analysis and calculation needs can be provided from a single place

                           

                          Like SSB for Rhys this is my pet topic. It really would be nice to see this.

                           

                          Wolfgang Purrer

                          I don't like to take care of speed and compression of the data in the archives and  right tags ..

                           

                          This was a nice article. Like Wolfgang tuning the PI points is a right royal pain in the ... and to be honest we don't do it particularly well. With nearly a million points across our PI servers it is understandable. It would be nice to have something in the product that made this easier; or better yet removed the need to manually tune the points entirely.

                            • Re: Coresight for PI - Server   "PI - Server Version Easy"

                              Michael, Wolfgang,

                               

                              Today I looked at one PI server that had no compression at all and another PI server that had incorrect/restrictive compression for various instrument types.  I certainly sympathise on the administrative overhead of tuning the compression, but I wonder if it is too difficult a task to define corporate rules that are applied to all PI servers by some system.  Would be a start to model all types of equipment, instruments, measurement points, ... that a corporation uses (e.g. Smart Plant Instrumentation) and define rules based on UOM at that level.  You may even define different rules for the same instrument type that is part of different processes or equipment.  

                               

                              Then transferring those rules to the individual PI servers where they are either adopted or rejected - there are some sites within an organisation that always have, maybe even need, their own way of doing things.

                               

                              Anyway, just thinking out loud.

                                • Re: Coresight for PI - Server   "PI - Server Version Easy"
                                  andreas

                                  @Rhys:

                                   

                                  Defining corporate rules would be a nice feature - however I do not see an easy way to do it. If you choose the UOM - you have to keep in mind that the UOM in PI is only a string, so the same that applies for the compression, is valid for the UOM - you have to make that extra step and configure it properly. If you choose the Span (what the field service often does) you have to make sure that it is properly set.

                                   

                                  So at the bottom line instead of configuring the compression settings you will have to configure other data properly and rely on this.

                                   

                                  There is a devnet tool "Archive Analyzer Add-In for Excel" in the library (vcampus.osisoft.com/.../library.aspx) that helps you determine the amount of data you are storing in the archive - and basically use this measurement to identify the tags that need some "tuning" for the compression.

                                    • Re: Coresight for PI - Server   "PI - Server Version Easy"
                                      mhalhead

                                      Rhys @ RJK Solutions

                                      but I wonder if it is too difficult a task to define corporate rules that are applied to all PI servers by some system

                                       

                                      To a degree we have defined rules regarding data logging. We are fortunate that most of our control systems are from a single vendor and we try and keep the versions in line. We also have our own block library. This has allowed use to develop a tool that extracts the point configuration information straight out of the control system based on templates; we get the following information:

                                      • Zero
                                      • Span
                                      • Descriptor - a combination of the block comments, the attribute comments and rules related to type, name and hierarchy.
                                      • Instrument tag
                                      • Point source (with some manipulation of the plant hierarchy) - we split our points up according to a S88 model.
                                      • Eng Units
                                      • Point name

                                      The tool allows us to keep the two systems in sync and generate reports on the changes both within PI and the control system. We developed the tool with a local company IDX (Industrial Data Xchange), a quick plug for IDX (I'll have to ask for a discount).

                                       

                                      The solution above goes a long way to making the solution managable; particularly as we a quite dogmatic in the design and commissioning stages of projects regarding the correctness of information. However, in reality this probably only covers 70%-80% of the work. Defining a set of rules that catches every possible scenario is virtually impossible.

                                       

                                      Andreas

                                       There is a devnet tool "Archive Analyzer Add-In for Excel"

                                      I'm aware of this tool. We have a slightly modified version. Although this tool is useful it still requires manual intevention and it is quite time consuming to run. I would prefer something that is a bit more automated. Something along the lines of a plugin to SMT that shows the worsest tags in terms of compression settings (both under compressing and over compressing) and allow the administrator to schedule them for "auto-tuning". The system would then go off and figure out what the "best" tuning parameters are and present those to the administrator; the admin can then either accept or reject the tuning parameters. I wouldn't want the system to automatically change the settings; imagine the chaos if you tune while the plant is on a maintenance shutdown! I remember seeing a tool like this from one of OSI's partners but I can't remember which. Anyway I believe that this should be part of the product; Pat Kennedy was talking about PI systems spanning 10's of millions of points at the UC. The adminsitration of systems this large becomes critical.

                                       

                                      Personally I would rather have an under compressed PI Point than an over compressed one. I can't recreate the lost information.

                                       

                                      Of course the other alternative is that OSI changes the compression algorithm from the Swing Door Algorithm to something like Wavelets or Fourier decomposition. No I'm not serious. Wavelets, Fractals, Fourier. ... work well for certain classes of data but they aren't a silver bullet.

                                        • Re: Coresight for PI - Server   "PI - Server Version Easy"
                                          Ahmad Fattahi

                                          Maybe not exactly the answer to the posed question but along the same lines: The Archive Activity Grid (AAG) is a tool to monitor the recent read-based activities around certain PI tags. It can report the most widely asked-for tags, list all tags who were read, number of events read, the most demanding clients, among other things. Please see this KB article for more information.

                                            • Re: Coresight for PI - Server   "PI - Server Version Easy"
                                              mhalhead

                                              Ahmad,

                                               

                                              This would be useful to hit the commonly used data and only tells you which tags people are using. The lesser tags would be caught; these tags are only needed when something goes wrong and it is problematic if the data isn't there. One thing that I should mention is that tuning really only affects your real/double tags; we set compdev to 0 on ints, digitals and some real tags like limits and setpoints; those enterprise rules. So the total number of tags to tune is generally small in comparison to the total; at a guess in the region of 20%.

                                                • Re: Coresight for PI - Server   "PI - Server Version Easy"
                                                  andreas

                                                  Michael Halhead

                                                  I would prefer something that is a bit more automated. Something along the lines of a plugin to SMT that shows the worsest tags in terms of compression settings (both under compressing and over compressing) and allow the administrator to schedule them for "auto-tuning".

                                                   

                                                  That would be a nice little tool - taking all integer and float tags, and look for high # of events and low # of events. Just out of curiosity I have attached a little code snippet for a command line tool that returns a comma separated list with increased exception and compression for undercompressed tags and decreased exception and compression for overcompressed tags. There is lot's of room for improvement, how about making a community project to build such a set of sysadmin tools?

                                                   

                                                  Michael Halhead

                                                  Personally I would rather have an under compressed PI Point than an over compressed one. I can't recreate the lost information.

                                                   

                                                  I totally agree on this - nowadays most of the PI Systems around can handle under compressed data very well. 

                                                   

                                                   

                                                   

                                                   

                                                    • Re: Coresight for PI - Server   "PI - Server Version Easy"
                                                      MichaelvdV@Atos

                                                      Andreas,

                                                       

                                                      I'm a little curious about the following statement in your samplecode:

                                                       
                                                       strWhereClause = "pointtype = " 
                                                                                   + PISDK.PointTypeConstants.pttypFloat16.GetHashCode().ToString()
                                                                                   + " OR pointtype = "
                                                                                   + PISDK.PointTypeConstants.pttypFloat32.GetHashCode().ToString()
                                                                                   + " OR pointtype = "
                                                                                   + PISDK.PointTypeConstants.pttypFloat64.GetHashCode().ToString()
                                                                                   + " OR pointtype = "
                                                                                   + PISDK.PointTypeConstants.pttypInt16.GetHashCode().ToString()
                                                                                   + " OR pointtype = "
                                                                                   + PISDK.PointTypeConstants.pttypInt32.GetHashCode().ToString();
                                                      

                                                       

                                                       

                                                       

                                                       

                                                      Why are you using the GetHashCode() ? This will give the hashcode (in this case I assume the value associated with this enum value). It seems wrong to me, and causes the query to be broken.

                                                        • Re: Coresight for PI - Server   "PI - Server Version Easy"
                                                          andreas

                                                          according to the PI SDK documentation

                                                           

                                                          PI-SDK Help


                                                          Legal Combinations:
                                                          pointtype = 12 OR tag = 'sin*'
                                                          tag = 'sin*' AND pointtype < 12

                                                           

                                                          it should work (and it works for me )

                                                            • Re: Coresight for PI - Server   "PI - Server Version Easy"
                                                              MichaelvdV@Atos

                                                              So, it seems I learned something today. You should use the HashCode from the PointTypeConstants in a getpoints query...

                                                               

                                                              Thank you!

                                                                • Re: Coresight for PI - Server   "PI - Server Version Easy"
                                                                  skwan

                                                                  Some random thoughts and questions ....

                                                                   

                                                                  Auto-tuning of PI tags.  I would think the rules to do this are specific to each individual and organization.  Does that mean we're talking about a rules/script type system to allow the application of these rules?  Are we talking about compression settings only?  What are some example rules that drive your PI tag attribute tunings?

                                                                   

                                                                  Are you using AF attributes to drive your PI tag attribute settings?  For example, use AF attributes to store your frequently "need to tune" PI tag attributes, use substitution parameters to tie the AF attributes to PI tag attributes.  Then any time you need to change the PI tag attributes, simply execute "create and update PI point" in PI System Explorer with your updated AF attributes and  you're done.  I can image you can use AF Formula DR or maybe even custom DR to execute your tuning rules.

                                                                   

                                                                  In many ways, it makes sense to store PI Interface settings in AF.  I'm having trouble envisioning using PI System Explorer as a way to manage PI Interfaces, i.e. PSE is not PI SMT.  Someone help me make that leap.....  I see interface configurations as basically a one time exercise.

                                                                   

                                                                  Have a good weekend!

                                                                   

                                                                   

                                                                    • Re: Coresight for PI - Server   "PI - Server Version Easy"
                                                                      mhalhead

                                                                      Hi Steve,

                                                                       

                                                                      Steve Kwan

                                                                      Auto-tuning of PI tags. I would think the rules to do this are specific to each individual and organization.

                                                                       

                                                                      IMHO the auto tuning of point should be a mathmatical algorithm. The only rule I would consider would be a % of data lose; the absolute error between the raw uncompressed data and the compressed data. You could possible include a rule regarding the minimum and maximum number of events per day. There are other factors that will impact the accurarcy of the data; for example the sample time (anyone remember Nyquist sample theory), the accuracy of the instrument, filtering on the controllers, the controllers sample rate, ...

                                                                       

                                                                      The creation of tags is where the scripting comes in.

                                                                       

                                                                      Steve Kwan

                                                                      Are you using AF attributes to drive your PI tag attribute settings?

                                                                       

                                                                      In our case we use AF to create the points on the central server. The points on the site systems however are created using the control system as the master. We currently don't represent everything in AF (we are working on this). Our central PI server doesn't do any serious compression (compdev is set to zero), we do an archive update as an exact match is more important (far less arguements). I suspect most companies are in a similar situation.

                                                                       

                                                                      Steve Kwan

                                                                      I can image you can use AF Formula DR or maybe even custom DR to execute your tuning rules.

                                                                       

                                                                      To tune the point you need to look at a fair range of the data; at least enough to capture the dynamics of the data. I won't have thought that a DR was appropriate.

                                                                       

                                                                      Steve Kwan

                                                                      In many ways, it makes sense to store PI Interface settings in AF. I'm having trouble envisioning using PI System Explorer as a way to manage PI Interfaces, i.e. PSE is not PI SMT. Someone help me make that leap.

                                                                       

                                                                      I thought that this was the plan? Currently the interface configuration is stored in MDB; so I assumed that this would ultimately move to AF possibly to the configuration database.

                                                                       

                                                                      The interface configuration information however is different to the point tuning.

                                                                       

                                                                      PSE is not SMT. However, why couldn't SMT become a plug in to PSE? I must admit that this isn't a big thing in my life. There are many other improvement I would like to see first like centralised message log, improved error/diagnostic messages on points and interfaces, .. You should also remember that not everyone that uses PSE will responsible for maintaining the underlying PI server.

                                                                       

                                                                       

                                                                        • Re: Coresight for PI - Server   "PI - Server Version Easy"

                                                                          Steve Kwan

                                                                          In many ways, it makes sense to store PI Interface settings in AF. I'm having trouble envisioning using PI System Explorer as a way to manage PI Interfaces, i.e. PSE is not PI SMT. Someone help me make that leap.

                                                                           

                                                                          You have a non-removable AF Database (the AF SDK won't allow you to remove it) of something like the "Configuration" AF Database for PI Notifications, maybe named "OSIsoft Interfaces".  Then you can leverage the WIS to control access.  As far as editing configuration options then for each interface you would have a UniInt style data reference that provides the configuration user interface for modifying the configuration options of each type and instance of an interface.  Obviously for each type of interface you have an additional data reference installed by the interface installation for the specific (i.e. non UnitInt) configuration options.  All using AF Templates and Template Hierarchies for consistency of course.

                                                                           

                                                                          On the interface nodes you have have an offline xml representation of the AF Database so the interface(s) can continue to run if the AF server is down/offline/kaputt, but you can't update the interface configuration until AF is back - much like the behaviour of interfaces today if PI MDB is unavailable.

                                                                           

                                                                          By the way, did you mean PI SMT?  I manage the interfaces via PI ICU.

                                                                           

                                                                          Michael Halhead

                                                                          I thought that this was the plan? Currently the interface configuration is stored in MDB; so I assumed that this would ultimately move to AF possibly to the configuration database.

                                                                           

                                                                          The interface configuration information however is different to the point tuning.

                                                                           

                                                                          PSE is not SMT. However, why couldn't SMT become a plug in to PSE? I must admit that this isn't a big thing in my life. There are many other improvement I would like to see first like centralised message log, improved error/diagnostic messages on points and interfaces, .. You should also remember that not everyone that uses PSE will responsible for maintaining the underlying PI server.

                                                                           

                                                                           

                                                                          Agreed, I was always under the impression interface configuration will move to AF.  It certainly makes much more sense than keeping it in PI MDB, especially as you now get PI AF as part of PI Server 2010 (and future versions).  In fact, I thought the newer AMI interfaces are already based on AF?

                                                                           

                                                                           

                                                                            • Re: Coresight for PI - Server   "PI - Server Version Easy"
                                                                              skwan

                                                                              Hi Guys,

                                                                               

                                                                              Thanks for the participation on this thread.  I am still digesting all your inputs, but before this gets out of hand, I just want to clarify an item from my previous post - with regards to interfaces, please note I was describing a separation between storing the interface configuration information vs. managing the interface.

                                                                                • Re: Coresight for PI - Server   "PI - Server Version Easy"
                                                                                  mhalhead

                                                                                  Steve,

                                                                                   

                                                                                  From a configuration perspective I would prefer this to be through the management client. IMHO there should be a single management client that handles all aspects of the PI system; rather than the current scenario where there a different clients for different aspects.

                                                                                   

                                                                                  I like Rhys thoughts on the interfaces using an XML cache on the interface nodes. The current bat files feel a bit old and using something like an XML file would provide better support for some of the "smarter" interfaces like UFL; which current uses two files the bat file and the ini file.

                                                                                   

                                                                                  Whether this is hosted within the PSE is an entirely different discussion. There would be a certain attraction to doing this; one less client tool to learn.

                                                                              • Re: Coresight for PI - Server   "PI - Server Version Easy"
                                                                                jeffjamesmcmahon

                                                                                <Selfless Plug>

                                                                                 

                                                                                Regarding a tag tuning product with a statistical foundation -- have a look at CompressionInsight - a tool my company is building that does just that - recommend exception and compression deviation settings based on a statistical analysis of snapshot data over time.

                                                                                 

                                                                                http://patterndiscovery.com/products/compressioninsight.html

                                                                                 

                                                                                </Selfless Plug>

                                                  • Re: Coresight for PI - Server   "PI - Server Version Easy"
                                                    Børre Heggernes

                                                    Wolfgang Purrer

                                                    Pos: I like to let my keyusers add tags and formulars without they risk the security and stabilty of my server.
                                                    Neg: I don't like to have to "maintenance" the tags by myself, because if i give the users  the right to create tags they can do everything on the server.

                                                     

                                                     

                                                    Seems no one has mentioned the obvious on this one (must admit to not reading all replies in detail):

                                                     

                                                    You can let you users do "tag maintainance" by configuring a role on the PI server if your server version supports WIS, i.e. .380 or higher. This way they can do certain tasks, but not "everything on the server"

                                                      • Re: Coresight for PI - Server   "PI - Server Version Easy"
                                                        Bryan Owen

                                                        To be clear, tag maintenance is an important role for a PI System.  A modicom of trust is requisite when granting permission to create or edit tag(s).

                                                         

                                                        As such for good accountability it is reasonable policy to assign privileged roles only for identities authenticated using WIS.  

                                                         

                                                        We are also seeing more adoption of change management procedures as part of the solution.  For decades, change management was a common use case for the PI-RDBMS interface.  

                                                         

                                                        There are many other technical approaches. I'd really like to hear about the latest solutions in use with the PI System.

                                                         

                                                        Similarly ideas to minimize risks associated with tag management are always appreciated, especially if a simple approach can be implemented.  Spending less time managing the system is a great objective!

                                                          • Re: Coresight for PI - Server   "PI - Server Version Easy"
                                                            wpurrer

                                                            Dear Borre and Brian!

                                                             

                                                            If someone can create tags, he can do:

                                                             

                                                            * copy data from other tags wheren't he isn't allowed to

                                                             

                                                            * create tags where he can access pointsources which he isn't allowed to

                                                             

                                                            * create output tags and write data to dcs, which he isn't allowed to!!!!!!!!!!

                                                             

                                                            I think if osisoft provides a secure way to let key users,.. manage tags it would be far more value.

                                                             

                                                            (because then he don't have to play secretary for the dcs and processeng. guys to enter PE and tags into the system).

                                                             

                                                            I'm the Lead for Plant Information Mgmt System in our company and my goal is to spend as few money, time and ressources on the server (with of course fullfill the purpose.) ... not to play communication infrastructure for tags creation.

                                                             

                                                            Its the same with folders and file on our centrale storage ... if the central storage isn't propally  managed i can put a lot of files (movies,...) on it and it would overload the system ... for this purpose we have right management, memory limits,... it works.. now everyone can create files....

                                                              • Re: Coresight for PI - Server   "PI - Server Version Easy"
                                                                wpurrer

                                                                Ps dear Brian, do you have a knowledge base article what are the minimum rights a pi interface needs?

                                                                  • Re: Coresight for PI - Server   "PI - Server Version Easy"
                                                                    Bryan Owen

                                                                    Right, to reiterate, tag maintenance is an important role for the PI System.

                                                                     

                                                                    a) Clarification regarding output points:

                                                                     

                                                                    Output points are disabled by default. Additional permissions and access to the interface node is required to enable outputs.  

                                                                     

                                                                    Once interface outputs are enabled, a very important change from a security perspective, procedural countermeasures can help manage the risk. [eg. Isolate to a specific instance and disable updates to the uniint disconnected startup cache]. In practice, the most common alternatives include dedicated servers.

                                                                     

                                                                    b) Least permission for interfaces

                                                                     

                                                                    Unfortunately, requirements are interface specific. We can pursue specific answer via technical support channel.

                                                                     

                                                                    In general, I believe it is an advantage to avoid an excess of unique DATASECURITY and PTSECURITY access control strings.  One of the most common minimum permission patterns for interface nodes (inclusive of buffering requirements) is:

                                                                     

                                                                    PIPOINT A:(R)

                                                                     

                                                                    PTSECURITY A:(R)

                                                                     

                                                                    DATASECURITY A:(R,W)

                                                                     

                                                                    This is consistent with documentation on p97 of Configuring PI Server Security.

                                                                      • Re: Coresight for PI - Server   "PI - Server Version Easy"
                                                                        wpurrer

                                                                        a) Clarification regarding output points:

                                                                         

                                                                        > is this recommendation written down somewhere in the documentation

                                                                          • Re: Coresight for PI - Server   "PI - Server Version Easy"
                                                                            Bryan Owen

                                                                            Hi Wofgang,

                                                                             

                                                                            Output point security is a good KB topic and should be integrated with mainstream product documentation.  You can also expect to see newer features emerge relative to potential abuse (intentional or not) of output point functionality.

                                                                             

                                                                            In the meantime here are a few public references:

                                                                             

                                                                            1. Support for cachesynch=0 is available starting Uniint Version 4.3.1.9

                                                                             

                                                                            -the doc does not mention why you might want to turn off updates just that the ability to do so was added

                                                                             

                                                                            2. NERC CIP Webinar (2009)

                                                                             

                                                                            -slide 18 places this feature in context of securing output points

                                                                             

                                                                            3. vCampus 2010 PI System Security

                                                                             

                                                                            -slide 38 also includes mention

                                                                             

                                                                            While I have made such recommendation in correspondence several times that really doesn't help everyone else.  Thanks again for calling out where we need better information.

                                                                              • Re: Coresight for PI - Server   "PI - Server Version Easy"
                                                                                wpurrer

                                                                                Hi Bryan .. it would be nice this not only becomes a KB - Topic, but becomes a feature to configure in the PI - Server. (Set rights on some attributes,....).

                                                                                 

                                                                                I had an issue a couple of weeks ago, the "buffer file" of the interface get corrupted after a crash.

                                                                                 

                                                                                And the interface didn't send any data anymore. (we lost a couple of data because of this issue)

                                                                                 

                                                                                Together with support we found a couple of KB - Articles and notes how to detect and solve this problem (even the techsupport was very fast => that means for me .. this incident happened very often) => but there was no single "enhancement requests" that the interface or pibuff should automatically detect on startup if the buffer is ok and if not just repair the buffer file (or rename it).

                                                                                 

                                                                                = We created on even for the PI - Server itself which has the same issue.

                                                                                 

                                                                                Less knowledge base articels more intelligence into the PI - Server.

                                                                                 

                                                                                 

                                                                        • Re: Coresight for PI - Server   "PI - Server Version Easy"
                                                                          Børre Heggernes

                                                                          Wolfgang Purrer

                                                                          If someone can create tags, he can do:

                                                                           

                                                                          * copy data from other tags wheren't he isn't allowed to

                                                                           

                                                                          * create tags where he can access pointsources which he isn't allowed to

                                                                           

                                                                          * create output tags and write data to dcs, which he isn't allowed to!!!!!!!!!!

                                                                           

                                                                          Sure he can do all of that, but that's assuming he has a user account on your domain which most likely means he is a employee. I guess it's all about how paranoid you want to be. There needs to be a modicom of trust, as Bryan so eloquently put it (yes, I had to google modicom).

                                                                           

                                                                           If a major concern of yours is dishonest/disgruntled employees then you have other issues to you must deal with. In my opinion a properly configure PI identity should prevent users doing something they shouldn't do by accident. Malicious acts are harder to deal with, especially inside threats