Repost from LinkedIn "PI Professionals" Group - "PI on Linux" Discussion

Blog Post created by richard Employee on Dec 18, 2014

The following is a repost of comments I made in a discussion "PI on Linux" in the LinkedIn Group "PI Professionals"


OSIsoft has evolved the PI System and greater Infrastructure across technologies and platforms since our earliest days of “PI” 30 years ago. From the humble beginning on HP to DEC/VMS to many flavors of UNIX and Windows NT/Server. All along we’ve made a commitment to paving a path that brings our customers, data and systems along – rarely leaving anyone behind.

Along these lines, we continue to explore emerging technologies and platforms. There are three areas in which we are making significant advances and evolving the PI System. First, we are embracing distributed computing (grid computing/high density computing) as a means to provide significantly greater performance and scale as well and opportunities for availability and durability. Second, we are embracing platform neutral technologies across the whole surface of the PI System. And lastly we are making a big push in our “integration” technologies. Allow me to expand on these.


In our research and engineering department, we are well underway in the process of developing our next generation technologies for distributed/grid computing (such as a distributed historian and historian as a service.) The results are very encouraging. (Unfortunately I am not at liberty to provide any more detail at this point, in other words don’t ask about features or dates.) I can say that this new server/service stack is committed to the Windows Server/Azure platform. Future deployments of the PI System whether on premises or in the “Cloud”, self-operated or as a service (SaaS) will take advantage of the features and strengths of the Microsoft Private/Public data center architectures. We are also deeply committed to both a solid evolution/migration strategy from our current offerings to our next gen as well as delivering a “Hybrid Infrastructure” model, allowing the PI System Infrastructure to span on premises and cloud as a logic whole. (We are also exploring the possibility of evolving the PI (3) System back to its roots in “Unix” – specifically Linux for stand-alone (machine oriented) and embedded environments – please contact me outside of this forum, email below, if you have specific use cases that would benefit and/or demand this.)


The surface of our system consists primarily of data ingress (interfaces/connectors), data egress (developer platform) and visualization (clients). Beginning in 2014 we are focusing our “interface” development on “PI Connectors” and our new data acquisition framework. This technology is primarily built on manage code (.NET) technology and has been built and tested on both Microsoft .NET and Mono. The open sourcing of .NET suggests that this was a good bet and should help us extend our PI Connector footprint across more devices and platforms. We are also researching a much lighter weight “contract” based ingress end point which will allow data acquisition from wearable devices and very light weight “smart” sensors. (Hopefully this will be available for experimentation at the Hackathon in SF 2015.) In both cases are goal is to reach any platform or device where meaningful “operations” data is available. 

For data egress we are focusing our data platform surface area again on platform and device neutral technologies. With the PI Web API release we establish the new “primary” development surface for the PI System. The PI Web API brings an HTTP(S)/REST based programming surface (including OData as a CTP/Beta) and support for claims based security. For relational access to the system, we continue to invest in ODBC/JDBC and significant improvements are in the pipeline for our SQL surface. While the managed code based, AF SDK, still continues to provide a solid/rich .NET based developer surface and is preferable is certain cases, our commitment to platform neutral technologies is critical for both our own efforts and clearly those of our entire ecosystem of customers, partners and developers. (Note that there are corresponding changes in developer platform licensing that together with our evolving technologies should create much less friction for developers and solution providers.)


Our visualization/client surface is being re-envisioned as Web Applications, Web Sites and Web Services designed for platform neutral use and consumption across a broad range of devices. This represents a complete embrace of HTML and Web technologies (and yes, rewriting the Silverlight/.NET and other Binary dependent code base.)

Integration is another significant area of development which focuses on integration with both computational and visual analytics as well as machine learning systems. The primary challenges with successful integration are effective extraction of “decision ready data” from the PI System and efficient transfer of potential huge amounts of data to target integration systems like Azure ML, Hadoop/HDInsight, SQL, HANA, etc.


As others have noted, while we are investing significantly in many platform neutral and open source technologies, we are strongly committed to the core Microsoft platforms and technologies. Our partnership with Microsoft creates significant value for our customers allowing OSIsoft to stay focused on building the best PI System possible and enabling the entire ecosystem from customers, partners, developers to shine.


-Richard Beeson CTO/CSA,