2. The Servers
Frank Batke
Hans Herbert Gimmler

3. concern and content Wrap up an Overview
Gathers, Organizes, Distributes, and Stores Data
Time series data – PI Archive
Data Access
Real Time Interfaces
Non Time series data – AF
MCN Health Monitor

4. Functional Groups of The PI SYSTEM
Visuals
Deliver Information for Meaningful, Better Business Decisions
Performance Improvement
Operational Visibility
Knowledge Management
Product Quality
Lean Manufacturing
Asset Management
Situational Awareness
Analytics
Turn Valuable Real-time Data Into Actionable Information
The
PI SYSTEM
Gathers, Organizes, Distributes, and Stores Data from Many Sources
Server
The
Structure / Asset Data
Real-time Data
Custom Data
IT Data
Relational Data
Web Services
ERP / Maintenance 4 4 4 4

5. The Analytics The Visuals The Server
Smart Clients
RtReports
PI Analytics
ACE
Sigmafine
PI Notifications
PI Activeview
PI DataLink
PI ProcessBook
PI BatchView
RtWebParts
RtReports Clients
RtPortal iViews
Thin Clients
Real-Time Interfaces
PI AF
PI Archive
MCN Health Monitor
The Server
Data Access 5 5 5

6. The Server PI Archive PI AF Real-Time Interfaces Data Access
MCN Health Monitor
Data Access
Piarchive 6 6 6

7. Event Storage – core today
time
High-Speed Data (IT)
medium Data (Process)
Slow Data (Lab)
Archive and
Annotation Files
Event Queue
Snapshot Table
Point Database
Tag
Where ?
Timestamp
When ?
Value
What ?
Questionable
Quality ?
Substituted
Changed ?
Annotation
Add. Bit stream
Event :=

8. Development PI Server with Partitioning
Slow
High-Speed
time
And further changes in the architecture (shared-memory snapshot table) allow higher throughput on read operation or point modifications.
Point Database with Snapshot Table, in Shared Memory

9. Performance Startup Time Archiving Rate Snapshot Rate
Archive Query Rate

10. PI Batch Database: ANSI/ISA-88
Batch S88 Hierarchy
Campaign
Batch1
Unit Batch
+Unit Batch
Operation
+Operation
Phase
+Phase
+State
State

11. Batch and Event Frames – Examples
Continuous Data
Batch
Day or Shift
Event Frame (in development to unify)
2007Aug06-ABC123
Start
End
Fixed Start / End

12. Batch & Event Frames – Features
Extensible schema
– for S88 (ANSI/ISA-88) and beyond (e.g. IEC )
Increased database performance
Tighter integration with all products
Model more types of processes and meet data storage demands
Customer defined vocabulary
Segregate Event Frame types
Histories Event Frames from external systems
Client side Event Frames
– Create locally
– Commit Event Frame or the rule to server

13. PI Archive All in one - Administration by PI-SMT
Example PI System Management Tool:
PI Server Collective
Large PI Server on 64 bit Platform
Batch Database Editor
System with
2,010,128
Points!
Interface Configuration
IT Monitor Configuration
Authentication
Users and groups (security principals) managed in Active Directory (or Local Windows) for initial release; subsequent release may provide option for other directory services
Windows performs authentication when user logs on to PI Server, produces user token
Windows user token goes to PI Server
PI Server uses SIDs in user token to map to PI Identities
Authorization
User requests operation on secure object in PI Server
Server looks up access control list (ACL) for secure object that user wishes to access
Server compares PI Identities on ACL with PI Identities of authenticated user; matches determine which access rights the user can have
Server determines whether user is allowed to perform operation based on resulting access rights
One management tool (with plug-ins) for all PI SYSTEMS in your Enterprise that runs from your desktop.
PI Server Configuration
PI Security Configuration 13

14. The Server PI Archive PI AF Real-Time Interfaces Data Access
MCN Health Monitor
Data Access
Real-Time Interfaces
Interfaces 14 14 14

15. Interfaces COM Connectors
Advanced Real time Interfaces
Secondary everybody can see if Master is down
Can write to multiple Servers and Collectives
Ease of Deployment and Maintenance
Remote Configuration and Monitoring
Auto Point Synchronization (APS) (between DCS / PLC / SCADA)
Disconnected Startup (Node is able to restart without connection to the PI Server)
Buffering and History Recovery (no Data loss)
Exception Reporting (unload the bus)
Automatic Failover, High Availability (HA)
Data Security
Standardized Logging and Debugging
PI Server Master Configuration Data
PI Server
INTERFACE
DCS
PLC
SCADA
OTHER
OPC Client
XML
Modbus Device
OLE DB Consumer
OPC SERVER
RtPortal Services
DDE Server
ODBC
Client
SQL Server
Oracle
.NET App.
ODBC DSN
ADO RECORD SETS
OPC HDA SERVER
OLEDB
WEB SERVICE IIS
RDBMS IF
MODBUS IF
OPC IF
OTHER
NET/DDE IF
SQL
Server PI
MANAGED
SDK
API
PI SYSTEM
ADO .NET
ADO
Interfaces COM Connectors
>370
APS
DCS
PLC
Other
ASCII
OLAP

16. New value (scan or exception based)
Collection & Storing
Data Compression
Swinging Door
Exception Reports
DCS
PLC
SCADA
OTHER
PI Server + - o x %
INTERFACE
Interface (on Data
Collection Node)
PI Server
Exception
Compression
Snapshot
PI Compression
Algorithm
PI Archives
New value (scan or exception based)
Exception Report

17. Compression Results Raw values scanned After Exception
After Compression
Temperature
Time

18. COM Connectors
PI COM Connectors allow other enterprise systems to use PI architecture and clients, delivering data between the PI Server and foreign databases or data historians without storing them in PI Data Archive
Each COM Connector obtains foreign system data using techniques provided by the foreign system vendor.
A COM Connector can be implemented as either an in-process or an out-of-process COM object. PI
NetMgr
Snapshot
Request
Response
RPC Call
RPC Response
COM
Connector
COM call
Return Data
Foreign
Data
System
Redirector
An example of RPC flow through the PI Snapshot subsystem
Foreign Data Systems
PI Server
Connector
COM 18

19. High Availability – concern and goal
Software Fault-Tolerant System
Interface Failover
Buffering
PI Server Replication
SDK Services (discovery, failover, and load distribution)
N-way Buffering of Non-Interface Data (e.g. PI-SDK)
Replication of Archive Edits among Server Nodes
Promotion of Secondary Nodes on Primary Failure (configurable)
Near-Independent, Physically Separated Servers
No hardware/network restrictions, no limit on Server nodes
General Benefits
Availability, end-USER sees one logical system
Scalability, system load can be distributed
Flexibility, accommodates your environment
For IT and Management
Reduced Total Cost of Ownership (TCO)
Allows Disaster Recovery Plans
Extra benefit: Hardware and Software just out of the box

20. PI HA Architecture - reality
System Management Tools
Thin Clients: RtPortal, RtWebParts,
Rich Clients: ProcessBook, DataLink, Custom Application…
PI-SDK
Secondary
PI Server
Primary
PI Server
Secondary
PI Server(s)
Configuration
Changes
PI Server Collective
Define – Collective, Primary, Secondary
Server
Discuss configuration flow
Interface
Buffering fanning data
Client
Applications conversation with collective is mediated by SDK to provide automatic failover capabilities.
Identical Time-Series
Data Buffering Services
PI Interfaces
Data Buffering Services
PI Interfaces
Failover Mechanism

21. Replication & Data Mirroring - vision
SDK Client
Batch Interface
Data Collection & Buffering
Displays, Notifications,
Analysis,
etc.
Failover
Real-time Data Mirroring
PI-to-PI
Interface
Secondary
Secondary
Step 1: 3-node Collective with PI Interface buffering (no manual data, no batch)
Step 2: Manual data coming to the Primary server, PI-to-PI from the primary to secondary servers
Step 3 (future): Server-Side Buffering, removes the need for PI-to-PI, brings batch replication and allows manual data to secondary servers
SDK Client
Metadata
Replication
Metadata
Replication
PI Collective with
Server Side Buffering
PI Collective
(PR1)
Primary 21

22. Enterprise High Availability - vision
HA for an enterprise deployment of both PI and AF

23. PI Server, Windows Security Integration 3.4.380.x
Goals:
More flexible access control
More secure authentication methods
Leverage Windows for account administration
Single sign-on (no explicit PI Server login required)
New Security Model
Support for Active Directory and Windows Local Users/Groups
Mapping of authenticated Windows principals to “PI Identities”
Extensible Access Control Lists for points, etc.
This slide and the next goes before the WIS diagram

24. “PIIdentities” 3.4.380.x Purpose What are PI Identities? Why?
Link Windows principals with PI Server objects
What are PI Identities?
A representation of an individual user, a group, or a combination of users and groups
All PIUsers and PIGroups become PIIdentities
Why?
To maximize flexibility and performance for controlling user access to secure objects within the PI Server

25. WIS: Simplified Diagram: see 3.4.380.x
Windows
PI Server
Active
Directory
Security
Principals
Authentication
Identity Mapping
PI Identities
Access Control Lists
Authorization PI
Secure
Objects
Authentication
Users and groups (security principals) managed in Active Directory (or Local Windows) for initial release; subsequent release may provide option for other directory services
Windows performs authentication when user logs on to PI Server, produces user token
Windows user token goes to PI Server
PI Server uses SIDs in user token to map to PI Identities
Authorization
User requests operation on secure object in PI Server
Server looks up access control list (ACL) for secure object that user wishes to access
Server compares PI Identities on ACL with PI Identities of authenticated user; matches determine which access rights the user can have
Server determines whether user is allowed to perform operation based on resulting access rights

26. Security Windows Integrated Security Security between IT Network
& Process Control Network
Secure Real-time Communication
Windows
PI Server
Authentication
Identity Mapping PI
Secure
Objects
PI Identities
Active
Directory
Industrial Data Center
Authorization
Security
Principals
Access Control Lists
Single Sign ON
Application has its own logical Connection
handeled through common PI Platform
using Secure Connection
- Monitored

27. The Server PI Archive PI AF Real-Time Interfaces Data Access
MCN Health Monitor
Data Access
Data Access 27 27 27

28. RtBaseline Services for Thin Clients
Common Standards
Data Access
RtBaseline Services for Thin Clients
OLEDB
OPC / HDA
ODBC
WebServices (i.e. RtBaseline Services) – Provide Data to Web Applications
OLEDB – Data Access via SQL Queries
ODBC – Data Access via SQL Queries
OPC DA/HDA – Data Access to the PI SYSTEM via OPC Standard

29. WebServices RtBaseline RtPortal PI Server RtWebParts RtReports
Thin Client
RtReports Generator
RtBaseline
Services*
Internet Explorer
PI Server
RtWebParts
RtReports
RtPortal iViews
Thin Client
RtWebParts
Transactional, Web service, and other data

30. PI OLEDB and PI ODBC SQL Queries SQL Queries
PI OLEDB provider allows applications (OLE DB consumers) working with PI data and configuration through SQL queries:
OLE DB Consumer
PI OLEDB Provider
SQL Engine
Client Application
SQL Queries
PI Server
OLE DB/ COM
PI ODBC driver allows applications working with PI data through SQL queries
SQL Queries
ODBC Consumer
PI ODBC Driver
SQL Engine
Client Application
PI Server
ODBC 30

31. OPC DA / HDA
PI OPC1 DA/HDA2 Server is an OPC server for OSI Software's PI data archives.
This server interoperates with any OPC client that is compliant with OPC standards.
This server can be installed on the same system as a PI archive, or it can be installed on a separate system.
It can serve data from multiple PI SYSTEMS simultaneously.
Note 1: OPC is a synonym for OLE (=Object Linking and Embedding) for Process Control. wikipedia: “(OPC) Foundation is an industry consortium that creates and maintains standards for open connectivity of industrial automation devices and systems.” More details on
Note 2: Data Access / Historical Data Access

32. PI & OPC UA OPC Unified Architecture (UA)
Access to AF Asset Models, PI Notifications, Event Frames
Data and information exchange between OPC applications in a secure environment
Enhanced robustness and reliability
Support for other data standards (i.e. ANSI/ISA-88 = S88 for Batch Control, S95 for Enterprise-Control System Integration, MIMOSA1, EDDL2, PRODML3, B2MML4, etc.)
Support for multiplatform interoperability (Windows, Linux, UNIX, Embedded OS, etc.)
Note 1: MIMOSA is the synonym for Machinery Information Management Open Systems Alliance (details on
Note 2: EDDL is using for Electronic Device Description Language which is defined in IEC 61804
Note 3: PRODML (Production Markup Language) is an XML-based standard from Energistics (
Note 4: B2MML means Business To Manufacturing Markup Language is an XML implementation of the ANSI/ISA-95 and IEC/ISO 62264 32

33. Mapping into OPC UA OSIsoft Technology OPC UA Technology AF Database
Address Space
AF Elements, Ref.
Nodes, Ref.
Real-time data serv.
Data Access
Historical data serv.
Historical Access
PI Notifications
Alarms & Conditions
HA, Failover
Redundancy, Robust.
PI Security
Security

34. The Server PI Archive PI AF Real-Time Interfaces Data Access
MCN Health Monitor
Data Access
PI AF AF 34 34 34

35. PI AF 2.0 PI Server 1 PI Server 2 PI AF 2.0 in the PI System
Data structured and organized by asset
Spans multiple PI Systems
Incorporates non time series data
Honeywell
ABB
Delta-V
Rockwell
Other …
Time Series Data Sources
OPC
Interface
XML
Modbus
SNMP
Generic Interfaces
Vendor Specific Interfaces
PI Server 1
Honeywell
ABB …
Time Series Data Sources
OPC
Interface
XML
Modbus
SNMP
Generic Interfaces
Delta-V
Rockwell
Vendor Specific Interfaces
PI Server 2
Other
Non Time Series Data Sources …
SCM
RDBMS
MES
PI AF 2.0
Asset-Equipment Centric access to the Data

36. Evolution Asset Element Module Unification “foreign” structures Tag
Connectivity
PI – Archive
Tag
Processbook Datalink Batch Performance Equations
Control Monitor
History
Asset
“foreign” structures
Unification
Notifications
Analysis Framework
2.x
Aliasing
Versions Hierarchies
Module database
Module
ACE Module relative Displays
Context
Templating
Flow sheeting
“foreign” data
Model analyses
Analysis Framework
1.x
Connections
Element

37. AF – Developed Once – Deployed Everywhere
Collections of reusable Elements
Connectivity
Hierarchy
Role based
Assets
Other Data
Process Data / Events
Template
ELEMENT
TIMESERIES
DATA SOURCES
TAG
Datareference
ELEMENT
ATTRIBUTE WITH UOM i.e. FlowRate l/h
ATTRIBUTE WITH UOM
ATTRIBUTE WITH UOM i.e. LastCalibration
UOM Units of Measurement
NON TIMESERIES DATA SOURCES
Datareference

38. AF – Putting AF into Best Practice
Shaping your data by:
Condensor
Heatexchanger
Column
Valve
Pipe
Pump
1. Defining types of assets Templates
Schema how to attribute Elements
2. Association to a “real” asset Elements
Created from Template
3. Describing the “real” asset Attributes
having Units Of Measurements (UOM)
can come via data references from everywhere
4. Physical/logical asset structure Hierarchy
5. Assets connectivity Models
Model : Collections of connected elements
Column661
Condensor661
P661_1
P661_2
HeatExchanger661
Valve661_1
Valve661_2
OpeningGrade PI Point: \\MOBILEVBC\Valve661_1.OpeningGrade
InspectionResult Table Lookup: SELECT InspectionResult FROM …
LastInspection Table Lookup: SELECT LastInspection FROM …
SerialNumber Table Lookup: SELECT SerialNumber FROM …
XZY Formula: A=OpeningGrade;[A*0.98]

39. The Server PI Archive PI AF Real-Time Interfaces Data Access
Health Monitor
MCN Health Monitor 39 39 39

40. Process Control Servers
Overview
What do you know about the Availability – Performance – Security?
Devices?
Applications?
Network?
Distribution Switch
Distribution Switch
Core Switch
Core Switch
Access Switch
Access Switch
Firewall
Firewall
Router
Router
Administration
MCN Switch
MCN Switch
WAN
Internet
Remote Locations
Router
Router
Lab
Switch
Switch
Process Control Servers PI

41. MCN HealthMonitor
Proactive Communication Infrastructure Monitoring and Informed Decision-Making
Management Console
IT Organizer
IT Overview
Monitoring Interfaces
Perfmon
SNMP
PING
TCP response

42. Dont mix server and services
Services to run in a PI server environment
Advanced Computing Engine to combine .NET programming power with PI event power with PI event power on PI reusable structures
PI notification services to combine Microsoft messaging mechanisms with PI event power on PI reusable structures
Rtbaseline services to combine portal visualization technology with PI event power on PI reusable structures
Sigmafine analysis Methods to combine mathematical balance methods with PI event power on PI reusable structures 42

43. everything is changing over time
Designed for time series and non time series data:
High Performance (storage AND retrieval) e.g. Timestamp Resolution ~ 15 µs; Sustained Read from Archive ~ values/sec and write values/sec depends on hardware + multiple users
Scalability (no practical limits in growth) System Size to Data Streams (32 bit) # to Data Streams (64 bit); Number of Users limited only by hardware
High Availability (backbone for business critical applications) High Availability for Server collectives and Client failover, actual Part of Platform Release 1 in development load balancing, peer-to-peer replication
Security (access to all data with no risk for operations) Configuration & Data Security by Tag, Element and User
Extensibility (be prepared for unplanned integration) Supported Standards like OPC DA, OPC HDA, OPC A&E, OPC XML-DA, ODBC, OLE DB, COM, .NET, SNMP, Perf.Counter
Crossing boundaries (technical, organizational, regional,... )
Protection of investment: Integration of legacy systems e.g. Data from 20 year old DCS shown in Excel Services
Unified data access to all operations

44. Dank U wel!