GENERAL QUESTIONS REGARDING CANARY

 

Please describe Canary's customer base.

19,000 installations of the Canary System in over 55 countries. No specific details available regarding vertical percentages but the Canary System is used in every major industrial automation vertical.

 

A brief history of Canary.

Canary started in 1985, providing trending solutions for industrial automation clients. It was quickly realized that a trending tool was only as good as the database behind it. In 1988 Canary began developing a proprietary database specifically for industrial automation. The Canary Historian has gone through more than 15 major release versions since that time and an entire suite of tools have been built around it to provide data logging, contextualization, asset modeling, dashboard creation, and automated reporting. Canary chooses to stay light and lean, passing on venture capital investment opportunities and refusing to obtain debt. Structured to quickly pivot to our customer's needs, Canary focuses 75% of it's resources on product development and support with a group of deployment services staff available for larger, enterprise-scale opportunities. The Canary Partner network, built of thousands of men and women who have certified on the Canary System, resell and integrate Canary software. This 'go to market' strategy allows for Canary to focus on product development rather than building out global
offices and sales staff. Instead, a core Canary team of account managers support regional distributors comprised of software and industrial engineers that in turn train and support both Canary Partners as well as international users. In North America, all support is handled by staff at Canary's corporate office in Martinsburg Pennsylvania.

 

What differentiates Canary from other time series databases?

The Canary Historian provides a highly scalable solution that doesn't use data interpolation as part of the archive process. Nearly every other data historian in the marketplace requires original data values be interpolated, averaged, or dropped for long term archiving. Canary has developed a proprietary algorithm that provides best-in-class data compression without every changing the originally logged values. Scalability is possible due to the performance of the Canary System. Writing data into the historian or reading data out of the historian has minimum speeds of 1.5 million updates per second and does not degrade over time. Since a single server can handle more than 2 million tags, enterprises can scale quickly without unnecessary overhead. Additionally, the reporting and trending tool Axiom, provides a simple and clean interface that gives your users the ability to create their own automated reports, monitor assets by condition, and the best trending tool available in the marketplace.

 

What OS does Canary work with?

The Canary System is developed on .NET and should be installed in a Windows environment.  Currently, data collection (Data Collectors and Sender Service) also require a Windows machine but are being developed for Linux in a 2023 release.

 

Can Canary be Cloud Hosted?

Yes.  Canary can be installed on a virtual machine or a physical server.  The multi-tenant nature of the database and store and forward technology built into the data collection mechanisms make Canary a perfect solution for cloud environments.

 

DATA COLLECTION

 

What protocols can Canary log data from?

For data ingestion into the Canary Historian the Canary System provides integration with OPC UA or OPC DA servers and MQTT Sparkplug B brokers. Additionally, Canary can ingest CSV files, read other MSSQL or MYSQL databases historically or in real time, and has both .NET and WEB APIs available for additional connectivity. Releasing in Jan/Feb 2021 are additional connectors that will provide for direct connections to Allen-Bradley PLCs, Siemens PLCs, and PLCs via modbus. 

 

How does Canary handle data logging with network connectivity concerns such as uptime, bandwidth, etc.?

Canary is designed around 'real world' struggles and can support poor bandwidth or network performance. Data is logged using store and forward technologies, providing local buffering as needed within the Sender Service.  When the Receiver looses connection to the Sender Service, all tags that are part of that session are immediately updated with a 'null' value and a 'NoData' quality score.  This informs all client tools that the data points are no longer connected to the system.  Upon reconnection to the Sender Service, all buffered data is brought into the historian and only the original 'NoData' quality score, signifying the time of connectivity loss remains.

 

Can you log to redundant historians or to both local and enterprise historians at the same time?

Yes.  From the logging configuration, you may organize tags into logging sessions (like folder structure). Each session can be pointed towards multiple Canary Historians.  This includes local Level 3 historians, Level 3.5 proxy servers, Level 4 corporate historians, as well as Level 3 and Level 4 primary and redundant historian servers.  The only thing that needs to be modified to change a logging session from a single historian to a multiple historian is the addition of the secondary historian's name or IP address within the logging configuration. For example the logging field for historian that once read "LocalSiteHistorian" would instead be configured to read "LocalSiteHistorian, "CorporateHistorian".  

 

DATA ARCHIVING & STORAGE

 

What is Canary's underlying database technology?

The Canary Historian is a file based storage system that utilizes Canary's proprietary NOSQL database technology.  Build around organizing tag groups into data sets, Canary creates new daily archive files for each data set within the historian.  These historical files can easily be moved, backed up, and transferred to other Canary Historians as needed.

 

What are the database's performance benchmarks?

Write and read speeds have been benchmarked at 2.8 million TVQs per second for writing and 4.6 million reads per second on 20 core servers with 2.1 GHz processing power and 32 GB of memory.  Lowering resources will result in lower speeds but with minimal resources, 8 cores and 16 GB of memory, write speeds above 1 million updates per second and read speeds over 2 million updates per second should be achieved.

 

How many tags can a single historian hold?

Tag storage hard limits are not set within Canary and some production servers have more than two million tags actively writing data.  

 

What type of data speeds can Canary process?

With the proper hardware and resources, the Canary Historian can capture continuous data feeds as fast as 10 milliseconds. 

 

How is the data stored and processed?

No interpolation is used in compressing historical data files.  Canary has developed a custom loss-less compression algorithm that maintains all data with its original granularity and frequency, yet still achieves an industry best 66% compression.

Canary prioritizes loss-less data archiving to ensure the data record is trust-worthy and properly preserved for future needs.  For instance, machine learning is dependent on large volumes of data, many points which would be erased or lost if interpolated.

When requesting data from the Canary system, the client can choose rather to request raw data or interpolated/aggregated data based on their specific data needs.  This approach provides the best of both worlds, complete data archives with 'data cleaning' options based on the client's needs.

 

Does the database need managed?

While someone will certainly need to take on 'management' duties of the Canary System, the database does not need a database administrator in the traditional sense.  Canary archive files are automatically validated and compressed by the Canary Historian.  This process ensures that each historical file is verified and complete.  Should any errors occur within the file during validation, the Canary Historian will automatically take the file offline and notify the system administrators.  System admins will then use included diagnostic and repair tools to correct the historical file and re-instate its availability.  

 

What type of storage requirements are necessary?

Data storage has less to do with tag count and more to do with change rate, data type, scan rate, etc. Canary writes data based on exception, meaning that even if a tag is polled/scanned every second, if it does not have a value change or pass the logging configuration deadband, the historian does not write a new value, but instead just updates the time stamp.

Since Canary is typically installed on a VM, adding disk space availability/resources is done as necessary. Based on our experience, the following estimations for system provisioning can be used:

100,000 tags (30% boolean, 20% analog integer, 50% analog float) with an average sample rate of 5 seconds and an average change rate at 30% stored for 7 years would use 6.64 TB of disk space.

Compare this to the same 100,000 tags with an average sample rate of 3 seconds and a faster 35% change rate and that same 7 year period would use 11.30 TB of disk space.

Note, data read performance does not change, even when the database grows by size and time.

 

DATA CONTEXT

Can you add metadata or properties to tags?

Yes, metadata can be added to tags from the logging session or from additional databases.

 

Does Canary support tag aliasing or asset modeling?

Yes.  Canary uses the Views Service to create tag aliases and asset modeling. This means Canary does not require the data to be archived in the historian in a specific format to work with asset modeling.  This keeps an organization from having to standardize their asset model prior to logging data. Instead, Virtual Views are positioned between the client requesting data and the Canary Historian.  A further benefit is that multiple Virtual Views can be created and access to them is permissions based, allowing system administrators to create custom asset models and tag aliases for individual or groups of clients.

 

How do you develop an asset model?

Asset structures can be formed in several ways. Regular Expressions can be used to match tag naming logic or metadata properties to both alias tag names, add/remove tag naming structure, or remove tags from the asset model. Additionally, metadata properties that can be used for asset hierarchy and grouping can be added to tags by reading existing SQL databases in real time. These databases can link specific tag names to asset types, as well as provide parent/child relationships of those asset types.

Since this process is dynamic, if new tags arrive in the Canary Historian, the above mentioned rules are automatically applied to the new tags and the models are updated. This provides for the 'auto discovery' of new tags and population of the asset model within 5 minutes of their arrival.

 

What are some features of Canary's asset model hierarchy?

Tags can be grouped into asset types, complete with multiple levels of parent/child hierarchy. Tags may live in multiple asset models, as well as multiple asset types. For instance tags may be grouped and organized as a 'Crusher' asset type, as well as live in a more specific asset type such as 'Primary Crusher', 'Secondary Crusher', or 'Tertiary Crusher'. This allows the user to monitor condition of all crushers, as well as specific crusher types, without having to create multiple models. Additionally, you may reference other previously created Virtual Views and Asset Models when building new views and new asset models. 

 

Can new tags be calculated from other historian tags?

Yes, the Calculation Server allows you to define calculations based on tags or asset types.  If defined on asset type, the new calculation can then be applied to all asset type instances. The benefit is that you develop a single calculation and then deploy it to all current, and future, instances of the asset type. These calculations then become part of the asset model and can be further used for additional calculation development.

Canary Calculation Server can be used for a variety of calculations, including OEE, asset condition monitoring, asset health assessments, and alarming.

 

Can event monitoring be built into the system?

Yes. A trigger tag or trigger calculation would need to be defined that determines whether the asset is in a true or false state for the desired process. These rules can be based on asset type and then deployed to all asset type instances. When the rule or rules are violated, the event tracking begins. When the event is over, predefined analytics are calculated around the event duration. The tags that are analyzed do not have to be part of the event definition, nor do they have to be included in the asset.

 

What type of notification is available for events?

Events can be configured to notify individuals or groups via an email.  This feature also is capable of linking to an SMS gateway for text notifications.

 

Do Canary Events provide for acknowledgement or escalation?

No, Canary is not designed as an alarm management tool.  It is recommended to use either SCADA solutions or a dedicated alarm management software solution for these features.

 

Can other actions be triggered from events?

Yes.  Axiom applications and trend charts can be delivered to a set client list via email based on event triggers.  This feature allows for users to be notified not just of events that take place, but also to have important data delivered to their email in PNG format so they can make better, and faster, operational decisions.

 

Can Canary Event data be shared with other third-party applications for automated workflows and ticket creation?

Yes.  Canary writes all Events into an SQL database which can be queried by other applications.  Additionally, Canary also exposes the Event database via API.

 

DATA VISUALIZATION

 

Has Canary created their own visualization tool or is it a rebranded tool from a third party?

Axiom, Canary's trending, dashboarding, and reporting package, was built by the Canary team specifically for the Canary System.  Axiom was originally released in 2014 and is continually developed and improved. Canary's previous trending tool, Trend Link, was an industry standard from 1996 until 2018.  Axiom has taken the proven features from Trend Link, modernized them into an HTML application, and added even more functionality enabling HMI dashboarding and automated reporting.

 

How are dashboards and reports creation handled?

Axiom is a complete blank slate. Users open the built-in HTML editor and drag and drop widgets onto the dashboard. They can click a widget and edit it's properties, including linking it to a tag in the Canary Historian, setting limits, adding data transformation expressions, etc. The application can then be saved in one of three folders on the server, 'Public', 'Private', or 'Read Only' (admins can only save here). No design experience is required, average screen development for a basic report or dashboard is just a few  minutes.

 

Can Axiom work on mobile or touchscreen devices?

Yes, Axiom is fully developed for touchscreen interaction and can scale from hand-held devices to large television monitors.  Additionally, the design editor also works for touchscreen deployments.

 

How many trends can be shown on a chart?

No hard limits are set to restrict the number of trends visualized.  Some clients have added more than 100 trends to a single chart.

 

Can trends be shown on the same scales or modified to show correlation with other trends?

Yes.  The user has full control on how to layout the trend chart.  This includes the ability to move and resize trends, linking them to other trends and overlapping them as needed.  This action is very fast and simple to do.  Additionally, trends can be resized to prioritize information.

 

Can trends be shown with aggregated/processed data?

Axiom trends and controls support more than 40 aggregates helping you transform raw process data into usable information.  Aggregates include average, minimum, maximum, delta, range, interpolation, duration on/off, count, total, standard deviation, variance, and several dozen more options.  Each tag can be carved into time buckets and then have an aggregate of the users choice applied to each time bucket of raw data values. 

 

Can high or low limits be visualized on trends?

Yes, both high and low limits can be visualized on each trend band.  High and low limits can be static or dynamic.  Dynamic limits are achieved by allowing the user to link the limits to another tag value.  When a tag passes the defined limit, the trend color can be modified as well as a fill applied to the trend in the area that is outside of the limit.  Additionally, multiple limit lines can be placed on the trend chart.  These limit lines can be color coded and drawn with different styles.

 

Can trends be calculated adhoc?

Yes.  Axiom provides a calculation toolset that allows users to create complex calculations with any trends that are linked to the chart.  A variety of functions and expressions are available, including logic statements.

 

Can trends and reports be saved?

Anything created in Axiom can be saved several server-side in Public, Private, and Read Only folders. Reports saved by administrators in 'Read Only' allows all users to open, and edit the report, but they cannot save overtop.  Instead, they may save the report in their own private folders.  All users have access to open and save reports in the 'Public' folder and each user has their own 'Private' folder as well.  Within folders, users can create additional subfolders and hierarchy.  Applications and charts can be copied and pasted into multiple folders as needed.

 

What types of dashboards can be built in Axiom?

Axiom dashboards can contain hundreds of ways to visualize your data.  You can create data tables, use trend charts or spark charts to show historical data, link values to graphical symbols like tanks, blowers, motors, valves, etc, or use gauges to illustrate values.  In all of these visualization tools, data can be shown in real time or in historical aggregate format.  Additional design widgets include text boxes, drop down lists, iframes, buttons, labels, images, and more.

 

Is Axiom scalable for enterprise level applications?

The 'Asset Template' feature allows you to design for a single asset type instance and then apply that design for all discovered instances. Simply put, design one time, see all assets that apply. This also offers a filter for viewing asset instances which allows you to only see, and sort, the assets whose values or condition fall outside/inside defined parameters.

As new assets come online, these reports automatically include them within minutes, requiring no on-going work or system administration. Additionally, these asset based reporting tools allow for multiple levels of 'drill down' incorporating parent/child hierarchies and allowing for 'asset pass thru' which gives you the ability to monitor assets
on a wide scale but still drill down to specific asset detail without having to design individual screens for each asset.

 

DATA CONNECTIVITY

 

How can data be connected to Microsoft Excel?

Canary provides an Excel Add-in that allows Excel users to browse tag data directly from the Canary Historian.  The Excel Add-in can pull data in raw format directly from the historical archive as well as from any Virtual View or Asset Model.  Beyond raw data, clients can request aggregated or processed data as well.  Additionally, Excel users can pull Event history from the Canary Event Service as well as build ad hoc Event data tables.

 

Can SQL queries be made against the Canary Historian?

Yes, via the ODBC Connector, clients can make 'SQL-like' queries against both the Canary Historian archive and Virtual Views with Asset Models.

 

What other methodologies exist to provide real-time data from the historian to third-party clients?

The Publisher Service is included with each Canary System.  Publisher can be configured to send value updates on change to third party clients via either MQTT Sparkplug or JSON over WebSocket.  System administrators may configure whether to send updates in batch format, or in real-time as the records are written to the historian.

 

What industry standards does Canary follow for providing historical data to third-party clients?

An OPC HDA Server is included with each Canary System which provides clients the ability to read the Canary Historian archive using OPC HDA client tools.  Additionally, Canary Systems includes a .NET and Web API for querying historical data in raw or aggregated format from the Canary Historian, as well as Events history.

 

LICENSING & BUSINESS MODEL

 

How does Canary license the product?

The Canary System includes all the software you need to collect, store, and analyze your organization's process data and is licensed as a complete solution.

 

What software solutions are included?

Each system contains data collectors, store and forward, a 100 tag Canary Historian, Virtual Views, the Calculation Server, Events Monitoring, a single Axiom license, a single Excel Add-in license, and an SDK for data connectivity including .NET/Web APIs, an OPC HDA Server, and a Publisher Service for JSON and MQTT data streams.

 

What is additional?

You can choose how many additional tags and client licenses you need.  Both are offered in a 'non licensed, unlimited' format if needed.  Finally, the ODBC Connector is licensed additionally on a 'per-server' basis.

 

Does Canary offer hosting services?

Yes.  Canary has several Cloud servers available that clients can log data to and read data from using Axiom, the Excel Add-in, and other third party applications.

 

For multiple Canary Server instances, are Enterprise License Agreements available?

Yes, Canary provides Enterprise Licensing Agreements to reduce the per server license costs for larger organizations.  These are recommended for organizations that will need 10 or more Canary Servers.

 

Does Canary provide any hardware for logging or storing data?

No, Canary only creates and provides software.  Hardware is not included.