Today is...
Wednesday, April 25, 2018
Welcome to, the global online
community of automation professionals.
Featured Video...
Featured Video
EtherCAT with CTC’s master lets your multivendor network play well together...
Our Advertisers
Help keep our servers running...
Patronize our advertisers!
Visit our Post Archive
OSI Pi versus Aspen Info Plus 21
Chemical company with batch processes has to decide if they buy Aspen tech or OSI.

I am working for a company producing material in batches (Chemistry) we have to decide if we buy the OSI Soft PI Solution or the Aspentech Info Plus 21 Solution. (we also checked other providers)

In the financial perspective we saw, that for OSI we have to pay for the license in the first year approx. 200000 Euro for our site. For each following year approx. 40000 for the maintenance and support from OSI. On the other hand with Aspen we pay only approx. 40000 Euro per year for the license including maintenance and support.

We know that we need for both systems additional FTE/money for customising.

Are the both solutions comparable when they are installed by each company regarding their functionality? Do I need the same effort for both systems for customizing?

Is OSI so much better so that we will benefit from this initial invest of 200 000 in 3 Years?

I hope to get some answers.


I am not familiar enough with Aspentech to be able to give you much detail or a direct comparison.

Obviously, you need to ensure that you are comparing apples with apples. OSI PI has many different packages, and the more features you buy, the more it costs.

For a chemical industry process I would expect that OSI PI are recommending a minimum of ....

1. A PI Server - This is the base historian. pricing is based on the number of points you want to record

2. PI Datalink and PI Process Book. - These are the basic tools for building reports and displays

3. PI Batch - This is the framework for recording batch data.

I would probably consider these to be a minimum for your application and use the functionality and price of these components for a direct comparison between your offerings. Also, you may want to check how many users can access data concurrently from both systems.

There are several additional PI packages that may be extremely useful to you (Like AF, Notifications, RtWebparts etc), but equally you may never use.

I also found this post that may be of interest. (Remove any spaces from the link)


Hello Holger

I have used both systems and read the link supplied in the previous posting. I would largely agree with the opinions on the link although I have not used the batch components.

I would also add a couple of other things.

Out of the box I would say that PI has the edge in uptime (using PI High Availability - even more expensive) and would require less development/learning time from a basic database point of view. Maybe also PI is ahead in the connectivity stakes but with OPC this may not be an issue for you. Out of the box IP21 has some nice developer type features for creating custom data structures and perhaps is a bit more flexible.

Both systems have a web type interface or you could use another product say from Matrikon (used to be ProcessNet but not sure what its called these days) for user visualisation. There are others.

There are other large historian products also that would be considerably cheaper. But here's the crux - which is better depends on your requirements. I have seen numerous project where clients have failed to think seriously enough about what they want and why. Who the users are, where the users are based, what kind of information the users need, where the data comes from, how it needs to be manipulated. You get the picture. Maybe you've done this and written a good specification.

Anyway its not just a financial decision, for some applications maybe PI is worth the extra money. Define your requirements, get the quotes in, do an analysis and see if you can live with some requirements dropping out, or if the extra is worth it.

Hope this helps


By Andrey Romanenko on 7 July, 2010 - 9:53 am

Dear Holger,

Depending on your requirements you may get a third party solution for a fraction of that costs. What is the number of data points you need to record? For how long? What do you want to do with the data? What kind of reports do you need? Do you need OPC AE support? We have already installed another plant historian and data analysis system in a large petrochemical plant.


Best Regards,
Andrey Romanenko
Ciengis - Process Optimization and Advanced Control

By Gerald Beaudoin on 7 July, 2010 - 1:09 pm

You will probably also want some of the add-ins that are available with PI. Check out the list. The one which I consider absolutely essential allows you to anotate the data. Very useful when you make changes or introduce any other anomaly to the data. Perhaps it is now part of the package...was not when we bought it.

Gerald Beaudoin

Dear Gerald,

whats the product name of this annotation tool?


I believe it is part of the SDK (Software Development Kit) that PI offers along with their product. Unfortunately the OSI PI that we purchased was obtained from a third party vendor (Rockwell). As such, the SDK was not included or even available directly from PI, since the software was not purchased from them directly.

Something to watch for as this is an essential feature for any industrial archiving.

Thanks for your answers,

I have read also the additional link.

Your answers helped me a little bit.

I know it is difficult to answer from the distance because you do not know my exact requirements to a system.

We have in our evaluation 2 other systems. One needs much more customisation than OSI and Aspen. The last is not usable because of the amount of datapoints/continous data we want to save. (needs too much disk space and Servers; factor 10)

The problem I still have are the 200 000 Euros.

With this money it would be possible to get an external expert for customising our system and pay him 2 years. He could make trainings and the documentation in that time.

From OSI we would start with the PI server and the following applications:

1. A PI Server - This is the base historian. pricing is based on the number of points you want to record

2. PI Datalink and PI Process Book. - These are the basic tools for building reports and displays

3. PI Batch - This is the framework for recording batch data.

Gives Aspen me the same tools at the beginning (out of the box) or do I have to program these applications also.
(Then the difference in the price would shrink)

My Idea to ask in forum with specialists was to find probably someone who was standing before the same decision.

Anyway the discussion is very helpful.

By jlalbrecht on 20 June, 2011 - 6:33 am

I just came upon these posts, and I'm interested in what you have (by now I'm sure already) decided to buy, and what were the deciding factors.

ACo (my firm) is a systems integrater for the last 17 years. We work with Aspen a lot, and PI some, and have done numerous comparisons for clients over the years.

No doubt, the license issue is a big issue for first timers. That being said, in general, it is a matter of the processes involved, the IT environment, and budget. There is not one "right" answer.

Anyway, I hope whatever you decided that it turned out well for you.


By bob peterson on 8 July, 2010 - 5:42 pm

I have been peripherally involved in a couple of cases where people installed these kind of systems.

I am not sure they are worth what you get for most plants. A lot of time the nifty bells and whistles the salesman demonstrated to those responsible for buying the thing turned out to be almost useless to the plant. They were neat but served no real purpose.

I think you would be well served by sitting down with all the people involved at your plant and finding out what everyone really needs up front, versus what the software salesman convinced them they needed.

It may turn out that some relatively straightforward database system within your existing IT infrastructure can be created that your IT department can maintain that will do what you need without spending a bunch of money up front and more every year for "upgrades" and "support". I almost cringe suggesting getting your IT department involved in anything, but chances are they have most of what you need already and might well be able to gin up the rest of it for a lot less than a packaged solution.

Chances are if the data is useful, a fair number of people are going to want it, and want it in different ways. If you own the thing, it will be forever your problem to deal with these requests and demands. Your IT people probably have a system in place to deal with such things that forces the people demanding such things to find a way to pay for them. If it belongs to you, and your department has to pay for it, everyone is going to want everything because they perceive it as free.

There are also the almost constant hardware and software upgrades you are going to endure. Better off to let it be someone else's problem that is better set up for it then you are. There are also a lot of other issues that are not real obvious. Data and access security, remote access, and long term offsite archiving are just a few. You IT people probably have those things solved already.

I am not suggesting these kind of software packages are of no value, but you need to take a long, hard look at it up front. Don't spend a ton of money thinking you are getting a usable system. What you are getting is a framework. Putting the pieces together might well cost you more than the software did.

By Paul Edwards on 9 July, 2010 - 4:43 am


Have you considered Wonderware Historian, it will be cheaper and faster as well as quicker to get running. I am biased though :-)

In my experience its the user tools that make the system and how easy they are to use. I am not an expert on either but I believe that Aspentech data retrieval is a lot slower than PI which could be an issue with many users.

You need to look at the client tools from the point of view of different users on the system. Casual users will want web based tools which they can use on an infrequent basis. This access needs to be licensed on a concurrent basis so you are not paying for a license for a user who only accesses once a day. You will also have regular users who want a detailed look at the data. these users may need a more feature rich client tool with their own dedicated license to always guarantee them a connection.

Why not ask the vendors to set up a pilot system for 30 days or so and identify various users and allow them to test the systems you are interested in. It will be much safer then committing to a large capital outlay only to find out the product is not quite what you want. If they will not do that then do not consider the product.



How many point in your pi system did you buy?
pls rtn to

Is there any alternative other then OSI PI, Info Plus 21 and Wonderware Historian?


There's a very interesting alternative: Plantstreamer Portal developed by Ciengis.

For further info, feel free to contact:

Best regards

My company has reliable and effective plant information system. We offer our plant information system with reasonable price and high quality product and service. We already deployed our plant information systems worldwide and we worked in this industry more than 10 years. If you have any query, please don't hesitate to contact me. Thank you-*

> Is there any alternative other then OSI PI, Info Plus 21 and Wonderware
> Historian?

By Peter Clout on 21 March, 2012 - 10:33 pm

I have fallen a little behind and I see that this post is 4 weeks old but I have to say that the choice of other historians depends rather on
the requirements. For some applications, one is restricted to a smaller range of choices. Of course, we offer one too!

Peter Clout
Vista Control Systems, Inc.

By AngusSimpson on 1 April, 2012 - 11:05 am

A little late but.............

With Honeywell's acquisition of Matrikon, a really strong alternative is Honeywell's PHD as the database with Operational Insight as the web-based client. It offers a great combination and the backing of a top company.

Check it out!

1 out of 1 members thought this post was helpful...

i just stumbled across this page now (May 19, 2012) and know this is a late response.

We have worked in the process control industry (power, paper, steel) for 20 years and have worked with all historians mentioned. All posts that say develop your own historian using SQL should be discarded. SQL is a relational database, historians are real-time series collectors that are optimized for stored massive amounts of data AND ... retrieving the data quickly.

Don't be fooled by people that say SQL can do the same thing, it CANNOT and will NEVER compete with the proprietary algorithms and data storage techniques that these companies like OSI, Aspentech, EDNA, iFix/Proficy historian use. Same with Wonderware inSQL, good for a few thousand tags, horrible for anything much bigger. Plus it's not an enteprise solution, you are not going to put a inSQL historian on your corporate network and have 30 data sources sending real-time time series data to it. Don't believe me, go and ask inSQL for 10 years of vibration that is saving every .001 mills and watch what happens.

As far as vendor, OSI is THE premier historian. They built their technology in the 80's when disk space was expensive. they developed algs that save massive amounts of data in small space because they had to. Now of days, with disk space so cheap, companies think they don't need to worry about performance anymore, just write everything out. That's all fine and dandy until look at retrieval times.

We have seen dozens of vendors abandon OSI PI for writing their own historian (Emerson is famous at it), then 5 years later they are back to OSI PI because their experiments fail. (mostly because they again think SQL is an alternative to real-time series data archives). So don't get fooled into buying a piece of crap, or developing you own unless you just have a few thousand tags and will never expand it.

Likewise, don't ever get IT involved and ask them to develop you something. They will use SQL. I can tell you of a company right now that did just that, 10 years later they have no idea how to handle all the data, so they prune it all so only has 6 months because performance is such an issue. They have a staff of 17 people in IT, DBA's, security that try to maintain it and cannot add anymore improvements because they are boxed into crappy code they developed 15 years ago.

OSI PI is best, most mature, most expensive. 200+ interfaces (although most just use OPC, Modbus, DNP these days). Love their product, hate their sales and pricing model. Depends on who you talk to in company on what price get. Negotiate hard with them, tell them it's a competitive bid, and will get much reduced pricing. If they know you only are looking at them they will take complete advantage of it. Private company (est $250M/year), Kennedy (owner) sold out 49% to private equity firm so return on investment is driving their business model now (hence the "get what you can mentality).

Aspentech, great product, good pricing, mature. They are bigger than think, $2B public company that make a lot of sofware products. IP21 just being one of them.

inSQL Wonderware. Run from this product, not enteprise and built on SQL. Invensys public company (london), $2B year.

iFix/Proficy (GE owns now). Used to be low cost alternative to OSI PI, with GE owning it now, forget that. Pretty good product, good web based tools, tools kinda clumsy to use. Huge Public company, $200B/year

EDNA-Instep. Small company, good product, config tools a little outdated (mostly config files), kinda like OSI PI in the late 80's where everything done on command line with pidiff. Private company, estimated $10M/year

Iconics Hyperhistorian. Intrigued by this product. They build a swinging-door alg on 64bit platform from ground up. They are know mostly for their SCADA/HMI software (Iconics Gensis), but have developed a historian that looks really good. We are going to tear into it over next upcoming months to see if this truly is the low-cost alternative to OSI PI. Private company, est. $100M/year

Matricon (now honeywell) ProcessNet/Operational Insight. Web based interface for OSI PI, others. They developed this because OSI PI was so far behind in their web development. Lawsuit between OSI and Matrikon, OSI wins because they were telling people to buy 1 processbook to develop graphics, then convert processbook to their web platform. Settlement involved all customers having to purchase the PISDK for the PI server ($$$) if they want to develop against it. Absurd because if they take away the PI SDK on the PI server, it won't work. Classic argument like MSFT saying need to have iExplorer in windows or windows won't work.

In any event, operational insight is decent product, a lot better than the crappy web parts that require sharepoint with OSI PI. But Matrikon has a sales mentality. Sell you everything they can even if don't need. Support is pretty poor from our experience. Other products they make are nightmare (tunneler). Hopefully Honeywell will correct, but have doubts.

All of the above is my humble opinion only, don't take it to the bank, based on my experience over the last 20 years when OSI was mostly used on VAX/VMS and had 10 people working in Cali for them. If disagree, that's ok, everyone likes what they use and hates change. Fortunately, we have worked extensively with every product above from installing, developing against, to maintaining and are vendor neutral.

Canary Labs ( has a real time Data Historian that they claim can do 1M updates per second. Anyone have an experience with this product?

By bob peterson on 7 June, 2012 - 12:44 pm

Good trick.

I got their email the other day too.

never heard of them before.

i wonder how they can pull off such a thing.


By Peter Clout on 26 June, 2012 - 8:30 pm

We have systems running in the field with our historian and recording
20M 16-bit samples/second continuously. Of course, this has to be designed as a system with regard to the hardware and software but it is running out there and collecting data. We believe it can go faster but this has not been proven yet.

Some values are recorded at many KHz.

Peter Clout
Vista Control Systems, Inc.
2101 Trinity Drive, Suite Q
Los Alamos, NM 87544-4103

By Jeff Knepper on 23 March, 2016 - 11:02 pm

> Canary Labs ( has a real time Data
> Historian that they claim can do 1M updates per second.
> Anyone have an experience with this product?

Canary can write at speeds over 2.5 million per second. We specialize in speed and reliability. Can cover systems from 5 tags to 1,000,000 plus.

>Canary Labs ( has a real time Data
>Historian that they claim can do 1M updates per second.
>Anyone have an experience with this product?

Our RAPID Database/Historian product supports storing of 4.5 Million events per second at API level. We are a reasonably priced alternative for storing and reporting process data, and we scale from very small to Millions of tags as needed. Check us out at, contact us at

hi rkb and all,

have you got some feedback after using Iconics Hyperhistorian?

we are looking for a historian for 50K tag counts, need a good reliable and competitive solution.


By Rajnikant Parmar on 16 July, 2013 - 8:48 am

Hello Kevin,

Before making your decision, have a look at our PRISM enterprise historian. It is reliable and fully featured historian to fulfill your needs.
For more info please visit,

We are BNF Technology, providing various software solutions for process plants.

With Regards,
Rajnikant Parmar

By Sudhaakar Gubba on 13 July, 2013 - 5:32 pm

I am working on an academic project from Columbia University. The project is related to provinding "comprehensive mobile solutions for remote monitoring of power plants". Looking for some information user needs, pricing model, competition, product feature set etc. If any one is willing to share their expertise/experience in this area, would you please share contact details. I really appreciate your help. For the past 15 years, I have been working in software application development. Prior to that I had 10 years experience in power plant operations and construction experience. Now my knowledge in this area is completely outdated in power industry.

Sudhakar Gubba

By Maryanne Steidinger on 6 September, 2013 - 2:20 pm

Your information on Wonderware Historian (aka InSQL) is dated, but I'm pleased to provide you with current information. The product's evolution is such that it has high tag capacity (1M +) and is rigorous enough for large, distributed, enterprise applications. We recently released a Tiered Historian which allows you to use a cloud-based hosting for the Tier 2 (reporting) layer, providing even more scalability.

RKB I very much enjoyed reading your post on data historians and the choice of suppliers; you seem very knowledgable indeed.

I know this is an old post and you are unlikely to read my response but on the off chance I would like to start a conversation with you about historian products.

A bit of background from me. I work in the Water Utility industry in the UK and we are about to start looking at procuring a historian to manage our real-time data in a better way. We are looking to develop a user requirement specification for the historian and then put the implementation out to tender. I have had OSI Soft in to chat but want to remain impartial at this stage.

If I get a respnse from you I will drop you my email and perhaps we can have a discussion, your insight could be very useful to us.
Many thanks,


By Indrajeet Pawar on 15 January, 2014 - 8:13 am

Dear Jody,

We are an Aspen Tech ISP (Implementation service Partner) from India, capable of licensing and implementing software from any domain in the Aspentech Suite, all around the world. I would really like to gauge your requirements and share some information about Aspen Tech's IP.21 solution model at the same time. Please share your email id and we will be more than happy to engage with you on this opportunity.
Waiting for your reply,

Indrajeet Pawar
Sales Account Manager
Helium Consulting

By Tapas Bhattacharyya on 3 April, 2014 - 4:23 am

The discussion about comparing the selection is very informative.
To do better analysis, the areas to be factored as Technical and commercial. Technology, standards, innovations versus use must be selectively compared.

When you finish with background and foreground, then turn to economics, and ROI to sustainability.

This methodology is generally applies here to.
In my opinion, OSI is better, but ASPEN is catching up.

I came across this post in search for data storage metrics for IP21.

I'm involved in a project were we will receive about 100 parameters from 10 locations every minute 24/7. The backend will be Aspentech IP21. That will give us about 1.44 million parameters every 24 hours. In 1 year that will give us about 525 million parameters.

What happens with the database if we are not allowed to delete ANY of the parameters we receive?

The amount of data will grow to unbelievable sizes. Yes, we maybe can archive data older than a given period, but we still need to keep it.

Are there any "limitations" or recommendations in terms of parameter history/storage for Aspen IP21?

Any hints are much appreciated.

Since I'm more into MS SQL, I'm not familiar with systems like Aspentech IP21, hence my lack of knowledge. I understand that they are not comparable...:)

Thanks in advance,

I work in OIL and GAS industry using both systems IP21 and PI OsiSoft. The data that is collected by the system is stored in archive and it is not expensive. So for 10000 tags or records with millions of records and years of data can be stored on single DVD. We have archive which holds 2 years of data. Older archives would be backed up to external storage devices.

Usually the data is stored for 7 years and after that it is deleted.

> I came across this post in search for data storage metrics for IP21.

> I'm involved in a project were we will receive about 100 parameters from 10 locations every minute 24/7. The backend
> will be Aspentech IP21. That will give us about 1.44 million parameters every 24 hours. In 1 year that will give us
> about 525 million parameters.

> What happens with the database if we are not allowed to delete ANY of the parameters we receive?

> The amount of data will grow to unbelievable sizes. Yes, we maybe can archive data older than a given period,
> but we still need to keep it.

> Are there any "limitations" or recommendations in terms of parameter history/storage for Aspen IP21?

By Praveena on 9 July, 2014 - 6:46 am
1 out of 1 members thought this post was helpful...

I'm working on OSISoft PI Historian for couple of years, So I would like to share my exp on this.

I prefer OSISOft PI to any other historian just because of its extensive features like data handling and Integration and when it comes to PI2012 Windows Integrated security. This product comes up with very good analytic tools like PI coresight, Event frames, ACE and we can also achieve condition monitoring, Identify exceptions
to obtain early warning of failures.

But still it depends on requirements and tag count. Guess u would need a package of PI Historain, PI DL, PI PB, PI INterfaces (PI OPC, PIto PI, PI UFL, PI RDBMS) PI Batch (for pharma industries), Rtreports, Rtwebparts.

By Chetan Patel on 20 August, 2014 - 3:27 am

go for GE Proficy Historian, Best in solution and pricing as well....

By Jeff Knepper on 30 June, 2016 - 2:47 pm

Believe it or not Fredrik, the 1,000 points that you are monitoring at 60 second intervals will result in a very small time-series database. Depending on the format, you are only talking about 5.3 or 9.3 bytes per instance. If the smaller R4 format, after a year, that's only a 4.7 GB file if not compressed. Any historian worth it's salt will "roll up" the file every so many days into a compressed data base. We have a similar customer who is recording about the same tag count and after 10 years they have a total database with under a 12GB footprint.

What is important is to stay away from SQL historians. If not, someone will have to manage that database. With a proprietary database like ours, there is no required data base management. Even if you want to pull trends with dates from 8 years ago, our 4.6 million TVQ/sec read speed will load it instantly. For a full performance breakdown, read here -