Is PLM Customization a Data Management Titanic?

February 26, 2010

PLM implementations are not simple. At the time when PLM vendors are working how to improve their out-of-the-box product offerings, PLM customization plays a very significant role. According to the analysts, customizations and services can be estimated as about 40-50% of total revenues in PDM/PLM domain.

What is behind these numbers and why it happens? In order to understand it, I think, we need to get short round-trip in the history of Product Lifecycle Management. The roots of PLM are in the first implementations made by large aerospace, defense and automotive manufacturers. This is the birth place of PLM and origin of PLM ideas. Since then, PLM started their journey downstream by proliferating ideas, software products and implementations. I can identify the following three trends in Product Lifecycle Management these days:

Maturity of the basic product offering
The PLM core functionality came to the stable form and mostly represented by product data management, lifecycle components and additional modules related to the business process activities – requirements, program, project, services and other processes. Interesting is that PDM and Lifecycle are considered as the most mature components of these portfolios.

Industry specialization
Initially, PLM started in aero/auto domains. However, nowadays it is moving towards all industries. In order to play industry game well, PLM vendors decided to invest into industry orientation. This trend can be characterized by a wide range of options starting from industry marketing and ending by providing packaged PLM solutions for the specific industries (i.e. Apparel, CPG, Food and Beverage, etc.)

Emerging trends
I can identify two main emerging trends – SaaS / OnDemand and Open Sources. Both are focused on how to satisfy needs of customers differently utilizing new software technologies  and deployment as well as by investing in the alternative form of business models.

When PLM industry focused mostly on providing out-of-the-box functionality, I didn’t find any technological trends focused on core data management capabilities of existing and future PLM systems. This is a very bad sign, in my view. Looking backwards, I can see significant improvements that were made in PLM software by the introduction of flexible data modeling. It allowed to decrease cost of PLM implementations, but created the huge amount of today’s customizations and implementations based on existing PDM/PLM platforms.  And this is a growing conflict between customized PLM software and upgrades to the coming releases of PLM portfolios.

I found the following Develop3D’s article as a very interesting. Al Dean is writing about replacement of highly customizable instance of MatrixOne by Open Source PLM Aras. There is more information about this event on Aras website. Read it. It looks like customer made the decision in favor of Open Source because of absence of alternatives to move to the next version of out-of-the-box MatrixOne version. I want to point out on the discussion about PLM software upgrades – PLM, Cloud, SaaS and Software Upgrades. My conclusion was simple – technology and architecture matter. If PLM data management capabilities could manage the upgrade event from highly customizable solution, I doubt the customer’s decision was to dump out existing vendors. Does it mean Aras has such technology? I don’t know. However, coupled with Open Source business model it crushed existing PLM implementation.

So, what is my conclusion? My hunch is that PLM vendors forgot to invest into data management technologies. PLM data management technologies were created 10-15 years ago. Since then, industry developed huge amounts of customized implementations. I see these implementations as Titanic pushing forward… Do you think they will be able to achieve port of destination or will die in front of icebergs of upgrades? I see it as a real and dangerous problem.

Just my thoughts…
Best, Oleg


PLM, Cloud, SaaS and Software Upgrades

February 25, 2010

I’m continuing to discuss various aspects of PLM on the Cloud and SaaS. The issue I wanted to discuss today is related to the software upgrades. This topic is considered as a painful in the enterprise software and requires additional significant investment and effort. Let think what SaaS/Cloud can change potentially? In my view, there are two separate aspects we need to analyze: technological and business.

Technology of SaaS/Cloud
From the technological standpoint, moving to cloud or SaaS solution still doesn’t mean change in comparison to what we have today when we run software on premises. If your PLM solution, for example, will be delivered on dedicated servers hosted in data centers, you are not going to experience any change in comparison to the web solution deployed in the orgnization. These servers will run the specific version of PLM software installed on them and you (or your provider) will need to care to upgrade them in the same way you are doing it today. However, if PLM solution will be delivered in the multi-tenant architecture, and you’ll share instances of the servers with other customers, you have a potential to experience non-stop migration between software versions deployed by vendors. Even in this case, the question of how it will happen to be heavily dependent on details of the system implementation, data modeling aspects and many others. So, in my view, the devil of SaaS/Cloud technology is in details. You’ll need to watch it closely and understand what can be supported by your software provider. Certainly, some of the software aspects related to data modeling will still require “upgrade point” anyway.

Cloud / SaaS as a business model
This is another aspect that I’d not connect directly to the technology of an upgrade. The most widely accepted situation when moving to SaaS model is to get software licenses paid by a subscription fee and not by license fees. These change of the business model will likely move your software expenses to a monthly/yearly payment. I had chance to read some interesting observation about that business from Burton Group blog yesterday. You can find it here. ( However, even this subscription looks like a big change, in practice, lots of SaaS contracts that promoted actively as monthly payments, will be considered as 1-2 years contracts. So, forklift event, Burton’s blog is writing about, will be created by vendor.

What is my conclusion today? Marketing is doing a great job around SaaS and cloud software. However, not all marketing presentations are translated into practical realities. It can stay in the power point slides. Even if SaaS/Cloud computing is providing an option to be different from software on premises, you need to watch details of a specific solution you are going to move to.

These are just my thoughts… What is your opinion on that?

Best, Oleg


CAD/PLM On The Cloud and Vendor Lock In

February 24, 2010

I found a very interesting Forbes Magazine’s story – The Future of Enterprise Software. The author is writing about coming Cloud Software revolution. The write up is saying- Cloud Applications are coming and vendors won’t be able to lock in customers with the specific software. Customer will be “free to stay” or “free to go” and use any software from cloud.

“While concern over lock-in has a long history in hardware and services, it’s relatively new in enterprise application software. Companies always figured they owned the software because they put so much work into customizing it. Now, it appears, they are willing to forgo that level of ownership for the ability to move freely among cloud providers for the lowest possible cost or the best security benefits”.

This article made me think more about lock in and CAD/PLM software. The story behind customer lock in with enterprise software is not simple. And this is because of massive software customization that happens in the enterprise. Customers are building enterprise solutions based on software provided by enterprise vendors – MRP, ERP, CRM, SRM. The resulting solution is something to create strong affiliation between customers and software vendors.

However, in the context of CAD and PLM is even more interesting. In addition to customization and solution building on top of software products, customer creates intellectual property (IP) that dependent on CAD and PLM software. So, this IP in the forms of CAD models, drawings, Bill of Materials, etc. is an additional strong link between software vendor and customer. This dependency always discussed by CAD and PLM industry as “the interoperability” issue. I think, this issue will come back as soon as we’ll enter to the era of cloud CAD and PLM software.

Let’s try to understand what is the fundamental behind vendor lock in. I think, the ugly truth behind this is the cost of new customer acquisition. It is very high. To get a new customer on board is an expensive process and vendors are trying to keep customers not only by providing them more value and additional products, but also by putting additional cost them to walk out and use alternative products. This cost is translating of CAD models, export engineering and product data. This process is not always straightforward and requires experience, deep knowledge of software on both sides.

Now, let’s turn back to the cloud software. What is that about? I think, on the fundamental level, this is also about how to decrease new customer acquisition costs. Cloud software as services can simplify a selling process, can provide an easier way to demo software and provide free trial versions available for customers. In the end, fewer vendor bucks will be spent to get another happy customer on board. Done deal! What’s next? Do you think customer’s walk away from cloud software will be different than a similar process with software installed on premises? No, I don’t think so… The rest is very similar, in my view. Customer’s data, customization and many other factors will keep users on these systems exactly in the same way as it was before.

So, what is my conclusion today? Cloud is the way to software vendors to simplify the selling process. This is a huge advantage. In addition to that, there are many additional advantages related to the software on cloud. However, vendor lock in is with us to stay….

Just my thoughts… What is your view? What are your expectations from cloud software business relationships on both sides – vendors and customers?

Best, Oleg


Do We Need Chief Excel Officer To Manage BOM?

February 23, 2010

I had chance to read yesterday Arena’s whitepaper “Whose BOM is it Anyway?”. You can find the link this whitepaper for free download here.The white paper is written as a Disaster Story from a manufacturing going trough the hassle of BOM management using Microsoft Excel.

In the end of white paper, there are 3 tips how to take control of your BOM.

#1 Provide Multiple Views into a Single Consolidated BOM. Every department needs different information from the BOM, so you need to present your BOM in a format that allows each department to access and incorporate the information that is vital to them, without having to fork off a new version for their own use. Consider using a tool designed for this purpose instead of something like Excel, and keep all departmental information related to the BOM in one centralized location that is accessible by all.

#2 improve Visibility of Changes through redlines and notifications. The ability to redline a BOM is critical. A single change in a BOM of 20 columns and 500 lines is very easy and too costly to miss. Redlines enable someone who isn’t intimately familiar with the BOM to pick it up and know instantly what needs attention.

#3 Establish a review and Approval Process that Everyone Agrees to and Stick to it. By consistently following your process, you can make sure that all your changes are reviewed by the right people across multiple departments. This will help you prevent costly errors that result in scrap and rework and ultimately delay your product getting to market.

In addition to these three tips, there is a very nice picture shows Bill of Material shows how Your Product Data is managed by Arena Solution.

This white paper stroke me to think again about my post related to an idea of single Bill Of Material – Seven Rules Towards Single Bill of Material. The main point of my post was about how we can eliminate multiple Bill of Materials in the enterprise and replace it with single one, that contains all we need from design to manufacturing.

As I can read tip 1 – “Provide Multiple Views into a Single Consolidated BOM” with option to have a multiple set of views into this BOM. This is really good news. However, I found the picture presenting bi-directional data exchange with CAD/PDM and ECAD/Component Library. So data need to be synchronized in two directions. In addition ERP/MRP is presented as Item Master Ref (which I believe a reference) to product data in single BOM and probably mean data never going to be synchronized from ERP back to PLM/PDM. This is something that rarely happen if PLM consider as a Single Consolidated BOM.

At this point, of time, I came with strange feelings, similar to what Carlos had from the same white paper. BOMs are starting to fly all over!  However, now, Bills are flying between Product Data storage and all other systems. All systems mentioned on this picture, need to be somehow synchronized. So, you can see flying gizmos inventing magic vacuum cleaners to suck Bill of Material data from the central place where Product Data is stored and moving it between and around with CAD/PDM, ECAD, ERP/MRP. How we can organize this data exchange? Oh, I’m sure there are magic XML (or maybe Excel) files that can go back and forth and do this work. My hunch is that this work is very complicated and the cost of such integration is not cheap.

Unfortunately, this is a reality of every engineering and manufacturing company. Flying Excels with Bill of Materials, or any other representations of the BOMs in any format, is the ultimate way to do real Bill of Material job in the organization. And, this is very expensive work. There are two alternatives I had chance to see in the organization:

(1) to invest big $$$ to make a comprehensive integration (I’ve seen such companies and they are doing very well);
(2) to hire Chief Excel Officer to take care of all BOM/Excels in the company (this is the mainstream situation).

So, what is my conclusion today? The white paper touched the real problem. There are no magic gizmo that can solve integration problem related to Bill of Materials. To create a single point of product data in the organization is not simple. Period.

Just my thoughts…
Best, Oleg


Regulation and Future PLM Renaissance

February 22, 2010

I’ve been reading white paper by Kalypso – 4 Reasons Why Industry Leaders Aren’t Losing Sleep. Max Cochet explained four best practices how industry companies need to manage in order to sleep well during turbulent regulation rules. Some of them made me think about opportunity regulation rules created for Product lifecycle Management. Before talking about it in detail, I’d like to put some quotes from Kalypso paper. You can read the full paper following this link.

Consumer goods industry leaders follow an informed vision. Regulations may be going in one direction today and in another direction a couple of years from now, so it doesn’t make sense to base a long-term strategy on upcoming regulations. Companies need to make million-dollar bets on strategies that fit their vision while allowing some flexibility. Adaptation, rather than reaction, is the most efficient method to handle changes to the business environment. a couple of years

Integrated product lifecycle management (PLM) systems and processes enable companies to manage their existing and future product data and act efficiently to adapt to regulations. Smart investments such as these enable best-in-class companies to keep their house in order and prepare for future transformations of their environments.

Now, why I found this paper interesting? Regulation becomes a serious driver to organize product record in the company. This is something a company cannot avoid and need to follow. At the same time, regulation is something that is going to change all the time. What Kalypso paper offers are to follow PLM strategy and create flexible PLM that can keep full product information record in the company and make it compliant to the future regulation rules. Sounds like very smart? Flexible PLM option makes a lot of sense.

Pre-configured PLM best practices and Flexibility
Looking on many websites of PLM companies these days, I see a huge effort into presenting industry solutions and best practices for available PLM solutions. I can find it on almost every PLM vendor website. So, I started to think how a recommendation to have flexible and integrated PLM can live together with pre-configured business practices. My hunch is that marketing is working extra-hours these days by trying to present how PLM solution can fit business requirements. However, flexibility as one of the basic requirements have been left behind.

PLM Renaissance, Regulation and Flexibility
Why I see regulation as one of the drivers for PLM renaissance? The huge amount of regulation rules has a potential to burden life of manufacturers in industries. The number of industries going under regulation is growing and these companies will be looking for a solution. At the same time, as it was mentioned in Kalypso’s paper, regulation is dynamic and going to change within the time. So, investing into specific solution for regulation doesn’t make a lot of sense. To invest into a flexible system (and PLM can be such a system) to organize data in the organization to make it compliant to the many regulation rules – this can be a good opportunity for PLM Industry.

Just my thought…  Does it make sense to you? What do you think?
Best, Oleg.


What Are Your Questions To PLM Virtual Assistant?

February 19, 2010

When somebody is asking me what is the number one PLM software problem, my instant answer is – complexity. Yes, PLM is a dumb complex. I think, discussions about the complexity of PLM are endless. Each time I hear about new PLM product, the first requirement is to make it simple.  There are two major reasons why PLM is complex. They come from opposite sides of PLM business – product related and people related. From the product related standpoint, we need to admit – engineering and manufacturing are a complex discipline. Product complexity is growing, so we are facing growing complexity in design, engineering data and manufacturing. On the other side, from people’s side, engineering nature it to see all problems in the complex way, analyze dependencies, etc. So, engineers are constantly coming with more and more complex requirements. The result is simple – when you operate PLM software, you need to think… And you think to think a lot…

The complexity of applications creates a lot of problems to the people operating this software. However, recently, I started to experiment with personal assistant on my iPhone. Siri (

Now, let me switch your imagination on. Do you think we can apply this interesting assistant behavior to help me to operate my complex PLM software? Or, in another way, can my PLM experience changed to be as simple as experience with the virtual assistant. I think that it may work. PLM virtual assistant may fit “a role oriented” work PLM products need to do in your organization. What about questions like -

1. What is the most urgent work I need to do today?

2. Who is responsible of the ABC part manufacturing?

3. Where is the last revision of my drawing?


Now it is your turn. I’m sure you can come with much better questions. So, now think about what type of questions you can ask PLM virtual assistant if you have one in your organization? Sounds crazy? However, let’s make a try…I’m waiting for questions you want to ask you PLM virtual assistant.

Best, Oleg


Social PLM Enterprise and Federated Identity

February 18, 2010

I’d like to put some thoughts about user identity management in the enterprise. In the beginning, you may think the topic is obvious. Enterprises already solved this problem long time ago. Even if your need to login into the specific enterprise system was significantly decreased, it sounds like a problem disappeared. You can think that directory management systems solved the problem. However, is it really true? I’d like to propose to review this problem with a different angle – social.

My guess is that next big competition in the enterprise will be around a user. I think, very soon, enterprise software vendors will learn lesson or two from web and social software. They will discover that the key to know who are your users, how they use systems, how they navigate between different tools and environments. And, of course how they connected between them. All social relationships in the enterprise became very interesting.

I hope you got my point… This is the effect of social networks and communities. For the moment, this information belongs to IT organization. Most of the enterprise vendors are seamlessly agreed to synchronize this information with the available LDAP or similar directory management tool. But, I think this situation can become different soon. My hunch is that first to discover this will be enterprise software companies pushing their social platforms to the enterprise. They will try to take ownership on directory services and expand it to the level of social networks. Another interesting problem is what technologies can be used to do that? My guess that possible candidates can be open standards, like FOAF, and future integration of them into enterprise systems. I think, very soon, we are going to discovered need in the federated identity management for enterprise social nets.

Now let me ask you, what about PLM? My assumption is that PLM is one of the potential originators of the federated identity management for enterprise social networks. Running processes around almost everything in the enterprise, can be a good reason to acquire enterprise social networking. If this task can be accomplished, PLM can drive significant advantages from managing people relationships, product and project belonging, etc. Who are alternative owners? My guess, ERP is a good candidate. They already own a big portion of enterprise cake, so they can do the same for identity management.

Take a look on long, but interesting video on digital identity.

Just my thoughts… I’d be interested to know your opinion on that.
Best, Oleg


PLM and BIM Interplay in Enterprise Data

February 17, 2010
Interesting news drove my attention yesterday. Bentley System made an acquisition of the company called Enterprise Informatics. When I was reading press release, my guess was that Bentley decided to expand their capabilities in data management. Especially, this expansion can be very interesting when working in construction and power process industry for asset management. So, that was the exact target for acquisition.
Bentley Systems, Incorporated, the leading company dedicated to providing comprehensive software solutions for the infrastructure that sustains our world, today announced that it has acquired Enterprise Informatics, Incorporated ( and Exor Corporation ( The Enterprise Informatics eB Insight software provides configuration and change management capabilities for mission-critical infrastructure asset operations for the energy, nuclear, rail, and government sectors.

I think, this is an interesting move.We are starting to see potential convergence of two initially separate segments – PLM and BIM. Each of them has different origins. PLM started in the large aerospace and defense companies. BIM came from need to manage complex construction processes. However, my hunch was that both need to use very similar underlined data management technologies. First, I wrote about that last year in my post – PLM and BIM: Common Roots or Common Future.

Looking on eB product from Enterprise Informatics, I figured out a significant similarity with core concepts between what was called Information Models for Enterprise and Product Lifecycle Business Processes. The marketing slogans are different. However, data management, change management, process orchestration are very similar. I put few slides from eB presentation (you can watch 5-minutes video on this link).

So, what is my conclusion? This is a very interesting potential interplay between construction industry and traditional PLM implementations. Technologies used on both sides are very similar.  My assumption back that time was that PLM and BIM will come to the certain point of commonality in their ability to manage complex data models, changes and processes. And this is what happens now. I’d expect potential movement from both PLM and BIM towards interesting projects in industries where they have no sole proprietary ownership as leading solution providers.

Just my thoughts…
Best, Oleg


Social PLM Options

February 16, 2010

Social is hot. Social networks, social applications, social enterprise, social CRM… social PLM (?). I put a question mark, since I wasn’t sure if to use a word social is an appropriated in this context. What means “social” for PLM? How it is different from what we’ve been doing in PLM during the last decade?

Collaborative PLM
We used to apply the word “collaborative” in PDM/PLM context many times. Despite the magic of this word, for me, it was always about “working together”. And, I think, this is true. PDM and lately PLM put huge focus on how to share data in the organization, how to access this data collaboratively. It started from collaborative review and move into more complicated areas of design, engineering, etc. Collaborative technologies in PLM passed three stages: 1/Share Data; 2/Exchange Messages; 3/Work in real time.

Social Trend
The trend for social software started from Web 2.0. Actually, Web 2.0 was a lot about how we can generate content on the web. It started as blogs, wikis and web sites provided different user experience (i.e. flikr etc.). People started to collaboratively create content on the web. In the beginning, it was purely about technologies like AJAX and user experience. It was a turning point point for Enterprise 2.0- how to bring so successful Web 2..0 user experience and new models of communication and collaboration to enterprise. Within time, it came to the point where content became secondary and actually social communication (or people) came to the first place. The “aha moment”, in my view, was when everybody understood the value of collective information about social communication and data that people generate. It wasn’t about accessing data anymore. It was about how people interact and how to get value from this interaction.

Social Trend and Enterprise Software

Enterprise Software discovered hidden potential of social trend and implication on what currently happens in the enterprise. The first who came into this game was CRM, and it is obvious why it happens. CRM is all about communication with customers. However, who are your customers? These are the same people on twitter, Facebook, linkedin and many other social networks. So, if CRM can leverage this social communication, stay connected with customers, get feedback and early alert – this is a huge value for enterprise. Take a look on the video of industry analysts Michael Fauscette and Natalie Petouhoff discussing the intersection of Social CRM and the enterprise. Recorded at the Social CRM Summit last week. I found it very interesting.

In parallel, I can see other enterprise domains are also moving towards better understanding of what social trend can bring to them. I think, we are in the early beginning of this trend. Some elements of social software in the way of user experience, people communication started to appear in the portfolios of enterprise vendors. Community software is one example of that experience. I see it very positive. Social networks provided a very significant push to the development of social software. However, it was transformed into multiple “social platforms” in the enterprise. It was a funny transformation in my view. Now, every big enterprise vendor, started to propose their own “social network”. We are still not at the mature level. However, all trials are very interesting.

So, what about Social PLM?

Finally, I wanted to get back and answer on the following question: where is the intersection of social trend and PLM? In my view, implications can come in the three possible areas: collaborative technologies, user experience and value of social networks.

Collaborative Technologies: Social software developed bunch of very interesting technologies for collaborative work. These technologies related to the ability to manage big amount of data, involve many participants simultaneously. I’m sure PLM will need to learn lesson or two from this space. It contains a huge amount of innovative ideas and technological achievements. We are in the early beginning of this discovery.

User Experience: Social software came with the next level of user interface, usability, behaviors, etc. These fresh ideas and very broad use adoption level, need to be translated into PLM. Today’s PLM user experience is far from the consumer software level. We can see it started to happen already. There are multiple examples of re-use of Facebook’s user experience, Twitter micro-blogging communication, etc.

Value of Social Networks: This is the most interesting space, in my view. We will need to think how to bring accumulated value of social network into PLM. This is especially interesting for manufacturing companies facing consumer space. And, almost all manufacturers these days are looking how to get connected to the end users. Such social innovation can be a very interesting experience for Product Lifecycle Management.

Just my thoughts…
Best, Oleg


PLM Social Disruption – Whom Should We Follow?

February 15, 2010

I want to get back to social applications and PLM topic. You can ask me why? Last week was the week Google Buzz was born. In my view Google did it, as usual, in  a very disruptive way. I remember, last year or so, Google introduced Google Profile. Nobody clearly understood what is that. I did one for myself. Last week everybody finally discovered what is Google Profile about – it is about your social identity. Btw, if you haven’t had a chance to read my Google Buzz notes, please find it here.

Google Buzz was a trigger, in my view, to start a discussion about what are fundamentals of social connection? What I like about social application is that all of them put people before processes. When I asked teenagers – why are you on Facebook? The answer was simple – because everybody is there. Interesting comparison was made – Facebook is the place to see and stay connected with friends and Twitter is the place to spy celebrities. This is an interesting observation. You are following people you consider as your social or professional connection. And existing social networks provide a clear answers – you can follow celebrities, friends, colleagues, brands- these are they way things become social.

Let’s get back to Product Lifecycle and Enterprises. The application of social tools in the enterprise is in the very early beginning. The most massive social attack on the enterprise is fast growing number of social platforms promising the creation of communities between users, customers, brands, etc. Chatter, 12sprints, BlueKiwi, PTC Social Product Development with Product Point… This is an incomplete list of all apps proposing creating of community of people like Facebook. Finally, we got the new one – “Facebook for files”. I got this definition from Chris at Vuuch in his post “Following A Friend Make No Sense”. So, “whom should we follow?” was the most important question for me to ask?

It is about people.
Files and other data were here before social systems came in. Somebody can put an assembly, file, word document. You never know somebody did something with that. If you will try to spy after all files, things can get crazy. But, if you are following people you’re working with, it can be more relevant. In the end, you are not working with millions. On average, you connected to couple dozens of people in the organization you work. So, stay connected with them, and you’ll get all you need.

It is about data you can get through your connections.
You work with people, so you can collaborate on data everybody owns. Design, bill of materials, changes, product definition, customer’s claims, etc. – all these collections of data can be available from the people are working with. So, stay online with your connections and you’ll have all up-to-date stuff. You’ll become a focal point for the data you own.

It is about openness and share.
If you are in the open social world – share what you have to say. If you’re in the organization – share what you need to work on. If you will do it, the most probably it will allow you to get connected to the right social focal points in your organization and extended enterprise.

What is my conclusion today? You cannot follow “data” or “files”, since data is not social. Files are not social. You can get connected to data by using your social connections. It doesn’t matter if this is an enterprise organization, your high school or college. The aim of social software is to bring “the relevant data” to you. So, in PLM – follow people you need to work within the organization. This will be your ultimate way to get the most relevant information you need. And this is, in my view, the purpose of social PLM (and other social apps)- to get your data by using your organization’s social network.

It will be interesting to know what do you think?
Best, Oleg



Get every new post delivered to your Inbox.

Join 244 other followers