Building an IOT Platform: Centralized vs. Decentralized Models


Last week I delivered a session at QCon New York about the emerging trends and techniques powering the next generation IOT platform as a service (PaaS). The session explored the current IOT platform market as well as some of the architectures and technologies that will be foundational to the adoption of IOT in industrial enterprise settings. Based on the great feedback received during the session, I am planning to write a series of blog posts about some of the most important ideas from my presentation. Let’s start with one of the key points of my presentation: centralized and decentralized capabilities in an IOT PaaS.

Most people tend to associate IOT PaaS models with centralized architectures in which a hub (typically powered by the cloud) controls the execution of nodes (smart devices). While this model is certainly essential to implement industrial IOT solutions, is rarely sufficient. Decentralized computing models in which nodes in an IOT interact without the control of a central authority are also fundamental to enable the next generation of IOT enterprise solutions. Let’s explore both models at a high level.

Centralized IOT Platform Capabilities

As mentioned in the previous section, most emerging IOT PaaS are generally associated with cloud architectures in which a central hub provides a series of backend services to smart devices. In this architecture style, smart devices act as a recipient or consumer of data while the central hub enables centralize services and capabilities as illustrated in the following figure:


Some of the key centralized capabilities of an IOT platform

  • Event processing
  • Enterprise system integration
  • Device discovery
  • Device management
  • Event Notifications
  • Real Time Analytics

Cloudfoundry as the Foundation for Centralized IOT

From the different technologies in the market, I believe Cloudfoundry provides a great model for powering centralized IOT PaaS architectures. As a platform, Cloudfoundry provides an open, platform agnostic model that abstract essential backend features of PaaS infrastructures while enables the flexibility of incorporating third party services and infrastructure capabilities. In the same way that Cloudfoundry is currently powering Web and mobile solutions we feel the model can be extended to the IOT space.

Decentralized IOT Platform Capabilities

While centralized services are essential elements in an IOT topology, they are not sufficient for powering industrial IOT solution. In addition to integrating backend services, there are many scenarios that require autonomous communication between smart devices in an IOT topology without the need of a central hub. We like to call that architecture style decentralized IOT as illustrated in the following figure:


From the functional standpoint, decentralized IOT models can help to enable some of the following capabilities:

  • Peer-to-Peer Messaging
  • Decentralized Auditing
  • Decentralized File Sharing

The Block Chain as the Foundation for Decentralized IOT

Implementing decentralized IOT capabilities requires an infrastructure that enables nodes in a distributed topology to perform autonomously. With autonomous computing comes the need to maintain trusted relationship between the nodes without a centralized authority. From the technologies in the market, the Bitcoin Block Chain provides a mechanism to enable this type of distributed model. While Bitcoin is the most famous application built on the block chain is certainly not the only one.

As a platform, the block chain provides the building blocks to enable nodes in a distributed topology to exchange data and perform tasks in a trusted and verifiable manner. I believe the block chain can provide the foundation to enable decentralized capabilities in an IOT topology. IBM Project Adept is attempting to take the first steps to validate this model. More about this in a future post.

Leave a comment

Posted by on June 17, 2015 in Uncategorized


Tags: , , , ,

Why a Tech Bubble Might Not Be as Bad as It Seems?


There are a lot of debates within the public market and technology circles of whether we are in another tech bubble. In the past, I’ve been written extensively about several arguments that explain why I don’t believe we are currently experiencing a bubble in the tech sector. While certain aspects of the private technology market like pre-IPO valuations are certainly “bubble-ish”, the public markets seem to be trading tech stocks at very reasonable valuations and you can even argue some of the market leaders like Apple or Google are trading below its real market value.

If any other reasoning fails to explain why we are not in a tech bubble we can use the number one rule of economic bubbles: “If everything thinks we are in a bubble, then is not a bubble” ;).

Despite of the previous reasoning, we can’t deny the fact that certain characteristics of the current tech market presents signs of a bubble. However, you can make the argument that is could be a positive bubble.

Don’t take me wrong, we all know that bubbles cause a lot of economic damages when they burst. However, people often ignore the simple fact that some of the most powerful economies in the world have been built on the back of a series of economic bubbles. Take a look at some of the most famous economic bubbles in history:


While some bubbles like the US 2008 housing bubble are just based on speculation in an existing market, other bubbles are based on the creation of new industries. Is the second type of bubble that, after it bursts, leaves behind an infrastructure that creates new generations of wealth and moves society forward. From that perspective, you can argue that, despite its negative consequences, bubbles like Britain’s railway mania in 1840 or the dot-com boom of the late 1990s created the necessary infrastructure to explore new frontiers and create brand new industries that became pillars of those economies.

When looking at the current state of the tech ecosystem, if we can get passed the crazy valuations in growth stage companies, we will see that the innovations in areas like mobile computing, data science, 3D printing, robotics, IOT, augmented reality, artificial intelligence etc, are likely to create a lasting infrastructure that will survive any potential bubble. While we should continue debating whether we are currently experiencing a bubble in the tech sector, we can relax a bit and think that it might be a good bubble ;).

Leave a comment

Posted by on June 9, 2015 in Uncategorized


Tags: , , , ,

EMC Acquisition of VirtuStream Closes the Door to a VMWare Spin-Off


Two days ago, storage giant EMC announced that it has reached an agreement to acquire VirtuStream for $1.2B. VirtuStream is not the most well-known platform in the cloud market but one that has developed a considerable penetration in the enterprise space. This move follows the recent acquisition of CloudScalingrepresenting another step in EMC’s strategy to develop a robust cloud portfolio.

While the acquisition brings very clear benefits to the EMC cloud stack from the capabilities standpoint, it has raised some interesting questions about the future of EMC’s crown jewel: VMWare. In recent months, EMC has been under pressure from activist investor Elliott Management to spin off VMWare as a separate company. The hedge fund manager has been critical about EMC’s federation structure under which companies like RSA, Pivotal and VMWare are run as independent companies. The debate about a potential VMWare spinoff has continue throughout 2015 in part fueled by the VMWare’s relatively poor market performance. The acquisition of VirtuStream seems to be a clear signal against those plans.


From a technical perspective, VirtuStream allows companies to transition applications to the cloud in a secure and compliant way. The xStream cloud management platform offers a consistent management   across public, private and hybrid cloud topologies. Those capabilities clearly align with the VMWare’s technologies like vCloud Air that provide a robust option for hybrid cloud enterprise topologies. From that perspective, the acquisition of VirtuStream only indicates EMC commitment to build a robust cloud portfolio in which VMWare is the central piece. In that sense, VirtuStream integration with vCloud Air will accelerate the path for customers migrating applications to a hybrid cloud topology powered by the VMWare stack. Regardless of how the integration between VirtuStream and vCloud Air materializes, its pretty obvious that EMC plans to keep both business running under its “federation”.

Despite the technical alignment and the fact that the price for the acquisition was relatively reasonable and the technology, investors clearly were not thrilled with the EMC decision. To prove the point: EMC shares were down 2.2% at $26.25 in mid-afternoon trading on Tuesday. EMC’s 52-week trading range is $25.07 to $30.92, and the market cap is $51 billion.

Leave a comment

Posted by on May 28, 2015 in Uncategorized


Tags: , , , , ,

4 Reasons Why Big Tech Companies Continue Splitting

bridgeComputer Sciences, the giant IT consulting company founded in 1959, announced yesterday that it will be splitting into two different companies. Under the new structure, CSC Commercial will serve Fortune 1000 companies while CSC US Public Sector will focus on servicing government entities. Following HP, Symantec and eBay, CSC is the latest of a series of tech companies that have announced splits in the last year. Other legendary companies like EMC have been under constant pressure to spin off VMWare into a separate entity. This phenomenon is becoming a strong reality of large tech companies to the point that legendary venture capitalist Marc Andreessen presented the provocative thesis that most tech companies older than 20 years will be forced to split.

Despite the media fascination with the split of large tech companies, it’s pretty obvious that those decisions are extremely complicated. Companies like eBay, Symantec, HP, CSC or EMC are legendary institutions in the history of the tech industry and many of them have enjoyed stellar performances in the public markets. In that sense, what can motivate or force such institutions to split into different companies? While the explanations are far from obvious, there are a few factors that might indicate the possibility of a split.

One Business Unit Starts Growing Faster than the Core Business

One of the main scenarios for a company split is when a business unit of the company starts growing exponentially higher than the others. Arguably, this is the case of eBay in which Paypal’s revenues recently became larger than eBay’s core business. In those situations, many shareholders see a split as a potential solution to allow the fast growing segment to continue evolving faster while the core business units can grow without the distraction of a faster growing little brother.

Different Business Units Evolve Under Drastically Different Business Models

Another cause of a company split could be seen when two groups of business units within a company evolve under completely different business models. In this scenario, we are not necessarily referring to a group of business units growing faster than the rest of company but rather under different business models. HP could be used as an example of this scenario as the enterprise software business units have grown following a business model completely different from the core printing business. When faced with this scenario, companies typically faced challenges enforcing common strategies and goals which makes the split an appealing scenario.

Wall Street Doesn’t Know How to Value the Company

In many cases, the stock price of large tech companies have suffered because Wall Street have difficulties valuing the company with many diverse business models. In those scenarios, sometimes shareholders believe they will benefit from a split that could simplify the valuation of the stock. An example of this scenario (not from the tech world) can be found on GE recently announcing the sale of the assets of the GE Capital unit.

Agility in New and Competitive Markets Becomes Challenging

Every sector and geography in the world can be potentially considered a technology market. Some sectors also experience very fast grow and become extremely competitive with new generation of startups entering the space. In those situations, business units of a large company can experience continuous challenges to expand into new areas or geographies or simply remain competitive in fast growing spaces. As a result, many shareholders see a split as an option to allow the company to regain its agility without the pressures and distractions of the other business units.

Leave a comment

Posted by on May 21, 2015 in Uncategorized


Tags: , , , ,

Not All IOT Platforms are Created Equal


A few days ago, I was part of a super interesting debate with several thought leaders in the internet of things (IOT) community about the evolution of IOT enterprise platforms. The core point of the debate was trying to identify the different channels by which enterprises will be exposed to IOT platforms.

In the past, I’ve been very vocal that the enterprise IOT will inevitably produce a new type of platform. That thesis is no longer a theoretical argument as we are already seeing the first flavors of IOT platforms starting to make inroads in the enterprise. Even though this generation of IOT platform represent the first iteration in the space, we can already see marked differences between different types of platforms. As an organization looking to embrace an IOT platform, this initial diversity can result very confusing.

IOT Platforms Provided by Traditional Enterprise Software Vendors

  • Overview: Traditional enterprise software vendors like Oracle or IBM are already heavily invested in extending their capabilities into the IOT space. As a result, these vendors are likely to bring to market IOT platforms that are deeply integrated with their existing technology stacks , vertical solutions and services.
  • Should Excel At: in my opinion, the IOT platforms produced by traditional enterprise software vendors are likely to be commercialized as part of industry specific industry solutions rather than standalone platforms. In that sense, this type of platform will excel at providing industry specific capabilities as well as tight integration with the products and platforms provided by the specific enterprise software vendor.
  • Should Not Be Great For: Complementing the previous point, we think IOT platforms provided by traditional enterprise software vendors will be commercialized as part of domain specific solutions and will require a decent level of professional services and training. In that sense, we can make the argument that those types of platforms won’t be great for building general purpose solutions without the involvement of the target enterprise software vendor.

IOT Services Provided by Platform as a Service Providers

  • Overview: Platform as a service (PaaS) providers like AWS, Azure or IBM are already providing specific services that abstract fundamental backend capabilities of IOT solutions. This trend is only going to increase in the near future as more and more cloud providers start building a presence in the IOT space.
  • Should Excel At: The IOT capabilities of PaaS solutions are typically provided in the form of individual services. In that sense, this type of solutions should see a broad adoption within the developer and startup communities. As a result, we should see a lot of the IOT PaaS services being adopted by startups providing smart devices or industry specific solutions as well as IT organizations building their own internal infrastructure.
  • Should Not Be Great For: The IOT capabilities provided by PaaS platforms are likely to lack the consistency of complete IOT platforms. Also, the cloud nature of this type of solution should present challenges for organizations building IOT solutions that require to be deployed within their premises.

IOT Services Provided by Smart Device Providers

  • Overview: From startups to big software companies like Cisco or Texas Instruments, IOT device providers are starting to build the first incarnations of IOT platforms that work consistently across their device portfolio. These IOT platforms will be tightly integrated with the specific family of hardware devices as well as the corresponding manufacturing toolkits.
  • Should Excel At: Platforms provided by IOT hardware providers should be best in class enabling solutions powered by those specific devices. Similarly, this type of platforms will provide consistent backend services and management experience for solutions powered by those smart devices.
  • Should Not Be Great For: While the IOT platforms provided by smart device providers should excel in solutions powered by those devices, they are likely to result limited for general purpose IOT solutions. In that sense, is unlikely that third parties will embrace this type of platforms for building new IOT solutions.

IOT Capabilities Provided by Enterprise Mobility Management Platforms

  • Overview: Enterprise mobile management platform(EMM) vendors such as BlackBerry or VMWare are starting to make the first inroads in the IOT space. If you think of mobile as a subset of IOT, the assumption that a lot of the current capabilities provided by those platforms should be adaptable to the IOT space makes some sense. As a result, we are likely to see that group of vendors providing hybrid platforms that enable both enterprise mobile and IOT solutions.
  • Should Excel At: If we use the enterprise mobile space as a reference, we are likely to see strong IOT security, device management and other operational capabilities included in the IOT platforms provided by traditional enterprise mobile management platform vendors. That model will allow the EMM vendor to extend their existing footprint in the enterprise mobile ecosystem into the IOT space with a consistent value proposition.
  • Should Not Be Great For: While IOT platforms provided by EMM vendors should excel in the security and management capabilities, they are likely to not result an ideal platform for developers building IOT solutions. For years, EMM vendors have evolved cultivating devops as their target customer which entails specific product delivery, sales and marketing models. EMM vendors should continue expanding on this model as they enter the IOT space.

New IOT Enterprise Platforms

  • Overview: Like any other transformational movement in the history of enterprise software, IOT will produce a new group of startups and platforms that will help enterprises build and manage the new generation of industrial solutions. These platforms should provide capabilities such as complex event processing, security, real time analytics, operational management while also providing a friendly interface for developers. We are already witnessing platforms like Kii, Xively, 2lemetry (Amazon), ThingWorx etc start to make progress in this space.
  • Should Excel At: From all the different channels explained in this article, the new wave of enterprise IOT platforms is likely to produce the biggest wave of innovation in the entire space. We should also expect to see these type of platforms being delivered using both on-premise and cloud models as well as leveraging open source vehicles. Additionally, these new group of platform is likely to provided the broadest levels of integration with new hardware manufactures and IOT solution providers.
  • Should Not Be Great For: As explained previously, the new generation of enterprise IOT platforms is likely to excel at innovation and openness. However, because this type of platforms are just evolving, we should not expect to see a lot of industry-specific solutions powered by these platforms in the immediate future. Additionally, as the enterprise OT space evolve, some of this new IOT platform startups will be acquired by larger enterprise software companies inheriting some new commercial and delivery models.
Leave a comment

Posted by on May 14, 2015 in Uncategorized


Tags: , , , , ,

The List of Companies that can Acquire is Smaller than you Think   


Yesterday, Bloomberg broke the news that have hired advisers to evaluate a potential takeover offer. The news spread incredibly rapidly and the CRM stock had to be halted due to volatility. When trading resumed, the stock was up 10% trading at an all time high.

As media outlets started speculating about the potential acquirers, there was a consensus that only IBM, Google, Oracle, Microsoft and SAP have the sufficient market cap and cash to afford what could be considered the second highest technology acquisition of all times. However, as we start analyzing each potential acquirer, we quickly realize that the list of smaller than we think.

Let’s take a look:


  • Pros: Microsoft seems to have over $95B in cash that could be deployed into M&A activity. The Redmon company has established a strong relationship with on the CRM side and an acquisition can help immediately help their Office365 and business suites. Other components of the platform like the social analytics and marketing platform can also be a great fit for Microsoft’s portfolio.
  • Cons: While the CRM suite seems to be a great fit for Microsoft, we can’t say the same about the the rest of the platform. Specifically, there is a strong competition between the Salesforce1/Heroku and Azure platforms which will be hard to reconcile. Additionally, keep in mind that Azure and Salesforce1 have been built in different technology stacks. Finally, a takeover offer doesn’t seem to be the style of the Satya Nadella and the current Microsoft board.


  • Pros: Acquiring will represent a string accelerator toi IBM’s SaaS business. Also the Salesforce1 platform fits nicely with IBM’s aggressive investments in the mobile and IOT spaces.
  • Cons: With only about $9B in cash, IBM doesn’t seem to have enough liquidity to embark in such an aggressive acquisition. Also, similar to Microsoft, IBM is heavily invested in their cloud platform which presents some serious overlap with the Salesforce1/Heroku stacks.


  • Pros:com can be a very strong and necessary addition to Google’s enterprise business. Additionally, Salesforce1/Heroku can help to expand Google Cloud’s capability set which is still trailing competitors like AWS or Azure.
  • Cons: Acquiring will be a strong shift from Google’s current trajectory making it’s enterprise business one of the most relevant business units of the entire company. Also, a hostile takeover doesn’t seem align with Google’s culture.


  • Pros: SAP has embarked in an ambitious effort to modernize its existing business suite. Acquiring could be the accelerator needed to effectively execute on those plans. The marketing and analytics suite seemed to be a perfect fit for SAP. Also, the Salesforce1/Heroku platforms can really help SAP’s struggling cloud business.
  • Cons: SAP seems to only have about $5B on hard which will require the German giant to take on some debt to pursue the acquisition.


  • Pros: Oracle seems to be a great candidate to acquire The CRM and business platform can really simplify and help Oracle’s chaotic SaaS business. Salesforce1/Heroku can be a great fit for Oracle’s Cloud stack which is lagging competitors like IBM, AWS and Azure. Also, don’t forget that leverages a lot of Oracle technology which will make the technical integration slightly less challenging. Finally the existing relationship between Benioff and Larry Ellison should not be ignored.
  • Cons: Oracle has reported to have around $14B in cash and another $30B in securities. In that sense, Oracle will have to assume some heavy debt to pursue the acquisition.

I hope the previous analysis makes sense. In addition to the previous list, Amazon, Alibaba and EMC could also be considered as potential acquirer although not at the same level of the ones included previous list. Is that enough for speculation? What do you think?

Leave a comment

Posted by on April 30, 2015 in Uncategorized


Tags: , , , , , , ,

Will Open Source Platforms Dominate Enterprise IOT ?

open source

Recently, I’ve been spending a lot of time thinking about the key elements that will be required to make IOT successful in enterprise settings. While IOT is clearly dominating the discussions in technology forums, the ecosystem of enterprise solution is still in very early days. As a result, many of the key elements of enterprise IOT platforms are still to be defined. When we think about those key elements of IOT platforms, we naturally tend to focus on technology. However, lately I’ve becoming convinced that the development and distribution models (open vs. closed source) is going to play an important role in how organizations embraced IOT platforms.

When we think about a way to categorize the different types of IOT platform, we can use a simple model that groups them based on its hosting topology (cloud vs. on-premise) and development-distribution model (closed sources vs. open source).


Following the previous taxonomy, we can clearly see that closed source solutions, whether deployed on-premise or in the cloud, are likely to see more adoption powering vertical-specific solutions in industries such as healthcare, retail, manufacturing, etc.

There are a number of reasons that can contribute to the adoption of closed source IOT solution in this model. For starters, vendors such as IBM or Oracle providing these type of IOT platform have a deep expertise in different verticals and a long history of developing solutions for customers in those industries. However, that’s not the only factor. The closed source nature of those IOT platforms makes it ideal for building solutions using the platform building blocks but almost impossible to extend the platform with new building blocks.

As an alternative to closed source IOT platforms, open source models can see a lot of adoption in the enterprise IOT environment. Below are some of the reasons that might facilitate the adoption of open source IOT platforms in the enterprise:

  • Adapters to custom devices: As IOT continues evolving, enterprises are likely to adopt all sorts of smart devices that go beyond the traditional development toolkits such as Arduino or RasperryPI. An open source platform will provide the right extensibility vehicles for supporting heterogenous devices in enterprise settings.
  • Integration with Enterprise Systems: Integration with corporate systems will be a fundamental component of enterprise IOT platforms. An open source model would allow enterprises and the developer community to implement new adapters to line of business systems so that they can be leveraged in an IOT solution.
  • Custom Protocols: Despite the increasing adoption of standard transport and encoding protocols within IOT device manufacturers, it is safe to expect some IOT vendors to create custom encoding formats or transport protocols that optimize their devices for certain conditions. An open source IOT platform will allow the vendors to incorporate their custom protocols, encoding mechanisms and other  elements into the platform so that they can be used in enterprise settings.
  • On-premise Hosting: Even though the cloud is destined to play a super important role in the adoption of enterprise IOT solutions, there will be many scenarios in which enterprises would prefer to deploy on-premise instances of an IOT platform. In those settings, open source models will offer the right levels of customization to support diverse enterprise environments.
  • New Foundational Capabilities: As enterprises reinvent themselves by embracing smart devices in industrial settings, the number of requirements for an enterprise IOT platform is destined to grow exponentially. An open source model will allow enterprises to customize and extend the IOT platform with new foundational capabilities that are required in their scenarios. As more and more foundational capabilities get added to the platform, other enterprises can benefit from the collective knowledge of the platform
  • Customer Confidence: Like any other emerging trend in enterprise software, the IOT industry is likely to produce a new group of startups that provide platforms for enterprise IOT deployments. Large enterprises typically express concerns when working with new startup platforms. Open source models typically help to alleviate those concerns as enterprises can help to maintain and evolve the code base if the platform vendor goes out of business.

These are just some of the reasons I think might propel open source enterprise IOT platforms as one of the most dominant models in the space. As I said before, I don’t have any clear answers but hopefully the ideas explained in this post will trigger some interesting debates.

Leave a comment

Posted by on April 24, 2015 in Uncategorized


Tags: , , ,


Get every new post delivered to your Inbox.

Join 65 other followers