Dr. Munirathnam Srikanth
Founder & CTO, ComputeNext
I grew up in Chennai, India (formerly Madras). My fascination for math and science came from my grandfather who was a school teacher. I studied Mathematics and Computer Science at BITS, Pilani, India and started working in the IT sector.
My interest in modeling complex systems – specifically probabilistic systems – led me to a Masters degree at Waterloo University, Ontario, Canada and subsequently to a PhD in Computer Science at SUNY, Buffalo, New York. My research focused on how context can be modeled to improve information retrieval.
After completing my PhD, I joined Language Computer Corporation to lead their R&D efforts in language understanding, textual semantics and knowledge acquisition. I explored the use of Semantic web technologies to capture and represent information from text to infer knowledge. I moved to Microsoft to work on Knowledge Web, an incubation project under Bing, where the effort was to capture and model structured and unstructured data and satisfy a user’s information needs in the form of answers to their questions or recommending things to do.
Natural language process and text analysis are compute intensive. The search for scalable solutions in those domains led me to cloud computing and to the problems of resource modeling and efficient cloud computing. As CTO of ComputeNext, I am working on modeling computing services and finding ways to satisfy IT users’ computing needs through a Cloud Marketplace.
ComputeNext, recently named a Gartner Cool Vendor in Cloud Service Brokerage, is a Bellevue, WA based cloud technology company that is pioneering the way organizations search, discover, procure and provision cloud infrastructure with its multi-cloud marketplace. As a leader in building federated cloud ecosystems, they enable end-to-end transactions across platform agnostic infrastructure. Sundar Kannan and Munirathnam Srikanth started the company in 2010, based on the belief that a few cloud providers cannot satisfy the world’s computing demands. Their mission is to empower cloud consumers with increased freedom and flexibility in the choice, management and utilization of cloud services.
JM: Based on the belief that a few cloud providers cannot satisfy the world’s computing demands, you are a strong proponent of a federated cloud ecosystem because it provides the customer with a choice of what cloud services they use, as well as greater interoperability. How would you define a federated cloud ecosystem to the audience, and what other benefits does it provide to customers?
MS: Yes, the main aspects of a federated cloud ecosystem are choice and customer empowerment. The ecosystem is defined based on its characteristics of a multitude of providers and platforms, location of service and the interoperability of services. Cloud providers with different cloud platforms offer competing services and customers can search, discover and use cloud services from geographically diverse cloud providers. This also includes cloud service providers that use one or more IaaS services to offer higher-order services to the ecosystem. We have built a cloud marketplace to be the framework that powers a federated cloud ecosystem.
Consumers benefit from the ComputeNext marketplace in many ways. They can compare cloud service products and providers, select the service that meets their needs, provision such services and use them. As a consumer, you can build higher-order services using the marketplace and offer it back as a product for other consumers to discover and use your services. This marketplace is transactional and provides value to all participants. With choice of cloud services and an independent marketplace offering insights about these services, you can find answers to questions like, “How do I choose the right service?” and “How do I manage if it is not right?” As business needs change, the marketplace offers options for you to change your service provider.
JM: In a 2012 GigaOM interview, you said that Amazon Web Services (AWS) does not have a compelling reason to meet the diverse needs of all of its customers because the customers are limited only to AWS as an infrastructure as a service (IaaS) provider. In your opinion, why is having the choice of both applications and IaaS providers such an advantage to the customer?
MS: Choice is always good for businesses to operate without being locked down to a vendor. AWS is a market leader and they provide a variety of cloud services that address a number of use cases. However, different applications have different computing, storage and locality needs that may not be satisfied by AWS. For example, for applications that need fast IO and solid state drive (SSD) storage, CloudSigma may be a better option than AWS. If you are collocated in a Switch data center in Las Vegas, you may find moving data and using compute from cloud providers like ProfitBricks and HP Cloud more economical than AWS. If your data has to reside in Canada, for compliance reasons, AWS is not a choice for you now.
Data gravity has been identified as one of the drivers for the decisions that businesses make on where they deploy their cloud services. The location of data and its consumers dictate the best place to provision compute resources. A business with most of its customers located in France, for example, will be able to efficiently run their services from a local cloud provider like Lunacloud in Paris. As their customer base increases south of France, they may decide to move their services and data to a cloud provider in Toulouse or to a provider in Spain. Cloud services are a means to achieve business efficiency. Cloud Marketplace can satisfy such diverse business needs, while a single provider will find it difficult to satisfy the world’s computing needs.
JM: As an early leader in federated cloud service brokerage, what are some of the biggest obstacles ComputeNext has faced since its inception in 2010, and what are some challenges you are currently facing?
MS: It is a blessing and curse to be one of the first movers in cloud service brokerage, especially as one of the first few true marketplaces. The classic chicken and egg problem of marketplaces was an apparent obstacle for us. Convincing business partners – initially hosting providers and early IaaS providers – to align with this new model of demand generation before we actually had a significant amount of customer traction. Simultaneously building the platform, normalization engine and connectors for all major cloud platforms, while doing business development to bring critical mass on the provider-side and convincing them to be patient before we turn on the demand generation. With the platform in place and demonstrating value to the marketplace, we were able to address all concerns of the provider-side of the marketplace. As of the end of this month, we will have 25+ providers in the marketplace.
The demand side is where we have been putting a lot of our efforts now. To showcase the value of the cloud marketplace and cloud ecosystem, we developed MediaPaaS – a vertical service to offer asset management and media services in the cloud. MediaPaaS brings together solutions from our independent software vendor (ISV) partners and offer their integrated services in a software as a service (SaaS) model for asset management, transcoding, fast file transfer and automatic content verification. Customers can upload their videos, collaborate and share their creative content, and eventually distribute to a content distribution network (CDN) or broadcasters for wider consumption. We are focusing on the challenge of identifying such use cases and bringing together partners that can participate in the ecosystem and provide value to customers through their own and/or integrated services on the cloud marketplace.
JM: I recently interviewed 6fusion Co-Founder and CEO John Cowan about the open marketplace they are creating for IaaS. Do your two businesses exist in harmony, with your marketplace allowing customers to effectively participate in choosing the most efficient cloud resources, and their marketplace allowing businesses to break their cloud capacity into standardized units to be traded, or am I misunderstanding that?
MS: Since the day we started executing on our vision, we didn’t intend to wait around for any standards body to define exemplary models and descriptive metrics for cloud resources. We wanted to take the lead and actually make it happen by unifying the inventory of heterogeneous cloud service providers. With close to 25 cloud service providers in our ecosystem and hundreds of users, we believe we have fine tuned an effective model and are allowing the end-user to drive the standard.
Of course, we are open to making it better and willing to work with partners like 6fusion to see if there are synergies to contribute. For example, both organizations are now part of the IEEE Intercloud Testbed to contribute to such models. The key difference is that we are already monetizing on these efforts, staying away from hypothetical models and instead focusing on end-user needs for an intercloud.
JM: What are the one or two biggest trends you would recommend looking out for in cloud technology in the next year or so?
MS: With cloud computing becoming ubiquitous to power and supporting the Internet of Things, there will be convergence of services and data in the cloud. Enabling a consumer to access the services and data from anywhere would require businesses to look at multi-cloud and federated cloud environments to move the services and/or data and provide value. The current container model of applications – Virtual Machines – is easy to bring up and scale in a provider. However, movement across providers has compatibility issues that one can expect to be addressed in the next year. Different types of container models are being explored in the context of Parallels, OpenShift and Docker.io. Abstracting out the VM image on which an application runs opens up the possibility of the application mobility from one provider to another. This will work well for the federated cloud marketplace where the customers can focus on finding the best places to run their workloads and deploying them based on business needs.