The cloud infrastructure space is going through a huge upheaval, as it moves from a largely hardware architecture to a primarily software-based model. This discussion digs into the issues and unearths some areas where there are clear opportunities for new entrepreneurs.
Sramana Mitra: Momchil, let us introduce you to our audience. Tell us about who you are. What are you doing?
Momchil Michailov: Thanks for having us. This is Momchil Michailov and I’m one of the co-founders of Sanbolic Inc. I happen to be in Boston today, but we’re both a North American and a European company. A lot of our development is done in Bulgaria in former Eastern Europe. The rest of the company is here in Boston.
Sramana Mitra: Are you normally based in Boston?
Momchil Michailov: When I’m not travelling, I’m in Boston. I spend a lot of time travelling both across EMEA and North America. I also also spend a lot of time in Bulgaria with the development.
Sramana Mitra: Tell us what Sanbolic is all about. What problems are you solving? What trends are you aligning with?
Momchil Michailov: We started Sanbolic almost 15 years ago. At that time, we worked with Motorola and Nortel on designing the next generation Baby Bells Network Operating Centers. The requirements had a lot to do with infrastructure elasticity, scale-out, and ultimately workload scale-out with a very broad spectrum of capabilities. It required a very sophisticated architecture that spanned from storage, networking, through compute that ultimately resulted in application availability and scale-out. That today is your public cloud provider. It took us 14 years to build the platform.
Our focus today is on bringing a software platform that delivers the scale-out and agility of what cloud providers use as a hyperscale infrastructure to the enterprises. There is a significant gap in the infrastructure development both in terms of where server architecture is and the virtualization layer available at the servers, and where networking virtualization is and where storage is. Server and manufacturers operate within the 10% margins. Networking and high-end components operate in about 20% profit margins. Storage vendors operate in the 56% to 57% profit margins, yet they use the exact same components that other vendors use. We found that customers are looking to change that architecture and reduce their dependency to that proprietary storage architecture. Cloud providers including Facebook and Google have proven that hyperscale architecture, where you use commodity and industry-standard components coupled with an agile system through software, is actually a viable model both in terms of its enterprise capability and economics. Our focus is to bring that to enterprise customers on premise and to customers who are not necessarily ready to move entirely to the public cloud but want to have a similar architecture and infrastructure cost.
Sramana Mitra: Let’s take some of your customers and do a few use cases that will help us deeply understand what you do.
Momchil Michailov: VMware started with the trend of software-defined data center. We basically enable customers to create this elasticity and scale-out capability in the storage layer. That’s the first piece. What’s interesting in that piece is that you can create that scale-out as well as layer all these advance software features on top of existing infrastructure. Our platform allows customers to migrate a lot closer towards a full software-defined data center without ripping and replacing their infrastructure. They can layer our platform on top of EMC’s storage or any storage that they currently use. We have customers that are running Cisco, Dell, and IBM infrastructure.
Sramana Mitra: Let me just make sure that I understand this. Mostly, your customers then are the data center vendor?
Momchil Michailov: Our customers are end-users that run their data centers. We’re not an OEM. We actually work with either end users like Verizon Wireless and Commonwealth Bank of Australia. These are customers that run very sophisticated large-scale data centers and we help them migrate their architecture towards the software-defined model instead of the monolithic storage they have today. We also work very closely with some integrators. That is part of the data center transition from proprietary storage appliances into a more distributed industry-standard architecture.
The usual pitch of a lot of established and legacy vendors is, “You don’t need this complexity. You don’t need an army to keep this thing alive. Google, Facebook, and eBay who are using this type of hyperscale architecture have an army of engineers keeping this thing alive. We bundle it free and give it to you as an appliance.” The reality is the connectivity in the server and storage layer at this point is very sophisticated. You can run these systems on-premise without an army of people. However, for customers that have smaller IT departments, we work very closely with system integrators that would enable this type of architecture. Most of our customers are either end users with larger data centers and/or system integrators that help end users.