Datadobi
  • Home

Exclusive Interview With Kiran Bhageshpur, CEO, Igneous

Towards subscription-based on-prem cloud storage model
By Philippe Nicolas on 2017.01.25

AddThis Social Bookmark Button

 

 

 

Who is Kiran Bhageshpur ?

He is:

He was:

  • VP, engineering, Isilon storage division, EMC
  • Senior director, engineering, Isilon
  • Director, enterprise software, Questra (now part of Parametric Technology)
  • Senior engineering manager, Lucent Technologies
  • Engineering manager, Latitude Communications (acquired by Cisco)
  • Started as software engineer at ABB-Systems Controls

Storage Newsletter: Igneous Systems is a growing company, could you summarize the genesis and background of the company? Year founded? Number of employees?
Kiran Bhageshpur: Igneous was founded in October of 2013 and we launched in October 2016. We spent the first three years in R&D, and we’ve got just under 50 people today - still primarily in R&D. The motivation for starting Igneous was a convergence of two major trends  (1) the growth in unstructured data that challenges the scalability of existing infrastructure, and (2) the emergence of IT-as-as-service. When we started Igneous, we combined the systems expertise of our founders with cloud expertise that we brought on board from our Seattle ('Cloud Capital') ecosystem.

You co-founded Igneous. Who are the other people behind the company? What are the storage products already developed by the team before Igneous?
Prior to Igneous, I was the VP of engineering for the Isilon storage division of EMC through its growth to over $1 billion per year in revenue. Prior to EMC, I held several engineering leadership positions through Isilon’s rapid growth from about $100 million per year in 2008. My co-founders were Jeff Hughes and Byron Rakitzis. Jeff is our CTO and was director of engineering at EMC’s Isilon’s division. Jeff was a pre-IPO engineer at Isilon Systems. Byron is our systems architect and was the first employee hired at NetApp, where he was responsible for one-third of the code in the company’s first product. He was integral to the development of NetApp’s Write Anywhere File Layout (WAFL) and more recently worked on their all-flash storage solution. Throughout his career, he has authored or contributed to over 30 patents in storage and file systems.

How is the company capitalized?
Igneous is a venture-backed company. Our investors include Madrona Venture Group, New Enterprise Associates, and Redpoint Ventures. To date, we’ve done a seed round and a series A, totaling $26.7 million raised.

What's the company vision?
It is to provide the advantages of public cloud services inside the enterprise data center. These include having nothing to buy, nothing to manage, and cloud-native services. Our tagline has been True Cloud for Local Data.

What are the challenges to solve?
Managing unstructured data at scale, with cloud economics. When I was at EMC, I got to see first-hand how hyperscale providers simply could no longer utilize traditional storage systems, and they built their own. Given that most organizations don’t have the appetite to build their own storage, we set out to make it simple by offering a service that lives inside their data centers with economics and agility that compare well with the public cloud.

What are the target market segments?
Anyone with a petabyte-scale or higher data that they don’t want to manage anymore. We are targeting organizations who are curating petabyte scales of data and higher. Of course, there are traditional suspects, including high tech, media and entertainment, life sciences, oil and gas, and others who buy a lot of scale-out NAS storage today. That said, we are also seeing growing data sets in other industries, including automotive, geospatial, and manufacturing, where growth in machine learning and artificial intelligence have pushed infrastructure requirements. 

Is Igneous as a software-defined storage player?
Software-defined, yes. SDS player, no. Even though we offer our own purpose built appliance, we utilize a software-defined architecture. With no need for specialty hardware (such as NVRAM) or proprietary interconnects (we use Ethernet and IPv6 internally), the guts of our architecture are software-defined. That said, we don’t consider ourselves an SDS player, as the competitors in this space associate themselves with a "bring your own hardware" approach. Instead, we consider ourselves an enterprise Infrastructure as-a-Service (IaaS) provider. We provide all hardware, software, and service with one price with cloud economics.

So do you sell and promote a service or a product?
As a service.

What are the advantages of your on-prem cloud approach? It’s a mixed of two worlds, right?
Correct. It has the best of both worlds. Like public cloud, you don’t buy equipment, install software, or manage systems. Igneous is pay-as-you-go and zero-touch. Like on-premises systems, you can access data at LAN speeds, secured behind the firewall, with predictable expenses.

Your company develops a product based on solid technologies, could you elaborate on a few of them?
From a technology perspective, Igneous incorporated four key attributes into the Igneous Data Service which it considers as "first principles" to enable cost-effective deployment of an on-premises IaaS offering.

  • On-premises data with cloud management. While the data always stays on the customer’s networks, Igneous continually monitors, maintains and troubleshoots its fleet of inventory using automated, software-based cloud management techniques employed in today’s hyperscale cloud environments.
  • RatioPerfect architecture. Behind the architecture is a nano-server design where each disk has its own dedicated CPU and Ethernet connection.  This design enables a highly distributed architecture that provides the scalability and resiliency of the public cloud but scaled to run on a customer premises.
  • Extensible data path. Large data sets typically require inline processing of data before uploads or downloads "complete."  With its extensible data path, higher level operations (such as auto-tagging for search) can be performed on a mirror of incoming and outgoing data streams without slowing down low-level system functions.
  • Cloud native services. To enable new application workflows, the architecture is built utilizing a modern micro-services approach and incorporates stream processing, an event-driven framework, and container services.

Your solution introduces a new design with a system embedded with each drive, give us some details?
The RatioPerfect architecture utilizes nano-servers to form a highly scalable and resilient storage layer. Each nano-server consists of a drive, an ARM CPU, and redundant Ethernet connections. The nano-servers are densely packed into a standard chassis with redundant power supplies and redundant Ethernet switches. Our dataBox units that support 212TB of usable storage are 4U (7-inch high) and pack 60 nano-servers each, enabling customers to have a very distributed, private storage cloud even with a single dataBox. Of course, this architecture scales horizontally, and multiple dataBox units can be combined seamlessly in a configuration to scale to exabyte capacities.

With such model, how do you protect data? We heard about DEEPR, what is it?
We utilize a variant of Reed-Solomon erasure coding to protect data. As such, the system can tolerate many concurrent failures, without impact data availability or durability. DEEPR stands for distributed erasure encoding with prioritized repair. It is our patented data protection method that utilizes our unique nano-server architecture to optimize repairs while maintaining optimal service availability and performance.

What are the interfaces exposed by the service? Any plan to add others?
Igneous Data Service utilizes a data access layer independent of the underlying storage layer. Today, this data access layer supports S3 API and FTP. Watch for future announcements in this area.

What about file services? Any possibilities or a clear wish to not play in this field?
While we feel there are inherent scalability limitations in POSIX semantics, we do believe there are a number of use cases for file services that are ideally suited for our infrastructure. Watch for future announcements in this area.

What kind of patents do you have? Around which topics?
Today, Igneous has been granted 12 patents. The core of our IP was all developed to enable us to support infrastructure at-scale and as-a-service.

When a file is written to the box, does it belong to one drive only or is it chunked into multiple pieces and sent to multiple drives?
Data is written across many physical drives to improve both resiliency and performance.

Could we replicate one box to the other? Is a cascading mode available? Could we build a scenario with multiple targets in different rooms, datacenters or sites addressed by one front-end ?
Yes. Replication between sites is possible. Each site has its own data access layer, and this data access layer can be used to facilitate communication among sites. As such, a data operation sent to one "front end" or data access layer at one site could trigger replication policies across multiple systems.

The competition?
From a budgetary perspective, we often compete against the status quo of traditional NAS and scale-out NAS systems like NetApp or EMC Isilon. From a functional perspective, we are often compared to public cloud services, such as Amazon S3. From a market fit perspective, we are similar to other on-premises IaaS providers, including Zadara Storage, ClearSky Data, and ZeroStack.

Wha's your link with open source?
We are both a consumer and provider of open source technologies. For example, as a consumer, we utilize Linux as our core operating environment and as a provider, we have contributed a number of projects to extend the use of Golang  or Go Programming Language (our preferred language for development). These are just a couple of examples.

What makes Igneous unique?
Our long-term investment in developing the right IP for delivering infrastructure at-scale and as-a-service.

Your business model?
Subscription-based and as-a-service.

How do you sell?
We allow customers to purchase through IT resellers in the channel. Today, we do not sell through any OEM relationships. We have developed some Independent Software Vendor (ISV) relationships with a number of other application and ecosystem providers, and see these relationships as core to delivering service value to our customers. 

What's the initial or minimal usable capacity of the service a user can buy?
Today, our offering starts at 212TB usable capacity and scales in 212TB increments. 

Price?
Our list pricing starts at under $40,000 per 212TB increment, with an automatic volume discount schedule based on capacity and contract duration.

What are your priorities for 2017 and later?
In 2017, watch for us to build out further services in core data management use cases. Beyond 2017, our plans our to extend our infrastructure beyond storage and data management itself.

 

Read also:
Start-Up Profile: Igneous
Infrastructure-as-a-Service for data you can't or won't move to cloud
by Jean-Jacques Maleval | 2016.10.11 | News

Articles_bottom