Exclusive Interview With Bob Griswold, Chairman, Open vStorage"We are a start-up in elastic block storage system policy driven, multi-datacenter, multi-cloud with built-in self healing and backup."
By Philippe Nicolas on 2017.01.12
Who is Bob Griswold?
- Chairman since one year of Open vStorage, based in Lochristi, Belgium
- Advisory board member of a steathy start-up in Sacramento, CA to be launched at end of January after closing a financial round
- Member of the board of Regents, Archbishop Mitty High School
- Independent investor/advisor of RMG Advisory Services
- VP sales, client partner at Cisco Aricent
- VP strategy and planning, enterprise group, Hewlett-Packard
- SVP, product line management, Seagate
- VP, segment and architecture strategy, and VP strategic service, Cisco
- President and CEO, Provment
- SVP, Terracotta
- CEO, BEA Systems Stockholm Engineering, AB
- Director, Sun Microsystems
StorageNewsletter.com: Open vStorage is not a newcomer but you’re pretty confidential, could you summarize the genesis and background of the company?
Griswold: About six years ago we started off doing a project for an OEM to build a hyperconverged infrastructure product (a.k.a. Nutanix, Simplivity). In early 2013 we realized that what the future really needs is a storage system that is hyperscale and multi-datacenter. Hence, we took the advanced storage technology that we built for the hyperconverged infrastructure project and started building Open vStorage.
Who are the people behind the company? What are the products or storage technologies they already developed?
The team behind Open vStorage has more than 15 years of experience in the data center world. We have developed many well-known products over the years, such as Qlayer (acquired by Sun in 2009), VirtualBox (acquired by Oracle in 2008), DataCenter Technologies (acquired by Veritas Software in 2005) and more recently Amplidata (acquired by HGST/Western Digital).
How is the company capitalized? VC money, others? And are you looking for additional VC round?
Currently the company is funded by its founders, private investors as well as revenues from current customers. We will be looking for additional investment later in 2017 to support our expansion into the North American market.
What's the company vision?
The vision is to provide enterprises and service providers an elastic block storage system that is on par or better than what the big cloud providers have without giving up complete control of their data. This means an elastic block storage system that is policy driven, multi-datacenter, multi-cloud with built-in self healing and backup. All this needs to be achieved at a cost that is lower than what it costs a customer to rely on the big cloud providers.
What are the challenges to solve?
In order to build the elastic block storage system I just described, we needed to solve a number of challenges. We realized long back that SANs weren’t the answer to what a futuristic block storage system should look like and neither were hyperconverged systems. By and large, all of the challenges we solved are challenges that Amazon Web Services, Windows Azure have solved when building their own multi-datacenter block storage system:
- Object resiliency with block performance and semantics - Object storage is known for its resilience, scaling and low cost. However, it is not for high performance, low latency workloads nor is it meant for providing block semantics. We’ve married both of these concepts in a way where the block storage does have very high performance, very low latency, and provides a block interface but yet provides the economics, resilience and scalability of object storage.
- Storage efficiency - In the era of large data sets, storage efficiency while maintaining reliability is increasingly important. We use erasure codes for blocks (both on flash and HDD), a technique we call NC-ECC, where with an additional 25%-33% overhead our system provides 10x more reliability and availability compared to a three-way replica.
- Hybrid cloud/truly multi-DC - Unlike clumsy replication methods, we are truly a multi-DC and multi-cloud solution where workloads can be moved or cloned across data centers or clouds without having to make full copies. With Open vStorage, a customer can move a running workload to a public cloud without moving their data to the cloud - true hybrid cloud.
- Application-driven elastic storage policies - By abstracting flash, HDD and cloud archives as a single pool and having fine grained intelligent policies for each logical disk, we eliminate the needs for silos of storage as well as the brittle, restrictive limitations of current SAN systems where LUNs, volumes, and RAID levels need to be declared before use and can not be changed on the fly.
- Eliminate conventional backups - The entire backup software industry only exists to solve problems with current storage technologies. As data sets continue to grow exponentially, backing them up is a very hard, labor intensive and error prone process. By snapshotting disks frequently, utilizing user defined retention policies, NC-ECC technology and a multi-DC approach we eliminate the need for conventional backups. Set up correctly, a customer can lose an entire data center without losing any data.
What kind of storage product is it?
It is an elastic block store without compromise - truly bringing the best of the block, object and cloud storage worlds with none of their drawbacks. We are confident enough to say that Open vStorage is the world’s fastest distributed block store that spans across multiple datacenters.
We hear a lot of things about software-defined storage. First is Open vStorage SDS? and second, what is your definition of SDS ?
Open vStorage is an SDS. SDS is an overused and abused word, whereby legacy monolithic software running on a dual controller box (similar to a legacy SAN or NAS) is also categorized as an SDS. Here are what we think are the fundamentals of what an SDS system should do:
- Modularity and distributed deployment - different machines should be able to run a subset of the components of the software given considerations such as resources, security, networking, etc.
- Reliability/resiliency completely in software.
- Leverage new innovations in hardware (e.g. 3D Xpoint, many core processors, etc) without major changes to the platform.
- Allow for mixed hardware/cluster deployments (e.g. different sized drives, differently configured machines).
- A very fast and low overhead data plane and a feature rich and programmable control plane that are clearly demarcated. (the SDN companies have really made this happen - not so much the SDS companies).
Why Open vStorage is so confidential?
We are not confidential, rather we are conservative and hence not well known yet. It takes years to build and stabilize a storage system of the scale we’ve built. The approach taken by us to get to a level of stability is 1) Open source our code and get feedback from the community 2) Focus on our four large current customer engagements that we’ve been working with for the past two years. We believe now, after having more than 15PB of storage in production, we are prepared for a broader adoption of our product.
Could you summarize what is Open vStorage all about?
It is an elastic block storage system that enables enterprises and service providers the ability to build their own storage infrastructure that is on par with AWS and Windows Azure, lower cost than any public cloud provider. With Open vStorage enterprises and service providers get cloud economics and scale with the control, performance and security that a traditional data center provides.
In fact, the company develops a product based on solid technologies, could you elaborate on a few of them?
Some of the solid technologies that we’ve built that really stand out are:
- Distributed, scale out, log structured block driver - think of this as ZFS, but with a scale out model and completely software based resiliency.
- NC-ECC - Our error correction approach that works across nodes, racks and data centers and is fast enough for using on flash/NVMe devices.
- Arakoon - A system like ours stores tremendous amounts of metadata. It is a homegrown triple replicated key value store we’ve built to hold all of this metadata.
What is your link with Open Source?
The core of Open vStorage is open source and licensed under the AGPL v3 license. Monitoring, automation and some additional performance features are only available for paying customers.
What makes Open vStorage unique?
The combination of performance while being able to distribute data across multiple locations with a minimal amount of overhead.
What's your business model?
Like most software companies our business model is a combination of licenses and support.
How do you sell? direct, channel, OEM? What are the partnerships you develop?
Currently Open vStorage is engaged with OEMs/service providers, having 15PB of storage in production expected to grow to 50PB by the end of 2017. In 2017 we will start offering our software to the broader market, through channel partners.
How many customers do you have?
For commercial customers, we currently have four large OEM/SP’s. We have far more users of our open source product.
What is the total capacity you operate? How many sites?
Total capacity for paying customers is currently close to 15PB, expected to grow to 50PB by the end of 2017. This doesn’t take into account the clusters which are pure open source. To give you an idea, last month we saw close to 100 new clusters come online.
Your revenue range?
We typically don’t disclose this type of information.
How is the product priced and licensed?
We package licenses for a production version of the software along with 24/7 proactive monitoring, troubleshooting and support for an annual fee. Alternatively customers can pay for a perpetual license and then pay an annual fee for 24/7 proactive monitoring, troubleshooting and support.
What about the international presence?
We currently have a presence in Belgium (our global HQ) as well as Silicon Valley. We do provide 24/7 comprehensive monitoring and support from Belgium for our customers worldwide.
What are the priorities for 2017 and the future of Open vStorage?
Currently we have four paying customers and a number of open source deployments. Our goal would be to add another 12-15 customers this year and have proof of concepts running with another twenty or so enterprises/service providers. We hope to have more than 75PB in production by end of 2017. Open vStorage will be focused on building the best possible elastic block storage system. Areas that we are excited about is 1) Computing at the edge with the advent of IoT, sensors, machine data - with its modularity, multi-tiering and multi-site capability OvS is in a good position to take advantage of this trend 2) New types of storage class memory that will allow us to dramatically improve the performance levels that we get from our product and allow customers to run large scale performance sensitive workloads on our system.