Open Source Adoption

Nithya Ruff is the Director of the Open Source Strategy Office at SanDisk. The company is becoming increasingly active in the Linux community. I sat down with Nithya at LinuxCon Seattle to learn more about SanDisk and its involvement with open source projects and communities.

Swapnil Bhartiya: How much have things progressed in a year in terms of Linux and open source at SanDisk?

Nithya Ruff: One year of focused Open Source Office effort has been tremendously positive for the company. I really advocate having an open source office because it allows companies to coordinate the efforts across all of the business units, across legal marketing, engineering, strategy, etc.

In fact, there are a few highlights for this year that I can call out: We have had more speakers than we have ever had in external events like LinuxCon, All Things Open, OpenStack, etc. So, it has been positive as it allowed us to show what SanDisk is doing in open source and the commitment that we have to open source externally.

Second, we have really increased the amount of blogging and social media interaction we are doing from an open source perspective, and that really has been positive as well.

And, third, the company continues to be very involved in the Ceph project. We have done a tremendous number of contributions and are in the top 10 list of contributors. We are also working with the Ceph community to get on the board of Ceph, for example, so that we actually contribute our views from customers – our view as a flash vendor – to that project.

We have also been maintaining the SCST SCSI project, and we also open sourced an Android storage library, which allows application developers to use external storage such as SD cards or microSD cards more effectively.

We also recently launched a sub-project in the Linux Foundation, which is Mobile Embedded Linux Storage, so that we can coordinate storage issues across vendors and work with the community on that.

SB: You just mentioned a project you are working on with the Linux Foundation. Can you tell us more about the project and what kind or level of engagement there is with the Foundation and the Linux community?

NR: One of the benefits and one of the great support systems that Linux Foundation provides is advice on how to collaborate across common topics with other people in the industry. It provides this neutral house, neutral umbrella, to collaborate and coordinate activities.

We discovered that most of the kernels tend to be focused on servers and data center type of issues, and we needed a place where we could discuss the kernel changes we needed for supporting mobile and embedded type of applications.

So, we worked with Micron, Intel, Qualcomm, Tuxera, and a number of other developers who develop or support different filesystems as well as eMMC type of protocols. We brought this group together, with the help of the Linux Foundation, and we actually kicked off a series of meetings at the Linux Foundation's Vault Conference.

The intent is to coordinate requirements across all of these vendors as it pertains to implementation of protocols like eMMC. We can then approach the kernel community with a single viewpoint on implementations, so the community is not addressed by multiple vendors with multiple implementations.

I think this is a true sign of maturity of our open source efforts but also of the kind of work the Linux Foundation does in helping vendors to work together.

SB: What kind of collaboration is there within the storage companies in terms of open source, open standard, and Linux?

NR: That's a great question about collaboration across the storage companies. The one example that I mentioned is something that we co-led with Micron. Micron and SanDisk – in other circumstances, we would say – compete, but we knew we had common ways we wanted to implement JEDEC (JEDEC.org a standards body), for example, like eMMC, so we felt that collaboration made a lot of sense.

Then, there were certain layers that we could collaborate on for this consistency of implementation, and there still were opportunities for us to differentiate ourselves in other layers of the stack. We didn't need to do everything separately, because we could create efficiencies and commonalities of standards when we work together. So, that's one big one.

I know there are other storage implementations happening, and we are taking a close look to see how we could work with other vendors under the umbrella of, say, the Linux Foundation. We also became a member of OpenStack, and through OpenStack, we have an opportunity to collaborate with other storage vendors in the Cinder project, for example, or in the Swift project. And we are taking a very close look at those projects as well.

SB: You just mentioned OpenStack. What role is SanDisk – or flash-based storage devices – playing in this whole new, IaaS, PaaS, and container world?

NR: SanDisk is a member of the OpenStack Foundation and we see a lot of promise in vendors and community working together to create common tools and framework for managing private, hybrid, and public clouds. This is the reason we are making sure that flash as a storage element is a part of cloud storage infrastructure and that customers get the full benefit of Flash performance.. The reason being that "storage as a service" is a big component that service providers are offering their customers. Then, storage is a big element of cloud computing and of services companies. Since flash is becoming such a primary storage element in data centers – whether it's fueling cloud or fueling hyperscale or service providers – we needed to make sure that flash was at the table and that open source software was working with flash.

Also, we wanted to make sure that we supported APIs that our customers are asking for, such a Cinder and Swift. You know Ceph is a key OS and filesystem that we support on our product InfiniFlash, which is a massively scalable flash storage system with half a petabyte of storage in a 3U box. Since Ceph is one of the most popular deployed Cinder implementations, we naturally were drawn into OpenStack, and we wanted to show that we had a high-performance flash system that was OpenStack ready.

SB: When we are talking about OpenStack, we are basically talking about enterprise customers. Flash is quite dominant in the consumer space, but what kind of adoption are you seeing in the enterprise segment as compared to the traditional storage system? And, if there is an increase in adoption, what's driving it?

NR: We are definitely seeing an increased adoption of flash in the enterprise. Flash started out more on the server side and more as a tier-0 type of storage to accelerate applications such as high-frequency trading and things that required very low latency and high IOPS. But, as time has moved on, we realize that storage requirements and architectures are changing. Customers want more and more data that is discoverable, accessible from anywhere, anytime on demand, because of mobility, because of social media, and the need to access information all the time.

We are finding that flash is becoming more of a primary storage mechanism, and it's replacing hard drives, especially for performance and access and discoverability, and hard drives are really becoming more for storage retention, archiving. So, we are seeing more flash at scale being deployed – at petabyte scale – used for streaming media, social media access, for big data analysis, and anything where data needs to be accessed and used and discovered.

We are seeing a tremendous growth in that part of storage – both on the performance side and on the capacity on demand side, which is also becoming very important in the enterprise.

SB: You mentioned OpenStack and the Linux Foundation, the two projects or organizations that SanDisk recently joined. Are there any other open source organizations you are working with?

NR: So far, it's been Linux Foundation and OpenStack because it addresses our needs really well. We are always taking a look at other organizations that we need to work with; for example, we have sponsored, spoken at, and attended the Open Compute Summit for the last few years.

SB: SanDisk is becoming part of the open source community – part of the phenomenon where open source is becoming a norm in the enterprise segment, where competitors are working together. So, what is driving this collaboration and adoption?

NR: When I ask customers why open source is so prevalent in the data center, in enterprise, the answers are not surprising at all. And it's not always to do with cost; cost is the least of the issues. The biggest issue is to prevent vendor lock-in; they want choice. They want to be able to know that they can go to any vendor for support for a distribution. Whether it's Red Hat or SUSE or Ubuntu, they know that Linux is the same.

The second issue is that a lot of open source software value, to me, is that it is modifiable – that you have the source code. So, a customer knows that they can change it if they need to; if there is a problem, they can fix it. Most importantly, when I talk to companies like Facebook, they will think it's important for them to take it and then scale it for their environment and make it production ready. Then, they share this code with others. Facebook has open sourced more than 200 projects and they are projects that they have scaled and hardened for their production environment They have scaled it, and they want to share it with others. So, the modifiability is an important characteristic.

Third, frankly, a lot of the innovation in cloud, in containers, in the data center is happening in open source. It's on the leading edge as opposed to proprietary. So, if you want leading-edge solutions, open source is where you really invest and you use.

The fourth reason is the interoperability of open source. The fact that open interfaces, open APIs, allow you to mix and match solutions from different vendors, and that these solutions work together is one of the objectives of projects like OpenStack, for example. The Linux Foundation is doing the same thing – bringing two container companies, for example, together and creating common interfaces, so we can actually work with that interface.

Cost should really be the least of the worries for enterprises when open source delivers innovation, modifiability, lack of vendor lock-in, and open interfaces that really prompt the tremendous adoption of open source in the enterprise.