Independence

The past 12 months have been a whirlwind of change for SUSE. Former parent company Micro Focus announced last July that it was spinning off SUSE as an independent company  [1]; since then, the leading European enterprise Linux vendor has been in a flurry of reinvention. In addition to embracing a new emphasis on growth, SUSE is also finding its way through the changes in the Linux space after IBM's acquisition of Red Hat.

SUSE's life as an independent company got off to a roaring start as the annual SUSECON kicked off in Nashville, Tennessee, just days after SUSE legally split from Micro Focus. We used this opportunity, as well as the openSUSE Conference in Nuremberg, Germany, to speak extensively with Dr. Thomas Di Giacomo, President of Engineering, Product, and Innovation for SUSE (Dr. T).

Linux Magazine: The new buzzword is "application delivery"; everything is streamlined into that. What does that mean for your company?

Dr. T: It means that, while, like our customers, we are modernizing the way we do software developments, it's not all or nothing. It's not like we're shifting from waterfall to full DevOps. We still have some SUSE engine that's been done for the last 25 years to raise SUSE Linux Enterprise – that's still the basics of the engine – and then we add some more agility with smaller teams that are moving faster, so we try to balance the stability with the innovation and then combine that in a hybrid type of model.

LM: This new organization of the teams is different. It used to be you had the development team, you had engineering, you had QA, you had documentation – they were organized more by their task and not by the product that they were working on.

Dr. T: That's correct.

LM: So now you have agile teams with the DevOps-style model for each product?

Dr. T: That's correct, exactly. The thing to consider when you do that [is] you designate more cross-product integration today than in the past. Fifteen years ago, open source was Linux and that's it. … Now you have to integrate containers with Linux, storage with Linux, so there's a lot of cross-product things. We also have teams that are working to integrate products to get our SUSE products, but also partner products, to work well together to build solutions.

LM: A lot of companies talk about co-creation – that they have their clients or partners be part of the development teams. How do you see that?

Dr. T: That's a very good point and something we've been doing.

LM: You do it with SAP, of course?

Dr. T: We do it with SAP; we do it with IBM still today, so we started with IBM in the late nineties when IBM was pushing for Linux, and that's where Enterprise Linux distributions came from. We worked with IBM; we worked with companies like Fujitsu. There's a commercial relationship, but there's, as you mention, co-creation. Sometimes it happens directly with the open source communities upstream.

We do a lot of things with many different companies inside the projects, and sometimes it's like joining forces together on specific needs. I think that's where SUSE is pretty good, as well, because we have engineers that we can work [with] directly, sitting in the offices of the partner. We are very flexible with that. We don't try to protect our intellectual property or those kinds of things. We work quite closely with partners.

LM: Let's talk about software-defined infrastructure: Do you believe that's the future for everyone?

Dr. T: Generally speaking, I think software-defined infrastructure [SDI] can be used by anyone for any use case at different levels. If you're a telecoms company doing software-defined networking, it makes a lot of sense. You have to decide based on your business what makes sense for you, but software scales a lot better than hardware, and it's more flexible, as well.

Thinking [of] SDI components that can be used by any type of use case, you just need to make sure that … you pick up the right things, whether it's VMs or containers, and some people find storage so very interesting for companies because they can save a lot of money on storage. If you have a large amount of data, software-defined storage will cost you half the price of non-software-defined storage, for instance. We have customers that have invested in non-software-defined things; they have to monetize their investment in about five years …, so they need to wait and to change more progressively. It depends on the situation of customers.

LM: Do you feel that this technology has a different evolution in different regions of the world?

Dr. T: It's very different country by country. In Japan, for instance, I think they are still … trying to find what's working for them; they're still very mainframe driven. The US is very public cloud focused. In Europe, it's a combination of all of that, so there are different speeds and different adoptions. The good thing is that, in most cases, there is open source somewhere, and companies in all countries are investing more and more in open source-based solutions. It depends also on the type of industry; not only do countries [determine the path], but [so does] industry, because of regulation – probably more regulation in government-type cases then in e-commerce or stuff like that.

LM: Your job is trying to find a balance between being innovative but not too far from the market. The gap must be as small as possible. What are your customers demanding now? What do they want for technology?

Dr. T: That's a very good point because all people in technology want to innovate and [do] new cool things, but their attitude is very different. If you have to do innovation that makes sense for businesses, it should be close enough for them that they can adopt it. If the innovation is too far away from what they are doing, the gap [to adoption] is sometimes too big.

What we try to do is … innovate in places that are adjacent to what customers are doing and make small steps on big bang-types of things, especially for mission-critical workloads. I mean, you cannot just give them a new innovation. Our job at SUSE is to make sure it is stable, as well – it's long term. [Technology is] bought for the next five, 10, 15 years, and it's not just something they play with and [in] six months exchange again.

What we see … are things like containers – obviously something that, more and more, is being looked at by companies. The adoption is not as big as the industry pretends it is, but we are seeing a trend of companies going there, [although] maybe not with their legacy applications. When they build new things, then they will try to use containers: hybrid things out of existing stuff and new things that they can innovate with a little bit more … because they don't have to sustain it, [and] maybe they don't have a lot of customers on the new things. We need to make sure that those two worlds can coexist …, which is, maybe, not "sexy" innovation, but … maintaining the whole thing together …, that's also innovation in a way.

LM: Would you say that software-defined infrastructure is going to be the big theme for the next 10, 15 years?

Dr. T: Yeah, from the infrastructure standpoint and also from the application standpoint. I would couple [SDI] with public cloud, as well, because you can do software-defined infrastructure in your private data center, but you can also benefit from what public clouds are doing and trying to bridge those two things together at the infrastructure layer. Then, also, the application delivery … is going to be important, because even if you have the best SDI infrastructure, if there is no way to deliver, develop, [and] manage the applications and data on top of that in a hybrid way, then you don't really benefit from the SDI. Again, that's why our containers are in between the infrastructure and development. It's missing parts, so just having containers is not completely helpful for developers. They need something more than that because it's still infrastructure related, but it's providing portability across different platforms.

In addition, when people are thinking about serverless and function-as-a-service, which is also SDI in a way, where you have independent functions that can run anywhere, you still have to talk together, and you still need persistent storage, because when the function dies or is not called, you lose the data.

LM: Is it part of your role to think ahead about what comes after that? Do you need to think 10 or 15 years into the future?

Dr. T: Three to five years I would say … and we really try not to do that in isolation, so taking customer feedback, but also with hardware partners, cloud service providers, etc. As an example, we have an optimized Linux kernel for Azure … to take advantage of the Azure infrastructure. We're looking at maybe expanding the .NET support at the Linux level for Azure, as well.

LM: When you say you have a kernel specifically for Azure to optimize their infrastructure, you mean the hardware.

Dr. T: Yeah.

LM: Azure uses a certain hardware base, and you're optimizing for that, so it's so big that it's worth it to optimize the kernel for that particular hardware, and they plan their hardware so far in advance that you know it's worthwhile to do that?

Dr. T: Yes, exactly, and we see a lot of shift of workloads from on-premises to public clouds. In a way, a cloud provider for us is like a hardware vendor from the SUSE perspective, and we want to make sure we optimize Linux for what they are using it for: Linux or Kubernetes or even storage.

LM: Now that you're an independent company, does that free you up technologically to do something different?

Dr. T: We were already making our own decisions on technology before! From a direction standpoint, it's not really changing that, but being independent with a growth investor … frees up investment for acquisition, so we can be faster. Let's say we believe that a particular technology is very interesting for the next five years. Organically it takes time [to develop], because we work on the open source and start by being in the upstream project, and it takes time to get the knowledge. … Now, we could be faster because we could acquire companies already in such a space.

LM: Let's talk about the Community side of things – what has been happening there recently?

Dr. T: Well, to mention a few of the major projects we are involved in, SUSE got certified with the OpenChain Project, an initiative that helps make open source licensing simpler and more consistent [2]. We have also been doing lots of upstream work on important projects such as openATTIC (now integrated into Ceph), STRATOS (open sourced by SUSE and now part of Cloud Foundry), and [OpenStack] Airship.

LM: SUSE has also recently launched a Developer Program?

Dr. T: Yes, we have launched this to address a gap we feel we have at the SUSE level. SLES is well known among admins but is lacking visibility in the developer community. We want to make sure developers are aware of all the great tools available, mostly from the openSUSE community, like Open Build Service and openQA.