Interview with Giuseppe Paternò: OpenStack for business
OpenStack is an open source cloud platform but what about its application in the real business?
How does it work with security and development? We have interviewed Giuseppe Paternò for the answer.
According to Forrest Reserch,Giuseppe Paternò aka “Gippa” is among the thirty most important OpenStack’s consultants in the world. He is indeed an IT Architect and Security Expert with an interesting background in the open source field. He is currently one of the candidates for the OpenStack’s board and he is one of the best open source influencers.
Giuseppe Paternò works for the most famous italian and foreign companies in the telecommunications, government and finance industries.
On the occasion of the elections for the 2016 Board of Directors that will be held on January 9, our Marco Cristofanilli, Seeweb’s SystemAdmin and open source and OpenStack expert, had the opportunity to ask some questions to Giuseppe Paternò.
Oh, that’s a long one 🙂 I started having a quick look at OpenStack with the Cactus release, but it’s only with Diablo in 2011 that things got serious. By the time I was working for Canonical and Ubuntu 10.04 LTS was the first distro shipping a “cloud platform”: it was based on Eucalyptus. On customers’ projects, Eucalyptus showed its limits and we started embracing OpenStack as the next generation cloud platform for Ubuntu. Coming from the field and knowing most of the engineering people behind it, I started contributing with some patches around the compute and networking code that was a bit unstable. That was just the beginning of a long road in the OpenStack world ….
Tell us about your vision of Openstack. What about Openstack’s strengths and eventual weaknesses and which are the topics you leverage with your customers?
Probably OpenStack’s best strength is to provide a common and standardized set of APIs to consume services. No matter what flavour of hypervisor, storage or network you choose, you can manage the lifecycle of the whole IT with the same tools.
OpenStack has certainly some weakness. In my opinion, too much “hype” on the project and too many vendors in the board have lead OpenStack in creating too many “as a service” projects to attract more customers rather than consolidating the core, that should be more compact and stable. It is certainly interesting to have projects such as monitoring as a service (Monasca), but I’d rather prefer little functionality that is rock-solid. I’d love to see an evolution similar to what happened to Linux… it took more than a decade to bring Linux into the enterprises.
That doesn’t mean I’m not interested in providing some layered services on-top. As a matter of fact, I would say that the future is application-centric. I dream of an OpenStack where the customer can connect, upload and spawn applications. All he/she needs to set is the SLAs: availability, expected performances, compliance (disaster recovery & security), etc… Would be it through containers? It might well be, as the great advantage of using containers in production is managing application lifecycle (upgrades/downgrades/distribution).
While initially was a strength to speed up the development of OpenStack, getting out a new release every six month is now becoming a concern. An OpenStack infrastructure, be it simple or complex, can’t be upgraded every six months. Enterprise customers want stability rather than features.
From an CIO perspective, OpenStack is not just a technology, is way more that that: is an “enabler” that helps re-inventing the way IT has been conceived so far. Ranging from automatic disaster recovery and business continuity to autoscaling, common metering tools and no more pains in capacity planning. You need a load balancer or a new network? Just instantiate one…. you don’t need to wait for someone to buy specific hardware, install and configure. And you can measure exactly what the internal department or customer is consuming. Quickly responding to business needs, saving license costs and optimization of the budget are certainly the most asked questions from CIOs.
Around the Openstack’s project there is surely a lot of “hype” and the vendors are riding the wave of the moment. Since you have a lot of international customers, what is the status of its usage in production? Is there a field (public, private or hybrid) in which Openstack is mostly used?
I’ll be straight and honest: despite all the success stories that every vendor is publishing, it’s only in 2015 that customers started embracing OpenStack seriously. Up to last year, there were only PoC or limited deployments, in general no serious production… with some exceptions, of course.
Probably the best targets for OpenStack are service providers and telecom operators. It’s undoubted that VMWare is still market leader, but things are changing. Most regional providers are afraid of walking away from VMWare as they have a brand they want to bring to customers. That said, they’re squeezed by license costs and margins are low. Storage is also a big issue for them, as customers are increasing the demand, and many service providers are starting offering both VMWare and OpenStack.
Telecom operators are quite a different story: even if the budget is lower and lower every year, license costs are still not a big issue. With the beginning of Network Function Virtualization (NFV), i.e. the virtualization of mobile networks appliances (SMS Gateways, HLRs, SBCs,…), OpenStack is an enabler that helps reducing costs and time to market. Just imagine to set-up a brand new MVNO with some clicks or an ansible script.
Outsourcers are niche players that also are taking advantages of OpenStack. While they work on private customers and through private links, on a daily basis the behaviour is very similar to regional service providers.
Probably the most interesting stories are the unpublished ones coming from other sources such as finance, betting and oil&gas. These are the ones that also can take advantage of OpenStack on hybrid: if you architect things in the proper way, you can have a disaster recovery, business continuity or full active-active with regional service providers, thus eliminating the needs of an in-house secondary datacentre. Having a common set of APIs and standards is the key to success.
Does Openstack have concrete possibilities to become shortly a standard commonly accepted by vendors? There will be the possibility to eliminate the vendor lock-in?
Infrastructure vendors such as HP, Dell, IBM and NetApp have already understood that OpenStack is the new standard and they have to cope with it. As a matter of facts, most of them have created or are creating plugins for their hardware. On the other side, ISVs are well behind: most business applications are not certified yet for OpenStack.
While technically the OpenStack project itself was meant to provide no lock-in, the truth is that commercial OpenStack vendors are creating a lot of limitations. And it’s not technical… they just claim they don’t certify some combinations, locking the customer in another way. Carefully selecting the right vendor for the right customer is my job and I need to ensure that my customers have the right balance for what they need, and not what some vendor wish to push them.
Seeweb has followed the Openstack’s project for many years, and our company is really enthusiastic about it. To see that over 6000 people coming from 60 different countries meet on the occasion of the Summit for listening, talking and planning about a technology is a relevant signal. Like it or not, OpenStack is becoming (or it has already become) the bridge language of cloud computing. Do you agree with our vision?
I can’t agree more. With the exception of the big ones (Amazon WS and Microsoft Azure) Openstack has already become the de-facto standard project for cloud computing. RackSpace and Dreamhosts are just an example of large providers embracing this standard. I do expect that everybody, sooner or later, will embrace OpenStack.
We have spoke on the advantages of OpenStack for the end customers, but it is underestimated that OpenStack is also a great advantage for software manufacturers and system integrators. With a single set of APIs, the aforementioned companies can focus more on their business rather than “chasing” each vendor’s willing.
Why did you choose to run for the OpenStack’s board of directors election and what kind of contribution would you like to offer to the community?
It all started as a challenge from customers and friends that works into the OpenStack engineering, which wanted to be represented in the community. The board is half proposed by platinum/gold sponsors and half elected from independent candidates. If you have a quick look at the independent candidate list, most of them are somehow related to big brands, so you will end up having the OpenStack foundation being ruled by vendors only.
I’m not questioning here about how they run the foundation, but it’s undoubted that they do their own interests. What about the interests of customers, service providers, ISVs and integrators? This is exactly the contribution I’d like to bring to the board: a link to the real needs of the field.
Although the board is focused more on the financials, it is also responsible for the overall strategy. I’ve got two main goals: the first is having a compact, stable and ultra rock-solid core (Nova/Neutron/Cinder/Swift), while having an application-centric view on the very long run. The second one is introducing the concept of Long Term Support (LTS) releases, following the Ubuntu approach: while releasing every 6 months is ok for the development and testing environments, having a stable release every 2-3 years can give enterprise customers the peace of mind they need while running production environments.