The cloud model is upending industry norms and changing the relationship between carriers and vendors. As such, Fran Heeran, who left the vendor world to head up Vodafone's NFV efforts, has a simple message for the vendors: move faster!
Heeran, the group head of network virtualization, SDN and NFV at Vodafone, says he's not seeing the aggression he'd like to from some vendors in bringing a truly cloud-native, microservice-based offering to them. Instead, carriers are having to compensate for vendors that are moving too slowly to transition to a cloud-native environment.
Heeran, who made the case for cloudification in part one of this Q&A, today explains why vendors need to more quickly embrace the cloud to stay in the game. (See Vodafone's NFV Boss: Term 'NFV' Holds Industry Back .)
Carol Wilson: You have been pretty vocal in expressing your frustration with vendors, and you came from the vendor community, but you've expressed concern that moving beyond virtualization to cloudification isn't happening fast enough. Are you seeing any more progress in that area?
Fran Heeran: I think there is frustration all around; I think we are not where we should be.
We started this process five to six years ago -- the vendors looked at the carriers and said "you guys are holding us back." That is simply not the case. We are looking to the vendors to say, "there's a path to cloud nativeness that is something we have spoken about very openly and very strongly for the last number of years."
There is a bit of disparity. I have to be fair about people's origins. There are vendors that started much later in this industry that are software-only vendors and they generally have products that are much newer and were built to be cloud-native from the start. They have the benefit of no legacy they were starting from scratch.
There are definitely some vendors out there who are definitely pretty well along the path and have a reasonably good proposition in terms of true cloud-nativeness, microservices and containerization. But they have no hardware or significant established installed base.
Then there are the other vendors who have been around longer -- to be fair, their software is a bit more established -- and it has been a bit of a longer journey for them because they have to take this robust, well-tested software and services and then take it through this transition to cloud. What I would say is it's still not really fast. We are still not seeing the aggression from some of our vendors in bringing us a truly cloud-native, truly microservice-based offering. And as a result we are having to compensate for that.
So if you look at how we built our platforms, in a cloud-native world, the role or responsibility for resiliency rests with the application and being able to handle failures and so on, you know, it's the pets versus cattle discussion.
But since we're not there yet, we are having to ask the platforms to build in the resiliency while we are on this journey, and it's a big order pushing very hard on platform vendors for that and then VNF vendors who are not getting to cloud-native fast enough. They tell an interesting story, but it is definitely not where it needs to be.
We and others have been pretty vocal in the industry on the need to speed that process up. I think what will happen, as much as we'd like to say, anyone selling to us on Monday morning, needs to be cloud-native, we all know that we can't be that prescriptive in the sense saying you are in or you are out because that is just the nature of the industry, legacy versus the new players.
I think we have a situation now where economics will be used as a forcing function. I spoke at an ONAP or Linux Foundation gathering and I talked about this. Even in our early tests within Vodafone we are seeing significant benefits in true cloud-nativeness, purely from a resource-consumption perspective. So the same application that has been fully containerized and microservice-based will consume less resources than one still relying on virtual machines, because there is a minimum baseline. And that translates into cost of ownership.
That means that someone bringing a virtualized application based on VMs probably will demand more resources from our cloud than somebody bringing a finely tuned cloud-native application. And that is taking into consideration when we do our evaluation and costing.
It is important one for vendors to understand because if your application is not as efficient running in your customer's cloud, that is going to be a significant detractor when it comes to the selection process and how we engage with them.
Our plan probably has to have a small number of choices, because it can be virtual; you can be truly cloud-native and if you choose the one that is not optimal, that will impact your competitiveness in the market in general.
Next page: The next big step toward cloudification
CW: What do you see as the next big step toward cloudification?
FH: I think we as the industry and the service providers probably have to provide more of a forcing function on this. Again, we are in an industry where if a vendor was bringing their VoLTE solution or SD-WAN solution, they would typically bring the whole stack -- hardware, software, features. There was quite a variance in terms of how the different vendors come at this.
In cloud that model doesn't work anymore -- we have to get more prescriptive on what a cloud looks like and then say these are the choices you have and that's it. You have to fit in this environment or you can't play. We need to decide as Vodafone, how many choices should we give them?
There are operational costs involved in the number of different choices we have, so I think what you will see happening across the board, not just with us but with others in the industry is a greater degree of prescriptiveness on the environment and a much more consistent environment. And that will be a function of economics and then getting vendors to move much more quickly to cloud-nativeness. One challenge we have is that if all of our clouds look differently -- that will be a vendor challenge. So we have to make sure that where there is no differentiation to be had, let's make the environment as consistent as possible, that would be good for all of us.
But that environment represents a very aggressive move toward true cloud and true cloud nativeness. So we have a job to do for the vendors and they have a job to do as well.
CW: Are you finding that the open source groups such as ONAP are helping in that process?
FH: I think they are. I think ONAP is. Parts of ONAP are certainly helping in the process.
ONAP needs to be pragmatic and manage the same set of challenges we all do which will be a mixture of VMs, containers and whatever comes after containers.
They are supporting all of it -- it is definitely helping to demonstrate the efficiencies of one over the other, which I think will be a realization for everybody. For example, if we can see and touch and see the differences up close, I think that will help contribute toward this.
It's not a forcing function -- I think ONAP's role is to bring consistency to the overall environment and how we orchestrate and manage versus mandate necessarily true cloud nativeness. That may come.
ONAP needs to be relevant to today's challenges -- managing that hybrid environment. It is definitely helping. I think the activity we are seeing around Kubernetes, Docker, Rancher -- as they mature we are definitely seeing the developers becoming more confident that that is the platform to aim for.
This is classic in our industry. We hear from different vendors about containers inside VMs, VMs inside containers -- you get the classic engineering debates, it doesn't help our cause. We need to be clear about a small number of environments and this is the direction we need to be going in. This is what cloud nativeness is, I think.
This is what cloud-nativeness is -- no secret, well-defined by the industry.
Here's the challenge: the move for a vendor to being full cloud-native in their portfolio doesn't actually represent a new revenue stream. But it represents the ability to stay in the game.
You generally find in those situations from an R&D funding perspective -- I have to go there to stay relevant -- do I put all my resources there or do I keep adding features overall to that legacy?
That is the debate happening at every vendor that has legacy to be worried about. I think they need to look again at the investments and get there faster. Or else the choices will be limited when it comes to selling. Because the carriers are now defining the environment they sell into, they'll need to get there faster than we see them going.
— Carol Wilson, Editor-at-Large, Light Reading