Fierce’s Multi-Cloud Networking 101 series takes a look into the future, using a series of progressive interviews with experts to help readers better understand the key foundational concepts and market landscape of multi-cloud networking. This emerging field aims to solve the problems associated with linking networks and applications across multiple cloud environments.
This week's interview features Pere Monclus, VP and CTO in the networking and security business unit at VMware. The interview transcript has been edited for length and clarity.
Fierce Telecom (FT): We recently did a Cloud 101 series focused on defining what the cloud is and what it can be used for. But there’s obviously more than one cloud provider. What problems are associated with using more than one cloud?
Pere Monclus (PM): There are multiple problems. The first one is all the access control and ownership of the resources of multiple clouds. And as you have a developer that has to develop in multiple clouds, the question is how do you make sure that his access, username, passwords and so on are kept secure, and when he deploys an application in one cloud versus the other cloud all those things are well understood. So basically the first one would be to manage and control known resources that are in different identity domains.
Then you go deeper. When you’re doing multiple clouds you deploy multiple applications, so next is how do you make sure those applications connect to each other and how do they talk to each other in a secure manner.
And the third part a lot of times is cost management. How do you make sure that now as you deploy things in multiple clouds, how do you understand what you have in the cloud and keep it relatively clean to make sure that you don't go out of control in terms of expenses.
FT: Ok, so what is multi-cloud networking? What does it aim to achieve?
PM: Right now we are living three major transformations. The first one would be what people tend to call “multi-cloud,” which is what we're discussing about deploying applications in multiple clouds and how they interact with each other.
The next transformation would be the notion of work remodeling, the work-from-anywhere event that happened in the world. There's lots of productivity models that before were happening from the employee laptop to the company application. Now those tend to happen from home to the cloud. So basically, multi-cloud may mean how devices connect to multiple clouds to get multiple applications.
And the third transformation is the model transformation. This notion of agile development, continuous delivery and all the aspects of making developers more proactive.
Multi-cloud in a very, very simplistic way would be this notion of running applications on multiple clouds, and how those connect in a secure manner end-to-end. So that could be one definition, but then there is the other definition, which is how do you have users from anywhere connect to any cloud or to the right cloud to fulfill that application request. And the third would be, how do I have enough automation in my development processes to be able to deploy my application seamlessly across clouds. So multi-cloud kind of fits within these three transformations.
FT: How is multi-cloud networking different from a traditional data center interconnect?
PM: In the data center we had over the last 20, 30 years well-defined standards, let's say what is a switch or what is a router, what is an IP address, what is a network address translation, what is DNS. So all these things were well defined in a way that now regardless if I had a vendor, say Cisco in one data center and Juniper in another data center, they would all use and connect through similar constructs.
Now when you go to multi-cloud, the concepts are slightly different. What happens is that the interface and management of the constructs required for this interconnect are cloud proprietary or they are specific to every single cloud. So now what you need is a way to connect them together and have a layer of expertise, which could be humans understanding the different environments, or basically some sort of abstraction that can seamlessly abstract the differences of each cloud.
Then it becomes more complex. Imagine that you have your cell phone that connects from home and needs multiple connectivity, and let's say you are a corporate employee. You have to have consistent security for all the employees, so can you really go from your cell phone to an enterprise application hosted in the cloud, or do you need to go through a PoP where you are going to filter the traffic? And now “multi-cloud” may take a completely different meaning, which is how to create some zero-trust network access to aggregate all the users regardless where they are to access the public cloud.
So, you'll have different definitions and different concepts but going back to your original question on how does it compare to data center interconnect, it would be because of the diversity of technologies and APIs and interfaces to do similar constructs across multiple clouds.
FT: Does multi-cloud networking only work with public clouds or can it also work for private cloud environments?
PM: Recently, there were these two kinds of definitions like “hybrid cloud” versus “multi-cloud.” Hybrid cloud was kind of whatever you have in the data center extended into the cloud to have a consistent technology. But we see that the world has become multi-cloud. So now I think when we talk about multi-cloud, it's any cloud. That could be your private cloud or public cloud, multiple private or public clouds. So the more we go as an industry, multi-cloud by definition is any cloud.
FT: Has Covid accelerated demand for multi-cloud networking at all?
PM: Definitely, COVID has. The access discussions that I was telling you before, the notion of working from anywhere and the implications of that, that has been hugely accelerated because basically, one way or another we all ended up working remotely, right? So that's one aspect.
Now the market of connects between clouds, we still see the notion of spreading apps across multiple clouds and demanding huge bandwidth between multiple clouds. We see it in certain instances, but a lot of times it's just an application hosted in the cloud where the users are on the internet. So that's some demand, but it’s not necessarily hugely accelerated by COVID. COVID was more the access technology to the cloud, not the intercloud. But yeah, we are seeing demand and we are seeing growth, but mostly this COVID phase accelerated the user to cloud.
FT: What is the biggest hurdle to multi-cloud networking today, and how can this be resolved?
PM: I would say the first one is the lack of expertise in the industry in all the clouds. For example, if you're an enterprise that you decide to start with Amazon, and you have your on-prem cloud and then you have one public cloud. You were expert on your private cloud, then you had to learn another public cloud. And when you go to the third cloud or the fourth cloud and so on, now there’s the notion of having to create teams of expertise across multiple clouds. The industry has a shortage of experts on all the public clouds. Now, how do you solve it? Traditionally they work through consultancy and companies that can help you but that's expensive. The other aspect is through technology, and there are companies developing network abstraction layers for multi-cloud connectivity that seamlessly allow you to abstract those clouds.