Q&A: Why cloud management is hard

In this article, Alessandro Perilli, General Manager of Cloud Management Strategy at Red Hat takes a look at why cloud management can prove a headache for some businesses.

1. We all agree that there isn’t a single, better, path to approach the cloud transformation journey. According to your experience, what are the most successful strategies that organisations are adopting to blend traditional applications/IT with cloud-native applications?

First of all, it’s critical for IT organisations to realise that a cloud-native architecture is not necessarily a good idea for every workload, application, or service they have in their portfolio. In many cases, there’s no return on the investment in refactoring the code of an existing application to become scale-out, non-persistent, microservices-oriented and so on. So embarking on a multi-year project to convert all your traditional apps into cloud-native apps is a recipe for disaster. If this is true, then we must come to the realisation that the IT environment must be designed and segmented to serve both kind of applications at the same time, leveraging the most appropriate platform for the job. That’s what we call “multi-tier data centre architecture”: bare metal and server virtualisation engines for traditional, scale-up workloads, IaaS and PaaS cloud engines for cloud-native, scale-out workloads. Once you have this kind of architecture in place, you need to be careful not to build management silos for each tier I just mentioned. And so you have to invest in management tools that support all tiers and provide a consistent experience across the board.

2. Many organisations express the need for multiple services, public or on premise, depending on the nature of the workloads/applications. Have you experienced situations where one cloud is a better fit?

I’ve seen it all. Some organisations going all private, then all public, then going back to all private. Other organisations spend a million dollars on a specific technology, just to throw it away after 6 months. And a lot more. At the beginning of what we could call the “cloud era” there was not much knowledge and a lot of uncertainty, both from the end user perspective and from the technology and service provider perspective. We’ve all learned hard lessons along the way, and now it’s safe to say that we all agree that you can’t have a single cloud just yet. Not for a few years to come, at least. So the best that IT organisations can do is to consume the private and public clouds that best serve their business needs. The best vendors can do, accordingly, is to accommodate that diversity, empowering the IT organisation with the tools they need to be cloud service brokers. This is why Red Hat is investing to expand the list of supported public cloud providers in CloudForms, our cloud management platform. Our newest release offers consistent management capabilities for both AWS and Microsoft Azure, with more providers on the way. We already committed to support Google Compute Engine next. Same for private cloud: CloudForms 4 now enables IT organisations to build private clouds on top of Red Hat Enterprise Virtualization and KVM, Microsoft System Center and Hyper-V, VMware vSphere, and OpenStack.

3. Is private cloud going to die? Why do so many private cloud implementations fail?

Maybe private clouds will die one day, just like maybe we’ll stop owning cars one day. But in both cases, the answer is: not in the short-term future. And to continue with my analogy: privately owned cars can break and some are way more expensive or more complex to manage than anticipated, but that doesn’t mean that we stop buying them. There are many reasons why private clouds fail. The most common is having wrong expectations. Organisations look at public clouds and expect to obtain the same level of simplicity, flexibility and scalability at a comparable perceived cost. The reality is that there are a lot of hidden costs in public clouds that are underestimated or even completely ignored. And then there are public clouds that have way less automation than they seem to from the outside. To create a private cloud that is truly competitive with a public cloud you should have data centres designed from the ground up for the job, and operation frameworks developed from the ground up for the job, and IT staff trained from the ground up for the job. Clearly, this rarely happens in large organisations, so a private cloud can’t really compete. The good news is that doesn’t have to. Private and public clouds address different business needs at the moment and both are viable and necessary for specific use cases.

4. Containers are supposed to be the new cool. What’s wrong with VMs?

Containers have been around for over a decade. Around 2006, hardware virtualisation won over OS and app virtualisation because VMs were able to satisfy the market needs of that time better than OS partitions and application layers. But market needs change over time, and now there is an emerging need for speed that didn’t exist 10 years ago. So containers, which are significantly faster than VMs in terms of provisioning, are more interesting than ever. But that doesn’t mean that there’s something wrong with VMs. At the moment, and probably for a very long time, VMs and containers are best for different application architectures and use cases. So, just like for private and public clouds, IT organisations are going to use both at the same time. This is why Red Hat evolved CloudForms to manage both VMs and containers, side by side and consistently, in a private or hybrid cloud.

5. For those that jump on the container bandwagon, is this going to be like when we moved from bare metal machines to server virtualisation, with a whole new set of tools to buy, deploy and master?

Every time a new platforms comes to market, it brings its own management layer. That layer is necessary to operate the basic lifecycle of the workload/applications/services hosted on the new platform. So, yes, it’s going to be similar to when we moved from bare metal to server virtualisation, which was challenging enough in terms of mastering new management tools and adapting some management practices (like capacity planning). The difference is that, this time, there are multiple platforms being introduced at once. From server virtualisation we are moving to both IaaS and PaaS private clouds. From the on-premises data center we are moving to public clouds from two or three different providers. From VMs we are moving to containers. So the IT organisations are overwhelmed by the amount of new tools to learn and are really demanding a single pane of glass that can consistently manage all these platforms in the easiest and most consistent way possible. It’s a massive challenge that Red Hat aims to fix with CloudForms, Ansible and the other management products in our portfolio.

6. What do you think are the challenges in running and managing a hybrid cloud?

As I say, the main challenge is having a consistent management layer across all platforms that compose your hybrid cloud, regardless of the vendor that provides them. The less consistent the management layer is, the more operations you have to repeat by using multiple management consoles, and the higher the complexity and operating cost of a hybrid cloud. And, of course, when I say management, I don’t mean just the connectivity to this or that cloud provider. That’s the absolute minimum. I also mean the intelligence necessary to make smart decisions on how to leverage the different clouds you are connected to. For example, smart hybrid cloud management tools should advise users on the best cloud to host the requested workload according to constraints like cost, technical compatibility, security level, regulatory compliance, and so on. It’s like having the chance to buy from three different supermarkets online. Not only I would love to have a single website that allows me to login, book a delivery slot, and pay only once rather than three times, I would also love to have that website suggesting the best supermarket to buy each item on my shopping list.

7. Why are so few organisations investing in cloud management?

Honestly, I think there’s a lot of disillusionment in cloud management tools because the very first products hitting the market years ago were incredibly complex to deploy, hard to integrate with existing IT environments, not flexible enough to adapt to the different business needs of various kind of organisations, and not bringing return on investment as quickly as customers expected. I saw a lot of organisations implementing cloud management platforms from top players in the market and throwing them away after just 6 months. Which is why Red Hat is shaping CloudForms to offer the opposite experience. That's why we have taken and we continue to take a number of decisions to make our cloud management platform as frictionless and modular as possible. For example, rather than asking customers to deploy 6-12 modules like some vendors do, CloudForms comes as a single virtual appliance that can be deployed in 20 minutes. We are working hard to fight a preconception that cloud management is difficult.

8. What’s best, fully integrated solutions or best of breed?

Of course there’s room for both, but we are seeing a growing number of customers asking for best of breed. Which is why we are shaping the Red Hat management portfolio after a number of design principles that include modularity and multi-vendor support. Our goal is to support multiple server virtualisation, private cloud and public cloud vendors, by providing a solid set of core capabilities that can be augmented, when needed, through plug-ins developed by our partners, which are best in class in their own areas of expertise. We do this today with both CloudForms and Ansible, and we’ll continue to further build our management portfolio after these principles. Fully integrated solutions remain a great option for those customers that start from scratch and want to develop a long term business relationship with a new vendor. We support them as well, with the Red Hat Cloud Infrastructure suite.

9. What happens now that Dell is acquiring EMC and VMware?

I think a lot will happen. In a survey set up in January 2015 and unveiled in November of the same year (so, much earlier than the move was announced), Gartner showed how mergers are a major reason for IT organisations to revisit their vendor adoption strategies. And this is the biggest merger in the history of IT, so I assume that it will take a significant amount of time before the three companies will come to an agreement on how to rationalise their offerings in a single portfolio and start offering it to customers. So I assume that a lot of customers are taking or will take a wait-and-see approach, and will use that time to re-evaluate their approaches to virtualisation, cloud computing, management, storage and every other area where the merging companies are operating. But it’s not just a matter to consider for end user organisations. Technology partners and the whole channel will react to this event so a lot of attention could be redirected and spread across a wide range of existing players and newcomers. Interesting times!

Alessandro Perilli, General Manager of Cloud Management Strategy at Red Hat

Image Credit: Brian A Jackson / Shutterstock