2021: Multi-Cloud and Edge Predictions
Stating that 2020 has been challenging feels like an understatement. Nobody anticipated COVID and it has certainly taken its toll. No one really knows how long we will continue to work from home, shop primarily online, and continue to avoid travel. What does remain clear though is that the ‘gold rush’ to the cloud will continue as enterprises realize that they must become more agile and responsive than ever before.
At the same time, the 2020 experience has also exposed the limitations of larger enterprises’ ability to speed up their digital transformation and cloud-native and public cloud transformation. Many companies are now realizing that this move will in fact be a long journey – meaning that they will need to run a hybrid cloud environment for at least the short term. Having said that, many enterprises still find it hard to achieve synergy between their hybrid and multi-cloud environments as noted in 2020 IDG® report:
55% of organizations use two or more public clouds, but 79% struggle to achieve synergy across their multiple platforms.
This multi-cloud complexity is expected to grow even more with the adoption of more distributed applications such as edge/IOT and machine learning as noted in a recent Flexera report:
2021 should see new approaches emerge to simplify this new reality in the following areas:
From Private Cloud as a Silo to Everything ‘as Code’
Many enterprise on-prem environments run on VMware, most have been built out of layers of tools and homegrown infrastructure developed over years to meet regulation and organization standards. This makes hybrid-cloud management of on-prem and public cloud environments very different, synergy between environments is hard to achieve – as noted above in a report from IDG.
2021 should see organizations investing more in modernizing on-prem environments, to a point where they can be managed using similar practices and tools that are used for public cloud environments. To achieve this, organizations will need to adopt more open source platforms that are in fact platform agnostic, such as Kubernetes and Ansible along with other open alternatives such as VRA and VRO so that they will not find themselves locked into a single VMware based solution. This will allow organizations to abstract on-prem resources, including homegrown frameworks with an intent based modeling language suitable to be managed as code and directly via CI/CD. This will give organizations better synergy in the way they manage tiered on-prem and public cloud environments as described in the slide below:
Using Multi Kubernetes Clusters to Simplify Multi-Cloud Management
According to the 2019 CNCF SURVEY, most organizations are using more than one Kubernetes cluster.
The support of Kubernetes by all major public cloud vendors and the availability of open-source Kubernetes distribution for private cloud environments can provide a common workload management platform across public and private cloud infrastructure.
In addition, public cloud providers have added official support for hybrid Kubernetes cluster deployments. Google with Anthos and AWS recently announced that they are opening EKS which in combination with Outpost provides an end-to-end Kubernetes management platform.
Using Kubernetes on-prem will allow organizations to move applications to cloud-native architecture while running on-prem- maintaining access to existing services and infrastructure. Once applications run as cloud-native services it will be easier to move them to public cloud as a next step. This phase should involve a longer transformation to transition applications using native cloud services as back-end infrastructure.
This in turn will lead to a reality where enterprises find themselves managing many Kubernetes clusters separated between on-prem, public-cloud environments as well as segmented by location, department and so on.
My prediction for 2021 will see a new focus on managing multiple Kubernetes clusters. I also expect to see independent tools that will focus solely on policy based workload management across different Kubernetes clusters. Multi-cloud could well see a divide between management tools built as an extension of a specific Kubernetes platform – as in the case of Google Anthos, and tools that will be platform agnostic, as in the case of Cloudify.
Edge to Become a Private Case for a Multi Cloud Use Case?
Edge can be seen as a remote extension of the cloud infrastructure towards a specific geographical location – often used to provide more responsive application services.
Traditionally these remote devices are managed through a completely separate stack, often provided by the edge platform provider itself.
Many modern edge platforms such as StarlingX are now based on Kubernetes. Public cloud vendors are also offering different flavors of edge platforms in addition to their on-prem cloud offering (Azure Stack and AWS Outpost). As Forrester notes via CRN.com:
In 2021, we will see new business models emerge that facilitate the deployment of edge, efforts by cloud platforms to compete, and AI and 5G facilitating the expansion of edge use cases
This will surely open up more ‘developer friendly’ opportunities to deliver a new class of AI and 5G applications. As noted in CRN 2021 prediction:
By integrating the network edge into their cloud strategy, developers have the ability to easily deploy services at the edge without having to be concerned with the operational overhead of managing more infrastructure, .. With integrated development and deployment pipelines, developers can move application services and functions from the cloud into network edge locations. This will help create more responsive and dynamic applications.
Cloud and Telcos: A New Romance
Carriers are taking the ‘if you can’t beat them, join them’ approach when it comes to public cloud and are now cautiously forming strategic partnerships with public cloud vendors, knowing that by doing so they are also taking a risk of cannibalizing their own business.
Public cloud providers see telcos as their next growth opportunity and are aiming to take their large datacenter footprint towards their infrastructure. This leads to a very interesting tech and growth dynamic – on the business side this allows carriers to effectively become a local reseller of public cloud services.
The move to a public cloud based infrastructure will also force carriers to take a more pragmatic strategy to the way they currently manage their infrastructure.
For carriers to be successful in this transformation they would have to examine their current standards (ETSI) and their heavy RFI based approach with a more agile alternative that is used by many cloud based businesses such as Netflix. Without changing these processes they are limiting their choice of technology and vendors and will never really be able to be competitive.
2021 should also see a public cloud disruption that will force carriers to take a much more pragmatic and innovative approach to the way they run infrastructure. As with any cloud disruption we’ll also see a new ‘born-in-cloud’ startup offering 5G network services for enterprises, similar to the way we have seen startups lead the SD-WAN wave.
Public cloud providers should also simplify the transformation of carriers in 2021 by taking a hybrid approach that can integrate public cloud offerings with existing carrier backend services – this way they are more likely to win telco opportunities. We have seen the start of this in 2020 with the acquisition MetaSwitch by Microsoft and interesting moves by AWS and Google are widely expected in 2021.
Final Notes: The Need for a New ‘Automation’ Paradigm Shift
2021 should very well see incremental improvements to make multi-cloud and edge simpler and more developer-friendly via the combination of managing everything ‘as code’ together with the move to Kubernetes and cloud-native architecture.
All this is a step in the right direction but will still remain complex as the demand to support more distributed workloads such as ML and IOT , multi-Kubernetes clusters still leaves a high level of complexity in the management and automation of such systems.
If we examine the current automation methods that have been used by almost any orchestration and automation tool we can see they are all centered around the creation of some sort of automation template. What we can also see is that each automation becomes more specialized toward a certain workload – leaving us with many fragmented automation silos.
To simplify this environment it is important to think of a new way in which we can bring these different automation tools together to break the silos. We also need to reduce the time it takes to develop automation templates – this can be done by allowing better re-use of services that have already been templatized. A bigger step would be to find a way to auto generate these automation templates.
As 2021 unrolls and the world slowly regains some kind of normality, we will be covering all mentioned automation and orchestration behaviors – so watch this space – and in the meantime, I wish you all the best for a happy and healthy festive period!
Nati Shalom, CTO and Founder, Cloudify