#2: A Hybrid Cloud Look Ahead for 2021
In 2015, Amazon.com CTO Werner Vogels said hybrid cloud was not an end-point for enterprises looking to close down their datacentres, but more of a temporary stop as they moved towards a whole shift of their IT infrastructure to the cloud. However, last year’s Re:Invent 2020 was the first time that AWS has formally recognized hybrid cloud and marked a drastic shift in their tone towards the cloud space generally. This was a significant change in the point of view as AWS has always reiterated that all workloads should be moving to the public cloud for the past 10 years and that hybrid cloud is not the end goal for many enterprises and businesses.
So how did we get to the point where AWS has softened their stance on hybrid cloud? There are numerous reasons to look at and we will be diving into them in this post which covers the main point shared by the podcast. I will be summarizing what Brandon Whichard (Offer Manager @DXC Technology and Co-Host of SoftwareDefTalk) talked about in the podcast about the evolution of hybrid cloud, common deployment misconceptions and challenges, and the major areas for improvement in 2021.
Migration or “life and shift” to the public cloud is not as simple as many would think
It is a common public sentiment that public cloud is more cost efficient than running on-premise datacenters. Brandon Whichard has often faced clients where they wanted to move their entire workloads to the cloud without rationalizing how the workloads would look like in the cloud or how much it will cost. This is one of the most common pitfalls amongst the clients he engaged with because the common perception of public cloud is that it will help to reduce current costs of running workloads. Yes, the shift from CAPEX to OPEX is a major feature of the public cloud but there are multiple considerations to look at from the types of services being consumed, refactoring of application codes, DR and so on. This misconception is often fatal; it can actually be more expensive to run workloads in the public cloud compared to running it on-premise.
The cloud makes this assumption that customers are aware of the infrastructure that powers the technology. When users run workloads in the cloud, they don’t usually think about network and WAN costs just to name a few. A user would be more concerned about consuming various services like storage and network and putting them together, wrestling with others over IP addresses but won’t consider the miniscule details that runs the infrastructure. Brandon Whichard went on to share stories of users not being aware of the various nuances of running public cloud. For example, it actually costs money to route network between different availability zones or leaving a GPU on for a week often resulting in enormous costs.
From this onset, it is easy to see why hybrid cloud proliferated. In fact, Andy Jassy has quoted that 5% of global workloads run in the public cloud while the other 95% are ran by enterprises. For companies that are digitally native and have familiarity with public cloud services, it would make sense for them to be all-in public cloud especially if they are focused on microservices and app development using containers. However, that is not the case with enterprises who are ingrained in on-premise infrastructure to begin with. This assumption that a user would know everything about public cloud infrastructure is then tested; there are people who built their entire careers around cloud competencies.
To add on, it is notoriously difficult to move applications deep rooted in existing technologies like VMs to the public clouds. It won’t happen overnight and will take time, resources and effort to make that full transition to the public cloud. Therefore, it is not realistic to assume that users would know everything about the public cloud or could perform quick migrations to all-in on a cloud vendor.
It is not easy to understand all the nuances of the public cloud
Public cloud is a massive giant on its own. To make a simple comparison, let’s take VMware for example. There are people who built their entire career around understanding and being an expert in the fields of VMware technology. Be it in the form of virtualization, storage, network, digital workspace or any other solutions, it takes years and years of time and resources to be experts in the various technologies that VMware has to offer. It is the same case with public cloud technology; it takes so much to be experts in cloud technology that many do not understand.
The AWS console might be one of the most powerful technology to exist today and it is also one of the most complicated. Brandon Whichard went on to mention that the biggest misconception about the console is that many users think handling the technology is manageable as long as they explore the console or they can just hire AWS experts to handle the console for them whether in-house or outsourced. However, the learning curve to handle this console alone is a massive undertaking that many underestimated.
From here, we can see that many are not aware of the learning curve involved when it comes to public cloud. It takes years of practice and learning to be experts in public cloud be it AWS, GCP, Azure and so on. Without understanding the details that goes behind running the infrastructure, it can become a major shock to most users when they realize that running everything on public cloud is a massive undertaking.
The idea about moving workloads between public clouds does not mesh with the difficulty to do so
Whether it is hybrid cloud or multi cloud, the main question often asked is “Can you ensure consistent process?” In the past, there would always be a centralized group that makes decisions for the company but the current norm now is the distributed nature of decision making where marketing would make their own decision, finance their own and so on for other departments. So for a user to be all-in on public cloud, cost management is one of the main reason to shift workloads between public cloud vendors. The biggest misconception about moving workloads between different public cloud vendors is that all public cloud are identical; this is entirely not true at all. So now there is a disconnect between ideal and reality as users will find that it is hard to shift workloads around yet maintain a consistent process with the distributed nature of decision making.
Each public cloud vendor will have their own unique capabilities. This means that there are different capabilities that users can tap on each public cloud hence users may have certain preferences or requirements that each different public cloud vendor can fulfill. The ones that are smart about handling multiple public clouds are those who are concerned with how they can manage these multiple public clouds in a centralized, transparent manner ensuring consistent process and efficient cost management.
To put this into perspective, a user who is all-in on AWS will be thinking about IAM (Identity Access Management) in a different way as compared to another user who is all-in on Azure. There are nuances about running workloads on public clouds that will alter how a user will perform configuration, deployment and maintenance of workloads. So for a user to shift workloads between different cloud providers, it becomes intensely difficult to adjust the perspectives gained from being on the previous cloud provider resulting in complications.
Then the next question is, what is the problem of going all-in on one cloud vendor? The answer is that there is no issue with it but it is all a matter of preference. Again, a cloud provider has many unique features that users may prefer so being all-in on one cloud provider may not suit them. Probably another reason why users want to shift between different cloud providers is to reduce the risk of concentration relying on one vendor alone. Lastly, we have touch on this and that is cost management. Ultimately there is nothing wrong with going all-in on a provider; it is just a matter of preference and perspective.
My take on hybrid cloud moving forward 2021
I do think that with a major cloud player AWS formally recognizing hybrid cloud (and even defining it in their own terms), hybrid cloud will definitely be a popular choice amongst enterprises with existing infrastructure. It is not a matter of if, rather a matter of when. Besides costs, learning curve and shifting workload challenges, I think another major proponent of hybrid cloud proliferating is regulations and security. While some of the countries’ data localization rules have loosened, it still remains a fact that most enterprises are required to store sensitive data within the country itself for the purpose of security and compliance.
Where I do see developments is the leading up of hybrid cloud to multi cloud. Simply put, an organization will probably have multiple cloud vendors in their infrastructure. From the IT perspective, the next concern will be about management of different clouds to ensure consistent operations and consistent infrastructure. As reiterated in the podcast, the smart ones are those who think about centralizing management of these clouds from a single point of view. As more companies edge towards digital transformation, I believe that more will embrace cloud in a hybrid/multi format as IT teams become more proficient with cloud technology over time.
Link to the podcast: https://open.spotify.com/episode/0MMENwtpIBeknT34Mx0wi3?si=LyW1SH9jRMqcsWPdUyxSRg
