Cloud & COVID: understanding the impact of the last year
Cloud and the pandemic
As many know we are living in a much different time than we were just a year ago. Lifestyles have changed, habits have changed, businesses have changed. From the technology perspective one thing that has not changed has been the focus on cloud computing. In fact, with the pandemic, cloud has taken on an even greater focus. Let’s look at where we are at with cloud computing one year into a global pandemic.
For many when the pandemic hit, cloud hosted resources became a lifeline. From the sense of being able to quickly provision resources and services without having to worry about procurement, setup, validation, etc. prior to execution, it worked. Organizations across the world were able to quickly, though not always easily, bring online resources for remote access. Whether it was VPN resources to allow remote workers into the network or VDI to allow users to use their home PC to have a corporate like experience, it was able to be accomplished in a short time frame. There was no shortage of challenges because getting it done can be a monstrous task under ideal conditions. These new resources have to be integrated into the existing computing structure of the organization and tested to ensure they work and work well. However, that is something that would have happened even if they were done on-premises. There would have been issues to deal with and configurations to update regardless so from that perspective, the cloud did exactly what it is advertised to do.
But let’s take a step back and look at cloud usage and where we still have room to learn. As much as many cloud providers would like you to believe, the cloud didn’t nor hasn’t solved all the computing problems organizations have.
One of the items that still is frustrating to many is the line that gets thrown around so much: “just put it in the cloud there is no end to how much we can spin up.” Where that thinking came from is still a mystery, but it sure looks good on a marketing slide. In many cases, relatively, it is true. The major cloud providers have an amazing amount of computing power at the ready for organizations to consume. But like all other businesses, the cloud providers are a business too. They cannot afford to have endless resources at the ready. This was demonstrated at the beginning of the pandemic. Many organizations were forced to provision resources in a different computing region than they desired because there wasn’t enough free capacity in the desired region to handle their request. Many cloud providers required organizations to power down their non-production development environments to free up resources for production workloads to be added. Cloud providers obviously can provide larger ability to scale to an organization than the organization can provide itself, but when every organization everywhere needs to scale at the levels needed last year, that’s a challenge even for public cloud vendors.
To illustrate let’s look at an example I was a part of to understand some of the challenges. We were looking at the possibility of expanding a VDI environment out to the cloud to address some scaling issues. Like most the VDI environments, it had grown beyond its projected scale because of the higher number of remote users. Scaling on-premises was looked at but not ideal for a number of reasons. Those included the fact that the ability to get infrastructure into the data center did not exist even if desired. Lead times from infrastructure manufacturers grew heavily during the pandemic, as one would expect, because of staffing difficulties and logistics of moving equipment. In addition, nobody knew if the increased demand was going to stay or for how long. Nobody wants to sink a large amount of time and money into infrastructure that might go idle in three months (if only it would have only lasted three months). So, the conversation was how to handle moving and provisioning a large number of VDI in the cloud. One of the cloud architects asked the question about the scale needed. But it’s the cloud, right? We just provision what we want and when right? Not when you might need to bring on 2000+ VDI simultaneously in the same region apparently. Even in the cloud one has to think about capacity constraints and serviceability.
COVID aside, there are some lessons to learn in regard to cloud. COVID though certainly highlight them. Let’s dig into those lessons.
If all an organization is doing is infrastructure as a service, that really isn’t cloud in my opinion. IaaS is an easy way to get comfortable with cloud concepts because it takes something that is well established in organizations and applies cloud models around it. But at the core, it really doesn’t address most organizational concerns nor should it. When you think about it when a company takes its on-prem infrastructure and ports it to IaaS, it is really no different than when organizations first started virtualizing physical servers in the early 2000s. Someone is still managing the operating system, security policy, system access, backup etc. just as though it were in one’s own datacenter. Only difference is that now it is on hardware in someone else’s datacenter. There are some use cases where this makes sense, but too often it is seen as the end state; IaaS as an end state at enterprise scale is something that is still challenging. Changes to the operational model still need to happen in order to make it viable long term.
Not all clouds are the same. Most organizations I’ve encountered have chosen a preferred cloud vendor, which makes a lot of sense at first. Getting staff trained and ready on one cloud provider is a significant undertaking so having focus is beneficial. However, some cloud providers have better services than others, so organizations need to plan to adapt to that. Either working within the constraints of their preferred provider or understanding where it makes sense to diverge from the standard. Also, within a cloud provider, not all services are available everywhere. So, while a company might want to host all their services in the US East region, they may need to adapt if a particular service isn’t available there. Understanding that there now might be latency, or they may have to move resources. Certainly, cloud providers work to make services available in as many or all regions possible, but again they are a business and if the demand isn’t there the service might not be either. Having a plan to address differences in cloud providers and being able to cross to different providers is a level of maturity organizations should strive after.
Cloud providers are not just yours. Many organizations used cloud providers as part of their business continuity strategies, but only accounted for a disaster within their business. Now multiply that by every organization having an event simultaneously and that changes the calculations. Most cloud providers can probably handle some of their customer base bringing large numbers of resources online, but not all of them. How will that impact planning for larger events like hurricanes that can impact large areas simultaneously? I imagine many organizations are updating their business continuity plans and cloud providers are planning ways to handle wider level events. This will be something to watch and consider.
What is the cloud?
So, what is the cloud and how can it best be leveraged during and post COVID? Cloud is not solely a hosting solution; it is an operational and consumption solution. The benefits come from the cloud in being able to fill gaps that one’s organization is unable to fill. Does it make sense for a tourism organization plan their infrastructure for the summer travel surge when the rest of the year it is significantly less and more predictable? When that organization can consume someone else’s resources for a period of time and plan their operations on having that resource available that becomes a better solution. But that takes a large amount of coordination to know what makes sense to keep on-prem and when and how to bring peak capacity online. But it is much better solution than putting everything in the cloud just because.
IT organizations could do well to look at the utility industry that has a similar operational paradigm. During the coldest parts of winter or the hottest parts of the summer, additional electrical production needs to come online to handle people using their furnaces and air conditioners more. In order to do so there are production facilities that are made to be brought online quickly, but not run 24x7x365. The power produced from those facilities is not as efficient or cheap, but it isn’t designed to be. It’s designed to be brought online quickly and handle surges. Cloud can, and should, be thought of in the same manner. Bring the resources online to fill short falls, pay the premium for the short term, but don’t pay for it as planned daily operations.
The other component of the operational and consumption paradigm is filling gaps in talent within the organization. For some organizations it may not make sense to have dedicated data scientists, or they may not be able to afford them. This is where focusing on what the organization can provide and provide well and using cloud for non-core services makes sense. If that service becomes more highly leveraged then it can be re-evaluated whether it can be brought in house more efficiently, but now the organization can get started with a solution without having everything up front. Consider it like a loan on solutions and resources.
Organizations should constantly evaluate what they are doing in the cloud and whether it makes sense to continue to operate that way. The benefit for cloud is not just being able to expand quickly, but also in being able to contract quickly. If a company deems that they have a stable service that will continue to exist for some period, now it may make sense to repatriate it on-premises. Now a capital investment can be justified. Think of it like moving to a new city for a job. At first you might not buy a house because you want to understand the area, find the best place to live among other considerations. Renting an apartment makes sense then to bridge that gap. Once you have the full scope of information, and know where you want to stay long term, then you can look at buying a house as it becomes more financially beneficial the longer you stay.
Wrapping it all up
One of the better analogies for the cloud came from a recent podcast from Douglas Brown. The business is a car on the highway going from point A to point B. The car needs gas to keep moving; the driver needs food and rest stops to help make the journey tolerable. The cloud are those services along that highway. They don’t directly get the car from point A to point B, but they help make the trip easier. Now the driver doesn’t have to bring extra gas cans along with them; they can stop at a gas station. The driver doesn’t need to bring a full lunch and dinner; they can pull off and stop at a restaurant and have more selection of what they want to eat.
Cloud is going to continue to evolve and our comfort and maturity with it will too. One thing that needs to happen is more need to understand that the cloud is not a destination, but it is a tool. Getting to the cloud should not be an end state for a company, but something it can leverage to move things quicker or more efficiently.
Let me know what you think and your experiences with the cloud both before and during COVID. Shoot me an email at Benjamin.firstname.lastname@example.org and let’s discuss!