Issue link: https://insights.oneneck.com/i/1437687
11 Scalability The scalability of architecture is an important factor when evaluating project viability. An architect will need to understand the starting size options for the different alternatives; this loops back to the entry point topic that was just covered. Will the alternative easily allow the design to start at a smaller size as required, or will an organization need to purchase more infrastructure than would be needed to satisfy a project's starting size— and not be able to utilize all the resources until the project grows into it? Aside from how small of a scale the alternative can start with, it is equally important to consider how large the alternative can scale to. If the desire is to start at 500 and still be able to scale to 10,000 users, what will the alternative look like at both ends of that spectrum? Will the organization be happy with the low or high points—or both? The scalability topic is not just a storage discussion. It also holds true for compute, networking, and possibly other layers within the design. If adjustments are made to the configuration of the compute layer to achieve a smaller density of VMs per host server, how might this affect different design choices when scaling? An example would be if the initial host design starts with 128GB of memory per host and the final choice is 256GB or larger, one will need to ensure that the right size DIMMs are used to allow for the configuration to be scaled in the future. If the wrong choices are made to save costs, it will affect the density due to constraints, or cost more in the long run with DIMMs that were unable to be reused. The architect should focus on how the solution will be able to start small, as well being able to scale to the largest point. But one cannot ignore all the points in between either, because depending on how one scales the deployment, there could be many scaling points in between the start and finish. It is ideal to look for something that is going to allow the design to easily scale in buckets of user counts that the project identifies, while not outpacing the deployment timeline and capabilities. The ideal scaling bucket size for a project may be in increments of 500-1000 users. But if the architecture alternative chosen scales greater than this, understand how this affects the costs and deployment.