Augmented Capacity Planning and Benchmarking
In solution design and infrastructure dimensioning process of IT deployments, capacity planning and benchmarking is always a critical tollgate of the project for a fast growing vertical.
And depending on the growth of the business, frequency of capacity review becomes immune irrespective of the workload. Undoubtfully, failure in capacity assessment can lead both software architect and solution architect towards design board once again, sometime even at half way of the project which most likely results to re-dimensioning of infrastructure one more time. What else solution engineering can do if the business gets a hard wall due to capacity choking?
Performance or response time of each of the microservice or module contributes on overall capacity and for better performance and optimized capacity, software architect plays the best role in optimizing design attributes, starting from technology selection to sketch microservices or to build pieces of codes. However, even though the microservices are being designed and deployed in a perfect way they are planned to, there are other factors which can be potential bottleneck for capacity.
For example, infrastructure orchestrations or backfiring impact from peripheral entities for a network jitter. In these contexts, solution architect naturally got a bigger picture of the territory and can supply productive ingredients to performance and capacity optimization of the deployments.
So, it is important to be able to zoom in and out from pixel level to 10000 feet view and be agile or augmented for performance assessment & capacity benchmarking. For private data center, we suggest to start load test with smaller set of hardware, preferably in testbed to figure out minimum units optimized for applications and gradually figure out big numbers with vertical x-or horizontal scaling of infrastructure.
However, scaling is not always linear and so is capacity. Hereby, it results to a great realistic value if we can do several capacity assessments with several sizes of infrastructure and then connect the dots to visualize the real-world pattern.
Mathematics is fun!
Elasticity or auto-scaling in cloud infrastructure solves the problem of capacity choking to a great extent. However, cloud may not remain cost effective any more if sizes and types of infrastructure selections does not fit software requirement in a great way. It may lead to downtime as well under excessive traffic if the auto-scaling parameters are not properly tuned with augmented capacity assessment, though infrastructure will still remain available. Please remember, it always remains a shared responsibility unless you are not using software from cloud providers as a service.
Technically speaking, if one can draw a picture connecting pixels, there will be no one else who can visualize the photons better. Through agile load test in your CI-CD pipeline, augmented capacity planning can gain mathematical control on bits & pieces of your workload. Irrespective of how much it peaks, you can confidently park service downtime for capacity choking much ahead of time.