As technologists and engineers, are we asking or answering the wrong questions about our new or potential systems too often? These wrong questions fall into different categories, but they all have right questions or, even better, statements that you or your customers can use to replace them. Here are two that have reached pet-peeve levels for me.
“What does the product do?”
When an operations or other business customer asks this question, they focus on the technology and not the business. If you select technologies based on this question, you will end up trying to force your workflow or needs into the mold of the technology. You will probably not meet your company’s strategic goals and you will end up with a failed project or unused system.
Replace this customer question with ones of your own. Ask your customer, “What do you need to do?” and “What can’t you do now with the current system that you want to be able to do?” This should start the requirements gathering process and will lead to a lot more questions, but this will be a good thing. You will end up with a complete understanding of what your customer needs and wants to do with a system. You can then search out technology solutions to their requirements, minimizing project failures and improving engineering alignment with corporate strategy.
“What is the peak ingest/transcode/transfer workload?”
Video server or transcode software vendors most often ask this question. With the increasing emphasis on cloud services, watch out for how cloud-computing vendors ask this question, too. Their intentions are good; they are trying to help you size a system for your needs in a worst-case scenario. With most workloads, you end up buying more capacity than you will use 99% of the time. Think about a system where a vendor asked you the peak workload question. Now think about how often you get to that peak. If your workload is bursty or hard to predict, it probably is not too often. (If you do have a predictable, “we record five things an hour, every hour” workload, then you do probably hit that peak.)
The better way to size a system for a bursty workload is to ask the questions, “What is the average ingest/transcode/transfer workload?” and “How long can I have an ingest/transcode/transfer wait before it starts?” Those two questions start the process of statistically modeling out your workload. You will be able to tell your customers and accountants things like, “With 2 transcode servers, 95% of the spots that we get will transcode immediately. If we add another server, that immediate transcode number goes up to 99%.” They can then make sound, confident business decisions about how to spend their capital budgets, knowing what the trade-offs will be. Remember, these are just the first two questions to ask when beginning capacity planning, but they are good ones to ask.
Your Wrong Questions
These two questions are not the only wrong ones broadcast engineers encounter. What are wrong questions that you encounter frequently? Have you found the right question to turn it into?