Posted: February 8th, 2016

Would you be satisfied with a service that had 99% availability? What if that was the level of availability from your cable TV provider?

Answer question A below before figuring out the numbers associated with question B.

A. Would you be satisfied with a service that had 99% availability? What if that was the level of availability from your cable TV provider? Does 99% sound good? What if it was from your electric company? In theory 99% sounds pretty good. Let me know your thoughts on this first before proceeding to question B.

B. Assume each month has 30 days. Determing how many minutes in a month and determine how much down time in minutes is 99% availability. (Therefore, what is 1% of the total minutes available in a month.) Figure out the same thing for 99% availability for the year. How much down time in minutes is that annually to still be able to achieve 99% availability. (For fun also figure out down minutes monthly and annually for 99.9% availability and 99.99% availability.) These are often used as targets for system designs.

2. For tuning, do some further research like Vinay did and provide some information on the financial risks of not doing proper performance tuning and the financial impact that can have on a company due to lost business. What type of performance issues are customers not willing to tolerate?

3. For capacity planning, this weekend is a great weeked for addressing scalability of a system, especially for a high usage event. What if your company decided to advertise in the commercials for the SuperBowl? Could your network and servers handle the traffic? Are there alternatives for a temporary solution since you would only need to hanlde that type of volume for a limited time and not year round? (The reason I ask this question is often advertising in the Superbowl generates more immediate traffic than most companies can handle. Over a decade ago Ford advertised a new web site where you could go in on line and configure their new pick-up truck. Within moments after the advertisement aired their servers crashed and were unreachable due to the volume of traffic. They were not prepared for the volume.) Does it really make financial sense to build out the system to handle that type of volume of hits if it is only expected to happen for a relatively short amount of time (hours or days) when the investment of infrastructure will will likely leave your with vast amounts of excess capacity if you invest to fully support that traffic all the time. What are your options?