From this Q, as need a cost effective solution ,which option most suitable ? i felt it should be scheduled for monday morning .
Based on past statistics of our web traffic, we observe that we sometimes get traffic spikes on Monday morning. What is the most cost-effective type of scaling should we use for this scenario?
It is not clear which Question you are referring to, however that need not stop us from thinking about the issues in play. 😉
My approach will be to separate cost from comparing the suitability of the options at least initially. Cost effective is a test that needs to be applied after you find solutions that meet the business need. A cheap solution that does not meet the business need is not cost effective, it is a waste of money. 😀
Spikes at a regular interval can be met by pre-staging resources by scheduling, or be reactively staging resources when needed.
Pre-staging (scheduled) is excellent if the behaviour is highly predictable in terms of when it starts and how much you need. It is also excellent if the ramp up speed (how fast the load comes on) is faster than the ability of your on-demand to create and set up the instances. e.g if you load goes for 1 to 100 instances needed in 10 seconds, a dynamic solution would not be able to accommodate that fast enough. However if it takes 2 hours to go from 1 to 100, then a dynamic solution probably could.
Dynamic solutions are better if the timing of the load varies, or the amount of the load varies.
As a designer you can calculate which will be best by looking at two aspects. The cost to the business of having resources running that you do not need, and the cost to the business of having insufficient resources to keep customers happy.
I would assume that there are some other hints in the question. Like maybe a comment about the rate that load comes on. The word ‘sometimes’ is a problem in the absence of other information. Does it mean ;
1 out of 30 days (in which case scheduling would be poor choice because most days the resource would not be used.), or
29 out of 30 days (in which case scheduling would be a good choice)
Reread the question and see if there are any other hints
Hi Rusty you are right,,… scenario mentiones spikes on SOME monday mornings, NOT ALL… so it has to be dynamic…