bit difficult to rely solely on tools to forecast pricing because generally you end up using different machines (additional DNS and other costs) than what you anticipated.
I am curious to know what you folks think about moving 3-5% as a POC and then extrapolating the pricing data to get more concrete values?
Of course this assumes you’ve planned it enough to stick to the same plan when you actually begin to move?
Always allow for variance in costing, e.g 10-25% and you can with relative accuracy determine most usage based costs like ingress, DNS queries, etc. You'll never get 100% accurate, but you can get close if you do your homework. Projection before PoC, then use the PoC to adjust estimates where required. Something else to note is that no planning is ever going to be perfect, however the more you do it the better you'll start getting at doing it. I generally aim to ensure that 80% is planned for in as much detail as possible. The 20% usually ends up being things that are hard to estimate or end up having foreseeable variations.
My sizing guide is always 1 to 1 map of physical to EC2 instance pricing, then add 20% to total cost. Once moved to the cloud, you can always use Cost Optimizer to reduce the cost later on
Обсуждают сегодня