Hey everyone, just saw the announcement from Microsoft Ignite about the new "Azure savings plan for compute." I've been trying to wrap my head around Reserved Instances (RIs) for our workloads, and this seems like it might change the game.
From what I understand, the new plan lets you commit to a consistent hourly spend on compute services (VMs, containers, serverless) over a 1 or 3-year term, and you get a discount on any usage above that commitment. This sounds way more flexible than RIs, where you have to commit to a specific VM series and region.
My main question is: when would you use this over RIs? If my workloads are somewhat consistent, are RIs still the better deal? And if my usage is all over the place—like bursty analytics jobs—does this plan make more sense?
I'm also curious about the actual savings numbers. Has anyone done the math yet? For example, with a 3-year RI for a D4s v3 in East US, you get ~72% savings. Does the savings plan get you close to that, or is it a totally different ballpark? I'm trying to model this in Excel but the pricing docs are still a bit fresh.
Here's a super basic example of what I'm trying to compare conceptually:
```sql
-- Hypothetical: Comparing monthly cost
-- RI Scenario: 1-year, single instance commitment
-- Savings Plan: 1-year, $100 hourly commitment
SELECT
'Reserved Instance' as plan_type,
estimated_monthly_cost_ri as cost
UNION ALL
SELECT
'Savings Plan' as plan_type,
estimated_monthly_cost_sp as cost;
```
Would love to hear from anyone who has dug into the fine print or has experience with AWS's similar model (I hear it's like their Savings Plan). Is this finally the tool that makes committed spend less of a headache for those of us without perfect crystal balls?