Heaven is Not a Cloud

Heaven is Not a Cloud
AWS Michigan
April 2018

1. Heaven is Not a Cloud Chris Short SJ Technologies AWS Michigan
2. whoami @ChrisShort devopsish.com
3. Getting Started What is “the cloud” and is it worth moving to? ➔ Definition wtf is cloud? ➔ Build vs. Buy Which is cheaper better? ➔ Timeline How much time do you have? @ChrisShort devopsish.com @ChrisShort devopsish.com
4. NIST Definition: "cloud computing is a model for enabling ubiquitous, convenient, on-demand network access to a shared pool of configurable computing resources (e.g., networks, servers, storage, applications and services) that can be rapidly provisioned and released with minimal management effort or service provider interaction." @ChrisShort devopsish.com
5. @ChrisShort devopsish.com
6. Slightly Easier Definition: On demand compute, storage, network, etc. optimized to work with high levels of redundancy and availability while being maintained and consumed by its users via API @ChrisShort devopsish.com
7. @ChrisShort devopsish.com
8. Assumption #1 Migrating to the cloud saves money! ➔ If done right, yes. If done poorly, it'll cost more ➔ Optimize for What Matters Do not optimize for anything that isn't important ➔ Discounts, discounts, discounts Plan and execute how to get discounts ➔ Beware the asterisks They mean more than just a footnote @ChrisShort devopsish.com
10. @ChrisShort devopsish.com
11. @ChrisShort devopsish.com
12. * Those little f#%^&!s will kill you @ChrisShort devopsish.com
13. * Server costs based on 3 Year Prepaid Reserved Instances and 9% discount (No Upfront RI’s and/or On Demand have increased costs) @ChrisShort devopsish.com
14. Assumption #2 Not running your own infrastructure reduces your bottom line! ➔ CAPEX and OPEX They still apply in the cloud ➔ Expectation vs. Reality Help a cloud provider predict their CAPEX they pass some savings to you ➔ The Six Ps Proper Planning Prevents Piss Poor Performance @ChrisShort devopsish.com
15. @ChrisShort devopsish.com
16. $160 Billion 2018 Public Cloud Capital Expenditures Set To Top $160bn data-economy.com @ChrisShort devopsish.com
17. @ChrisShort devopsish.com
18. The Six Ps Proper Planning Prevents Piss Poor Performance @ChrisShort devopsish.com
19. Assumption #3 Lift and shift is a legitimate first step towards moving to the cloud! ➔ If done right, yes. If done poorly, it'll cost a LOT more ➔ Lift, shift, optimize Lift, shift, optimize is the new black ➔ Discounts, discounts, discounts Plan and execute how to get discounts @ChrisShort devopsish.com
20. Lift and Shift Legitimate first step towards moving to the cloud! @ChrisShort devopsish.com
21. @ChrisShort devopsish.com
22. Some Clouds Just Want to Watch You Burn Money Plan on discounts Plan on discounts not being automatic Plan on discounts not being automatic and needing some work to actually get @ChrisShort devopsish.com
23. Clean up the mess of cloud you created! Optimize! Optimize! Optimize! Make this part of the lift and shift… Optimization loops have to be tight @ChrisShort devopsish.com
24. Planning is Vital Planning, planning, planning ➔ PDCA Plan, Do, Check and Act ➔ Monitoring Monitor everything! Find holes in monitoring and fill them ➔ Cloud Native Embrace cloud concepts and use them ➔ Business Needs Dedicate to spend optimization and business metrics @ChrisShort devopsish.com
25. Plan, Do, Check, Act PDCA is a concept from Lean. Think of it as your scientific process to iterate your way through your cloud migration. Build the check and act times into your migration schedule. Plan it, do it, check it, act on the data. @ChrisShort devopsish.com
27. @ChrisShort devopsish.com
28. Business Needs That cool new feature comes after the things that pay for people's food and shelter. Business Logic Monitoring Dedicating technical resources to work with Finance team Developer input on system specs is no longer necessary in the cloud @ChrisShort devopsish.com
29. Vendor Lock-In Your multi-cloud dreams will be fraught with roadblocks ➔ Don't Fool Yourself Cloud providers don't want to make jumping ship easy ➔ Eat Your Own Dog Food The best way to avoid lock-in is to not move too far from core functionality ➔ Use Case for Vendor Tooling Augmenting staff; but put it in the backlog @ChrisShort devopsish.com
30. Don't Fool Yourself Microsoft, Amazon, and Google aren’t thinking of ways to make changing providers easy The more reliance on things that are not basic compute, network, and storage the further the vendor lock-in. @ChrisShort devopsish.com
31. * No carbon based life forms harmed in this GIF @ChrisShort devopsish.com
32. When to Use Vendor Tooling... Don't have specialized staff? Welcome to AWS ElastiCache. Put something in the your backlogs!!! @ChrisShort devopsish.com
33. Conclusion They won't let me talk forever Cloud: Great (mostly) Money: Good (mostly); keep more of it Planning: Awesome! Vendor Lock-In: Meh… (but not terrible) @ChrisShort devopsish.com
34. Focus on the Hard Stuff Computers are easy; people are hard. Bridget Kromhout, Principal Cloud Developer Advocate, Microsoft Tech is easy. Meat is hard. Nirmal Mehta, Chief Technologies, Booz Allen Hamilton @ChrisShort devopsish.com