Designing for Operations - The Practice of Cloud System Administration - page 46

2.4 Summary
Services should include features that benefit operations, not just the end users. Features re-
questedbyoperationsstaffareaimedatbuildingastable,reliable,high-performingservice,
whichscaleswellandcanberuninacost-effectivemanner.Eventhoughthesefeaturesare
notdirectly requestedbycustomers,thebetteroperational effectiveness ultimately benefits
the customer.
Operations staff need many features to support day-to-day operations. They also need
full documentation of all the operational processes, failure scenarios, and feature controls.
They need authentication, authorization, and access control mechanisms, as well as rate-
limiting functionality. Operations staff need to be able to enable and disable new features
withtoggles,globallyforroll-outandroll-back,andonaper-userbasisforbetatestingand
premium services.
Services should degrade gracefully when there are problems, rather than become com-
pletely unusable. Services that import data from other sources must allow the operations
staff to apply controls to those data sources, based on their data quality or other criteria.
Systems are easier to maintain when this functionality is designed into them from in-
ception. Operations staff can work with developers to ensure that operational features are
included in a system, particularly if the software is developed in-house.
Exercises
1. Why is design for operations so important?
2. How is automated configuration typically supported?
3. List the important factors for redundancy through replication.
4. Give an example of a partially implemented process in your current environment.
What would you do to fully implement it?
5. Why might you not want to solve an issue by coding the solution yourself?
6. Which type of problems should appear first on your priority list?
7. Which factors can you bring to an outside vendor to get the vendor to take your is-
sue seriously?