●
● Software Engineer & Dev Advocate at Dynatrace
● Writing code & talking about writing code
● I attended a sewing course once
Who Am I?
Slide 3
Slide 3 text
To Do List
Slide 4
Slide 4 text
To Do List v2
Slide 5
Slide 5 text
No content
Slide 6
Slide 6 text
No content
Slide 7
Slide 7 text
No content
Slide 8
Slide 8 text
No content
Slide 9
Slide 9 text
No content
Slide 10
Slide 10 text
No content
Slide 11
Slide 11 text
No content
Slide 12
Slide 12 text
No content
Slide 13
Slide 13 text
No content
Slide 14
Slide 14 text
No content
Slide 15
Slide 15 text
No content
Slide 16
Slide 16 text
No content
Slide 17
Slide 17 text
No content
Slide 18
Slide 18 text
No content
Slide 19
Slide 19 text
No content
Slide 20
Slide 20 text
No content
Slide 21
Slide 21 text
No content
Slide 22
Slide 22 text
Canary
● Testing in production
● Can be lower cost
● Quickly catch on to any
issues
Slide 23
Slide 23 text
Choosing who
gets what
Slide 24
Slide 24 text
No content
Slide 25
Slide 25 text
“Feature flags are a software development
technique that allows teams to enable,
disable or change the behavior of certain
features or code paths in a product or service,
without modifying the source code.”
- From the OpenFeature website
Slide 26
Slide 26 text
To Do List + Clear Completed Feature
Slide 27
Slide 27 text
No content
Slide 28
Slide 28 text
The ENV_VAR swamp
0
Dynamic configuration
1
Dynamic evaluation
2
Operationalized feature-flags
3
FEATURE FLAG MATURITY MODEL
Slide 29
Slide 29 text
Dynamic Evaluation
● Feature flag value is determined dynamically
● Based on contextual information
● Flag evaluation logic is centralized
● Flag evaluation logic is independent of application
Slide 30
Slide 30 text
Dynamic Evaluation
● No more if customer in legacyCustomers
● Reduced blast radius
● Experimentation that doesn’t require developer intervention
● Compliance agility
Slide 31
Slide 31 text
Feature Flag Architecture
Slide 32
Slide 32 text
“is an open specification that provides a
vendor-agnostic, community-driven API for
feature flagging that works with your favorite
feature flag management tool or in-house
solution.”
- From the OpenFeature website
Slide 33
Slide 33 text
Feature Flag Architecture
Slide 34
Slide 34 text
Dynamic Evaluation: evaluation context
Slide 35
Slide 35 text
Dynamic Evaluation: implicit evaluation context
Slide 36
Slide 36 text
OK, so we have the contextual data, but what do we do with
it?
Targeting: “The application of rules, specific user overrides,
or fractional evaluations in feature flag resolution.”
Dynamic Evaluation: How do we use the data?
Slide 37
Slide 37 text
Dynamic Evaluation: How do we use the data?
Slide 38
Slide 38 text
Dynamic Evaluation: How do we use the data?
Slide 39
Slide 39 text
Dynamic Evaluation: How do we use the data?
Slide 40
Slide 40 text
1. flagd is OpenFeature’s cloud-native reference implementation of a
feature-flag provider
a. Written in Go
b. Easily containerized
c. Can source flags from various syncs (files, HTTP endpoints,
Kubernetes CRDs)
2. Flags and targeting defined as JSON and a custom flag evaluation
DSL build on JSONLogic
Dynamic Evaluation: flagd
Slide 41
Slide 41 text
Dynamic Evaluation: flagd
1. “enable-mainframe-access” flag is a simple
boolean flag with true/false variants
2. By default, it returns false (“off” variant)
3. Has a targeting rule that returns “on”
variant if email address supplied in context
ends with “@ingen.com”
4. Let’s check it out in the playground
Slide 42
Slide 42 text
In the context of feature flagging, dynamic evaluation means feature
flags are evaluated during runtime, and use contextual data as the
basis for their resolved flag values.
Dynamic Evaluation
Slide 43
Slide 43 text
The ENV_VAR swamp
0
Dynamic configuration
1
Dynamic evaluation
2
Operationalized feature-flags
3
FEATURE FLAG MATURITY MODEL
Slide 44
Slide 44 text
A collection of APIs and SDKs used to collect
telemetry data in a vendor agnostic way.
Events (Logs)
● A point in time without duration
● Allow for arbitrary data and data types
● Logs are a particular type of event
Slide 47
Slide 47 text
Traces
● Collection of related spans
○ Operation with a duration and timestamp
● Arbitrary data stored as attributes
● Spans linked together in a tree
● Essentially 2 events - Start and End
Time
Spans
Slide 48
Slide 48 text
Metrics
● Numeric data aggregated from a series of events
● Usually original events are dropped
● Usually attributes are more restricted
● Requires keeping state on the client
● Possible to generate later from events or traces
Slide 49
Slide 49 text
Sneaker shop architecture
Slide 50
Slide 50 text
Scenario
Response times
are reasonable
Slide 51
Slide 51 text
Scenario
As load
increases….
response time
become worse.
Slide 52
Slide 52 text
Drilling into a trace
The database is
the culprit.
Slide 53
Slide 53 text
The DB is the bottleneck
Slide 54
Slide 54 text
Let’s add read replicas
Slide 55
Slide 55 text
Here’s the plan
Put access to the read replica
behind a feature flag
1
2
3
4
5
Enable the read replica for a
small number of users
Analyze the impact
Enable the read replica for
everyone
Remove the feature flag
Slide 56
Slide 56 text
Adding the feature flag
Identifies the
feature flag
Controls the
database
connection
Evaluation context
Slide 57
Slide 57 text
Collect telemetry
Monitors flag
evaluations with
OpenTelemetry events
Slide 58
Slide 58 text
Enabling feature
Matching flag key
Enabling for 25% of
sessions
Slide 59
Slide 59 text
Starting the rollout
Uh no! That
shouldn’t happen.
Slide 60
Slide 60 text
That was close…
Back to normal.
Slide 61
Slide 61 text
Failure rate by flag variant
It only failed when
the read replica
was enabled.
It only failed when
the read replica
was enabled.
Slide 62
Slide 62 text
Aggregate error messages
Something is
wrong with node 3
Slide 63
Slide 63 text
Let’s try that again…
Slide 64
Slide 64 text
Phase: Read replica disabled
Slide 65
Slide 65 text
Phase: Read replica enabled for 25%
Slide 66
Slide 66 text
Phase: Read replica enabled for 50%
Slide 67
Slide 67 text
Phase: Read replica enabled for 75%
Slide 68
Slide 68 text
Phase: Read replica enabled
Slide 69
Slide 69 text
Recap
Rolled out an important
performance fix
1
2
3
4
5
Controlled the impact of
unforeseen problems
Validated assumptions
Rolled the new feature out for
all users
Continuously monitor impact