Slide 1
Slide 1 text
Holly Cummins
Øredev
November 6, 2024
less waste, more joy,
and a lot more green
how quarkus makes java better
Slide 2
Slide 2 text
@holly-cummins.bsky.social
what
#RedHat
Slide 3
Slide 3 text
@holly-cummins.bsky.social
what
#RedHat
Slide 4
Slide 4 text
quarkus applications start fast
quarkus + graalvm 0.014 Seconds
quarkus + open jdk 0.75 Seconds
traditional cloud-native stack 4.3 Seconds
https://quarkus.io/blog/runtime-performance/
rest application
Slide 5
Slide 5 text
quarkus + GraalVM
13 MB
quarkus + OpenJDK
74 MB
traditional Cloud-Native Stack
140 MB
rest application
https://quarkus.io/blog/runtime-performance/
quarkus improves memory utilization
Slide 6
Slide 6 text
@holly_cummins
machine
quarkus
quarkus
quarkus
quarkus
quarkus
quarkus
quarkus
quarkus
quarkus
quarkus
quarkus
quarkus
quarkus
quarkus
container orchestration
machine
traditional cloud-native java stack
traditional cloud-native java stack
traditional cloud-native java stack
traditional cloud-native java stack
quarkus applications have high deployment density.
quarkus native,
but quarkus on
jvm is also way
smaller than
traditional java
Slide 7
Slide 7 text
@holly_cummins
traditional cloud-native
java stack
traditional cloud-native
java stack
traditional cloud-native
java stack
traditional cloud-native
java stack
node.js
node.js
node.js
node.js
node.js
node.js
node.js go
go
machine
go go go
go go go
go go go
go go go
go go go
go go
go go
quarkus
quarkus
quarkus
quarkus
quarkus
quarkus
quarkus
quarkus
quarkus
quarkus
quarkus
quarkus
quarkus
quarkus
… and not just when comparing to other java frameworks
container orchestration
machine machine machine
https:/
/developers.redhat.com/blog/2017/03/14/java-inside-docker/
Slide 8
Slide 8 text
let’s talk about throughput.
https://www.redhat.com/en/resources/mi-quarkus-lab-validation-idc-analyst-paper
48 concurrent
connections
traditional cloud native stack 3555 req/s
Slide 9
Slide 9 text
let’s talk about throughput.
quarkus native 3212 req/s
https://www.redhat.com/en/resources/mi-quarkus-lab-validation-idc-analyst-paper
48 concurrent
connections
traditional cloud native stack 3555 req/s
Slide 10
Slide 10 text
let’s talk about throughput.
quarkus native 3212 req/s
https://www.redhat.com/en/resources/mi-quarkus-lab-validation-idc-analyst-paper
a trade-off of throughput
against footprint
48 concurrent
connections
traditional cloud native stack 3555 req/s
Slide 11
Slide 11 text
@holly-cummins.bsky.social
native compilation trade-offs
quarkus native 3212 req/s
https://www.redhat.com/en/resources/mi-quarkus-lab-validation-idc-analyst-paper
48 concurrent
connections
traditional cloud native stack 3555 req/s
throughput
startup time +
footprint
Slide 12
Slide 12 text
@holly-cummins.bsky.social
native compilation trade-offs
quarkus native 3212 req/s
https://www.redhat.com/en/resources/mi-quarkus-lab-validation-idc-analyst-paper
48 concurrent
connections
traditional cloud native stack 3555 req/s
throughput
startup time +
footprint
Slide 13
Slide 13 text
but …
traditional cloud native stack 3555 req/s
quarkus native 3212 req/s
https://www.redhat.com/en/resources/mi-quarkus-lab-validation-idc-analyst-paper
a trade-off of throughput
against footprint
48 concurrent
connections
Slide 14
Slide 14 text
but …
traditional cloud native stack 3555 req/s
quarkus on jvm 6389 req/s
quarkus native 3212 req/s
https://www.redhat.com/en/resources/mi-quarkus-lab-validation-idc-analyst-paper
a trade-off of throughput
against footprint
48 concurrent
connections
Slide 15
Slide 15 text
but …
traditional cloud native stack 3555 req/s
quarkus on jvm 6389 req/s
quarkus native 3212 req/s
https://www.redhat.com/en/resources/mi-quarkus-lab-validation-idc-analyst-paper
no trade-off, just better :)
a trade-off of throughput
against footprint
48 concurrent
connections
Slide 16
Slide 16 text
@holly-cummins.bsky.social
we beat the trade-off.
throughput
startup time
+ footprint
Slide 17
Slide 17 text
@holly-cummins.bsky.social
we beat the trade-off.
throughput
startup time
+ footprint
Slide 18
Slide 18 text
@holly-cummins.bsky.social
we beat the trade-off.
throughput
startup time
+ footprint
it’s a double-win.
Slide 19
Slide 19 text
No content
Slide 20
Slide 20 text
@holly_cummins
#RedHat
developer joy
Slide 21
Slide 21 text
@holly_cummins
#RedHat
developer joy
-auto-provision services
Slide 22
Slide 22 text
@holly_cummins
#RedHat
developer joy
-auto-provision services
-live coding
Slide 23
Slide 23 text
@holly_cummins
#RedHat
developer joy
-auto-provision services
-live coding
-continuous (fast!) testing
Slide 24
Slide 24 text
@holly_cummins
#RedHat
developer joy
-auto-provision services
-live coding
-continuous (fast!) testing
-less boilerplate
Slide 25
Slide 25 text
@holly-cummins.bsky.social
how
#RedHat
Slide 26
Slide 26 text
@holly-cummins.bsky.social
Slide 27
Slide 27 text
@holly-cummins.bsky.social
Slide 28
Slide 28 text
@holly-cummins.bsky.social
Slide 29
Slide 29 text
@holly-cummins.bsky.social
profile shave
shave
profile
shave
profile
shave
profile
shave
profile
Slide 30
Slide 30 text
@holly-cummins.bsky.social
profile shave
shave
profile
shave
profile
shave
profile
shave
profile
shave
profile
Slide 31
Slide 31 text
@holly-cummins.bsky.social
performance is a
series of puzzles
Slide 32
Slide 32 text
@holly-cummins.bsky.social
“OAAS: obsession as a service”
- Francesco Nigro,
obsessive performance engineer
Slide 33
Slide 33 text
@holly-cummins.bsky.social
why don’t java apps do well on
techempower benchmark?
example performance puzzle
Slide 34
Slide 34 text
@holly-cummins.bsky.social
sometimes,
the root cause
is not obvious
Slide 35
Slide 35 text
@holly-cummins.bsky.social
sometimes,
the root cause
is not obvious
Slide 36
Slide 36 text
@holly-cummins.bsky.social
clue 1
Slide 37
Slide 37 text
@holly-cummins.bsky.social
we ran hyperparameter optimisation
clue 1
Slide 38
Slide 38 text
@holly-cummins.bsky.social
we ran hyperparameter optimisation
clue 1
it suggested this:
Slide 39
Slide 39 text
@holly-cummins.bsky.social
we ran hyperparameter optimisation
-XX:TieredStopAtLevel=1
clue 1
it suggested this:
Slide 40
Slide 40 text
@holly-cummins.bsky.social
we ran hyperparameter optimisation
-XX:TieredStopAtLevel=1
clue 1
it suggested this:
disables C2 compiler
Slide 41
Slide 41 text
@holly-cummins.bsky.social
we ran hyperparameter optimisation
-XX:TieredStopAtLevel=1
well-known for making
long-running apps slower
clue 1
it suggested this:
disables C2 compiler
Slide 42
Slide 42 text
small number of cores
Netty http handling
clue 2
Slide 43
Slide 43 text
lots of cores
Netty http handling
clue 2
Slide 44
Slide 44 text
what’s going on here?
lots of cores
example: type pollution clue 2
Slide 45
Slide 45 text
@holly-cummins.bsky.social
to find out,
Francesco Nigro and Andrew Haley read
20,000 lines
of ASM
ASM
clue 3
Slide 46
Slide 46 text
@holly-cummins.bsky.social
cache pollution
diagnosis
Slide 47
Slide 47 text
@holly-cummins.bsky.social
“type pollution”
diagnosis
Slide 48
Slide 48 text
@holly_cummins
if (a instanceof Thing)
problematic pattern:
Slide 49
Slide 49 text
@holly_cummins
if (a instanceof Thing)
problematic pattern:
!!
Slide 50
Slide 50 text
@holly_cummins
if (a instanceof Thing)
problematic pattern:
affects : !!
Slide 51
Slide 51 text
@holly_cummins
if (a instanceof Thing)
– Quarkus core
– Netty
– Hibernate ORM
– Hibernate Reactive
– Vert.x
– Smallrye Mutiny
– Smallrye Common
– Vert.x Web
– Infinispan
– Camel
– Drools
– Optaplanner
– Java Class Library
problematic pattern:
affects : !!
Slide 52
Slide 52 text
@holly_cummins
if (a instanceof Thing)
– Quarkus core
– Netty
– Hibernate ORM
– Hibernate Reactive
– Vert.x
– Smallrye Mutiny
– Smallrye Common
– Vert.x Web
– Infinispan
– Camel
– Drools
– Optaplanner
– Java Class Library
problematic pattern:
affects : !!
– Java Class Library!
Slide 53
Slide 53 text
@holly-cummins.bsky.social
avoiding the problematic pattern sped quarkus up a lot
(in many-core case)
Slide 54
Slide 54 text
@holly-cummins.bsky.social #RedHat
private static int extractSize(Object it) {
if (it instanceof Collection) {
return ((Collection>) it).size();
} else if (it instanceof Map) {
return ((Map, ?>) it).size();
} else if (it.getClass().isArray()) {
return Array.getLength(it);
} else if (it instanceof Integer) {
return ((Integer) it);
}
return 10;
}
slow on many-core systems
Slide 55
Slide 55 text
@holly-cummins.bsky.social #RedHat
private static int extractSize(Object it) {
if (it instanceof Collection) {
return ((Collection>) it).size();
} else if (it instanceof Map) {
return ((Map, ?>) it).size();
} else if (it.getClass().isArray()) {
return Array.getLength(it);
} else if (it instanceof Integer) {
return ((Integer) it);
}
return 10;
}
// Note that we intentionally use "instanceof" to test interfaces as the last resort in order to mitigate the "type pollution”
// See https://github.com/RedHatPerf/type-pollution-agent for more information
if (it instanceof AbstractCollection) {
return ((AbstractCollection>) it).size();
} else if (it instanceof AbstractMap) {
return ((AbstractMap, ?>) it).size();
} else if (it instanceof Collection) {
return ((Collection>) it).size();
} else if (it instanceof Map) {
return ((Map, ?>) it).size();
}
return 10;
}
slow on many-core systems
30% more throughput
Slide 56
Slide 56 text
@holly-cummins.bsky.social #RedHat
private static int extractSize(Object it) {
if (it instanceof Collection) {
return ((Collection>) it).size();
} else if (it instanceof Map) {
return ((Map, ?>) it).size();
} else if (it.getClass().isArray()) {
return Array.getLength(it);
} else if (it instanceof Integer) {
return ((Integer) it);
}
return 10;
}
// Note that we intentionally use "instanceof" to test interfaces as the last resort in order to mitigate the "type pollution”
// See https://github.com/RedHatPerf/type-pollution-agent for more information
if (it instanceof AbstractCollection) {
return ((AbstractCollection>) it).size();
} else if (it instanceof AbstractMap) {
return ((AbstractMap, ?>) it).size();
} else if (it instanceof Collection) {
return ((Collection>) it).size();
} else if (it instanceof Map) {
return ((Map, ?>) it).size();
}
return 10;
} but …
Slide 57
Slide 57 text
@holly-cummins.bsky.social #RedHat
private static int extractSize(Object it) {
if (it instanceof Collection) {
return ((Collection>) it).size();
} else if (it instanceof Map) {
return ((Map, ?>) it).size();
} else if (it.getClass().isArray()) {
return Array.getLength(it);
} else if (it instanceof Integer) {
return ((Integer) it);
}
return 10;
}
// Note that we intentionally use "instanceof" to test interfaces as the last resort in order to mitigate the "type pollution”
// See https://github.com/RedHatPerf/type-pollution-agent for more information
if (it instanceof AbstractCollection) {
return ((AbstractCollection>) it).size();
} else if (it instanceof AbstractMap) {
return ((AbstractMap, ?>) it).size();
} else if (it instanceof Collection) {
return ((Collection>) it).size();
} else if (it instanceof Map) {
return ((Map, ?>) it).size();
}
return 10;
} but …
Slide 58
Slide 58 text
@holly-cummins.bsky.social #RedHat
private static int extractSize(Object it) {
if (it instanceof Collection) {
return ((Collection>) it).size();
} else if (it instanceof Map) {
return ((Map, ?>) it).size();
} else if (it.getClass().isArray()) {
return Array.getLength(it);
} else if (it instanceof Integer) {
return ((Integer) it);
}
return 10;
}
// Note that we intentionally use "instanceof" to test interfaces as the last resort in order to mitigate the "type pollution”
// See https://github.com/RedHatPerf/type-pollution-agent for more information
if (it instanceof AbstractCollection) {
return ((AbstractCollection>) it).size();
} else if (it instanceof AbstractMap) {
return ((AbstractMap, ?>) it).size();
} else if (it instanceof Collection) {
return ((Collection>) it).size();
} else if (it instanceof Map) {
return ((Map, ?>) it).size();
}
return 10;
}
not a fan of the fix
but …
Slide 59
Slide 59 text
@holly-cummins.bsky.social #RedHat
private static int extractSize(Object it) {
if (it instanceof Collection) {
return ((Collection>) it).size();
} else if (it instanceof Map) {
return ((Map, ?>) it).size();
} else if (it.getClass().isArray()) {
return Array.getLength(it);
} else if (it instanceof Integer) {
return ((Integer) it);
}
return 10;
}
// Note that we intentionally use "instanceof" to test interfaces as the last resort in order to mitigate the "type pollution”
// See https://github.com/RedHatPerf/type-pollution-agent for more information
if (it instanceof AbstractCollection) {
return ((AbstractCollection>) it).size();
} else if (it instanceof AbstractMap) {
return ((AbstractMap, ?>) it).size();
} else if (it instanceof Collection) {
return ((Collection>) it).size();
} else if (it instanceof Map) {
return ((Map, ?>) it).size();
}
return 10;
}
not a fan of the fix
non-idiomatic
but …
Slide 60
Slide 60 text
@holly-cummins.bsky.social #RedHat
private static int extractSize(Object it) {
if (it instanceof Collection) {
return ((Collection>) it).size();
} else if (it instanceof Map) {
return ((Map, ?>) it).size();
} else if (it.getClass().isArray()) {
return Array.getLength(it);
} else if (it instanceof Integer) {
return ((Integer) it);
}
return 10;
}
// Note that we intentionally use "instanceof" to test interfaces as the last resort in order to mitigate the "type pollution”
// See https://github.com/RedHatPerf/type-pollution-agent for more information
if (it instanceof AbstractCollection) {
return ((AbstractCollection>) it).size();
} else if (it instanceof AbstractMap) {
return ((AbstractMap, ?>) it).size();
} else if (it instanceof Collection) {
return ((Collection>) it).size();
} else if (it instanceof Map) {
return ((Map, ?>) it).size();
}
return 10;
}
not a fan of the fix
non-idiomatic
difficult to maintain
but …
Slide 61
Slide 61 text
@holly-cummins.bsky.social #RedHat
private static int extractSize(Object it) {
if (it instanceof Collection) {
return ((Collection>) it).size();
} else if (it instanceof Map) {
return ((Map, ?>) it).size();
} else if (it.getClass().isArray()) {
return Array.getLength(it);
} else if (it instanceof Integer) {
return ((Integer) it);
}
return 10;
}
// Note that we intentionally use "instanceof" to test interfaces as the last resort in order to mitigate the "type pollution”
// See https://github.com/RedHatPerf/type-pollution-agent for more information
if (it instanceof AbstractCollection) {
return ((AbstractCollection>) it).size();
} else if (it instanceof AbstractMap) {
return ((AbstractMap, ?>) it).size();
} else if (it instanceof Collection) {
return ((Collection>) it).size();
} else if (it instanceof Map) {
return ((Map, ?>) it).size();
}
return 10;
}
not a fan of the fix
non-idiomatic
difficult to maintain
machine
efficiency
my team’s
efficiency
but …
Slide 62
Slide 62 text
@holly-cummins.bsky.social
sometimes, you need to
un-connect the dots
Slide 63
Slide 63 text
@holly-cummins.bsky.social
sometimes, you need to
un-connect the dots
Slide 64
Slide 64 text
@holly-cummins.bsky.social
sometimes, you need to
un-connect the dots
Slide 65
Slide 65 text
@holly-cummins.bsky.social
African wild ass
sometimes, you need to
un-connect the dots
Slide 66
Slide 66 text
@holly-cummins.bsky.social
sometimes, you need to
un-connect the dots
Slide 67
Slide 67 text
@holly-cummins.bsky.social
sometimes, you need to
un-connect the dots
Slide 68
Slide 68 text
@holly-cummins.bsky.social
sometimes, you need to
un-connect the dots
Slide 69
Slide 69 text
@holly-cummins.bsky.social #RedHat
challenge assumptions
Slide 70
Slide 70 text
@holly-cummins.bsky.social #RedHat
string comparison: who says you
have to read strings left-to-right?
challenge assumptions
Slide 71
Slide 71 text
@holly-cummins.bsky.social #RedHat
string comparison: who says you
have to read strings left-to-right?
QUARKUS_REPEATED_PREFIX_FOO
QUARKUS_REPEATED_PREFIX_BAR
challenge assumptions
Slide 72
Slide 72 text
@holly-cummins.bsky.social #RedHat
string comparison: who says you
have to read strings left-to-right?
QUARKUS_REPEATED_PREFIX_FOO
QUARKUS_REPEATED_PREFIX_BAR
challenge assumptions
Slide 73
Slide 73 text
@holly-cummins.bsky.social #RedHat
challenge outdated
architectural ssumptions
Slide 74
Slide 74 text
@holly_cummins
Don’t be dynamic.
Slide 75
Slide 75 text
@holly_cummins
Don’t be dynamic.
(Wait, what??)
Slide 76
Slide 76 text
@holly-cummins.bsky.social
#RedHat
old Java frameworks were
optimised for …
Slide 77
Slide 77 text
@holly-cummins.bsky.social
#RedHat
long-lived processes
old Java frameworks were
optimised for …
Slide 78
Slide 78 text
@holly-cummins.bsky.social
#RedHat
long-lived processes
annual (!) deployments
old Java frameworks were
optimised for …
Slide 79
Slide 79 text
@holly-cummins.bsky.social
#RedHat
long-lived processes
annual (!) deployments
late-binding
old Java frameworks were
optimised for …
Slide 80
Slide 80 text
@holly-cummins.bsky.social
#RedHat
long-lived processes
annual (!) deployments
late-binding
re-configurable without restart
old Java frameworks were
optimised for …
Slide 81
Slide 81 text
@holly-cummins.bsky.social
#RedHat
application frameworks were
optimised for dynamism
Slide 82
Slide 82 text
@holly-cummins.bsky.social
#RedHat
application frameworks were
optimised for dynamism
dynamism has a cost
Slide 83
Slide 83 text
@holly-cummins.bsky.social #RedHat
load code onto a server
then
Slide 84
Slide 84 text
@holly-cummins.bsky.social #RedHat
load code onto a server
then
Slide 85
Slide 85 text
@holly-cummins.bsky.social #RedHat
load code onto a server
then
public static void main(String... args) {
int exitCode = new CommandLine(new
ModuleBuildDurationReport()).execute(args
);
System.exit(exitCode);
}
Slide 86
Slide 86 text
@holly-cummins.bsky.social #RedHat
load code onto a server
then
public static void main(String... args) {
int exitCode = new CommandLine(new
ModuleBuildDurationReport()).execute(args
);
System.exit(exitCode);
}
Slide 87
Slide 87 text
@holly-cummins.bsky.social #RedHat
load code onto a server
then
Slide 88
Slide 88 text
@holly-cummins.bsky.social #RedHat
cloud apps are immutable
now
Slide 89
Slide 89 text
@holly-cummins.bsky.social #RedHat
cloud apps are immutable
now
Slide 90
Slide 90 text
@holly-cummins.bsky.social #RedHat
cloud apps are immutable
now
CI
Slide 91
Slide 91 text
@holly-cummins.bsky.social #RedHat
cloud apps are immutable
now
CI
Slide 92
Slide 92 text
@holly-cummins.bsky.social #RedHat
cloud apps are immutable
now
CI
public static void main(String... args) {
int exitCode = new CommandLine(new
ModuleBuildDurationReport()).execute(args
);
System.exit(exitCode);
}
Slide 93
Slide 93 text
@holly-cummins.bsky.social #RedHat
cloud apps are immutable
now
CI
public static void main(String... args) {
int exitCode = new CommandLine(new
ModuleBuildDurationReport()).execute(args
);
System.exit(exitCode);
}
Slide 94
Slide 94 text
@holly-cummins.bsky.social #RedHat
cloud apps are immutable
now
CI
Slide 95
Slide 95 text
@holly-cummins.bsky.social #RedHat
cloud apps are immutable
now
CI
Slide 96
Slide 96 text
@holly-cummins.bsky.social
#RedHat
a highly dynamic
runtime in a container
is pointless
Slide 97
Slide 97 text
@holly_cummins
Java dynamism
Slide 98
Slide 98 text
@holly_cummins
Java dynamism
build time
Slide 99
Slide 99 text
@holly_cummins
Java dynamism
build time runtime
Slide 100
Slide 100 text
@holly_cummins
Java dynamism
build time runtime
Slide 101
Slide 101 text
@holly_cummins
Java dynamism
packaging
(maven, gradle…)
build time runtime
Slide 102
Slide 102 text
@holly_cummins
Java dynamism
build time runtime
Slide 103
Slide 103 text
@holly_cummins
Java dynamism
build time runtime
Slide 104
Slide 104 text
@holly_cummins
Java dynamism
>
build time runtime
load and parse
• config files
• properties
• yaml
• xml
• etc.
Slide 105
Slide 105 text
@holly_cummins
Java dynamism
>
build time runtime
Slide 106
Slide 106 text
@holly_cummins
Java dynamism
@
@
>
build time runtime
• classpath scanning and
annotation discovery
• attempt to load class to
enable/disable features
Slide 107
Slide 107 text
@holly_cummins
Java dynamism
@
@
>
build time runtime
Slide 108
Slide 108 text
@holly_cummins
Java dynamism
@
@
>
build time runtime
build a metamodel
of the world
Slide 109
Slide 109 text
@holly_cummins
Java dynamism
@
@
>
build time runtime
Slide 110
Slide 110 text
@holly_cummins
Java dynamism
@
@
>
build time runtime
start
• thread pools
• I/O
• etc.
Slide 111
Slide 111 text
@holly_cummins
Java dynamism
@
@
>
build time runtime
ready to
do work!
Slide 112
Slide 112 text
@holly_cummins
what if we start the application more than once?
@
@
>
Slide 113
Slide 113 text
@holly_cummins
what if we start the application more than once?
@
@
>
@
@
>
Slide 114
Slide 114 text
@holly_cummins
what if we start the application more than once?
@
@
>
@
@
>
@
@
>
Slide 115
Slide 115 text
@holly_cummins
what if we start the application more than once?
@
@
>
@
@
>
@
@
>
@
@
>
Slide 116
Slide 116 text
@holly_cummins
what if we start the application more than once?
@
@
>
@
@
>
@
@
>
@
@
>
Slide 117
Slide 117 text
@holly_cummins
what if we start the application more than once?
@
@
>
@
@
>
@
@
>
@
@
>
so much work gets
redone every time
Slide 118
Slide 118 text
@holly-cummins.bsky.social
Hibernate
speed example: JTA auto-wiring
Slide 119
Slide 119 text
@holly-cummins.bsky.social
Hibernate
speed example: JTA auto-wiring
Class.forName(“LikelyJTAImplementation”);
Slide 120
Slide 120 text
@holly-cummins.bsky.social
Hibernate
speed example: JTA auto-wiring
Class.forName(“LikelyJTAImplementation”);
Slide 121
Slide 121 text
@holly-cummins.bsky.social
Hibernate
speed example: JTA auto-wiring
Class.forName(“LikelyJTAImplementation”);
Slide 122
Slide 122 text
@holly-cummins.bsky.social
Hibernate
speed example: JTA auto-wiring
Class.forName(“LikelyJTAImplementation”);
Class.forName(“APossibleJTAImplementation”);
Slide 123
Slide 123 text
@holly-cummins.bsky.social
Hibernate
speed example: JTA auto-wiring
Class.forName(“LikelyJTAImplementation”);
Class.forName(“APossibleJTAImplementation”);
Class.forName(“AnotherJTAImplementation”);
Slide 124
Slide 124 text
@holly-cummins.bsky.social
Hibernate
speed example: JTA auto-wiring
Class.forName(“LikelyJTAImplementation”);
Class.forName(“APossibleJTAImplementation”);
Class.forName(“AnotherJTAImplementation”);
…
Slide 125
Slide 125 text
@holly-cummins.bsky.social
Hibernate
speed example: JTA auto-wiring
Class.forName(“LikelyJTAImplementation”);
Class.forName(“APossibleJTAImplementation”);
Class.forName(“AnotherJTAImplementation”);
Class.forName(“NicheJTAImplementation”);
…
Slide 126
Slide 126 text
@holly-cummins.bsky.social
Hibernate
speed example: JTA auto-wiring
Class.forName(“LikelyJTAImplementation”);
Class.forName(“APossibleJTAImplementation”);
Class.forName(“AnotherJTAImplementation”);
Class.forName(“NicheJTAImplementation”);
Class.forName(“VeryNicheJTAImplementation”);
…
Slide 127
Slide 127 text
@holly-cummins.bsky.social
Hibernate
speed example: JTA auto-wiring
Class.forName(“LikelyJTAImplementation”);
Class.forName(“APossibleJTAImplementation”);
Class.forName(“AnotherJTAImplementation”);
Class.forName(“NicheJTAImplementation”);
Class.forName(“VeryNicheJTAImplementation”);
…
~129
auto-wiring attempts
Slide 128
Slide 128 text
@holly-cummins.bsky.social
Hibernate
speed example: JTA auto-wiring
Class.forName(“LikelyJTAImplementation”);
Class.forName(“APossibleJTAImplementation”);
Class.forName(“AnotherJTAImplementation”);
Class.forName(“NicheJTAImplementation”);
Class.forName(“VeryNicheJTAImplementation”);
…
~129
auto-wiring attempts
every single start.
Slide 129
Slide 129 text
@holly-cummins.bsky.social
it’s not just JTA
this happens for lots of internal service bindings
Slide 130
Slide 130 text
@holly-cummins.bsky.social
JVM
footprint example: Hibernate
Slide 131
Slide 131 text
@holly-cummins.bsky.social
JVM spends time
loading classes for
specific databases
JVM
class for
unused
database
class for
unused
database
class for
unused
database
class for
unused
database
class for
unused
database
class for
unused
database
class for
unused
database
class for
unused
database
class for
unused
database
class for
unused
database
class for
unused
database
class for
unused
database
class for
unused
database
class for
unused
database
class for
unused
database
class for
unused
database
footprint example: Hibernate
Slide 132
Slide 132 text
@holly-cummins.bsky.social
JVM spends time
loading classes for
specific databases
JVM
class for
unused
database
class for
unused
database
class for
unused
database
class for
unused
database
class for
unused
database
class for
unused
database
class for
unused
database
class for
unused
database
class for
unused
database
class for
unused
database
class for
unused
database
class for
unused
database
class for
unused
database
class for
unused
database
class for
unused
database
class for
unused
database
turns out they’re
never used
footprint example: Hibernate
Slide 133
Slide 133 text
@holly-cummins.bsky.social
JVM spends time
loading classes for
specific databases
JVM
turns out they’re
never used
JIT spends time
unloading classes
footprint example: Hibernate
Slide 134
Slide 134 text
@holly-cummins.bsky.social
Hibernate
example:
~500 classes
which are only useful if you're running an Oracle database
loaded and then unloaded
Slide 135
Slide 135 text
@holly-cummins.bsky.social
Hibernate
example:
~500 classes
which are only useful if you're running an Oracle database
loaded and then unloaded
every single start.
Slide 136
Slide 136 text
@holly-cummins.bsky.social
the true cost of loaded classes isn’t just memory + start time
Slide 137
Slide 137 text
@holly-cummins.bsky.social
the true cost of loaded classes isn’t just memory + start time
method dispatching:
Slide 138
Slide 138 text
@holly-cummins.bsky.social
interface
the true cost of loaded classes isn’t just memory + start time
method dispatching:
Slide 139
Slide 139 text
@holly-cummins.bsky.social
unused
implementation
the one we
want
interface
unused
implementation
unused
implementation
the true cost of loaded classes isn’t just memory + start time
method dispatching:
Slide 140
Slide 140 text
@holly-cummins.bsky.social
unused
implementation
the one we
want
interface
unused
implementation
unused
implementation
the true cost of loaded classes isn’t just memory + start time
method dispatching:
Slide 141
Slide 141 text
@holly-cummins.bsky.social
unused
implementation
the one we
want
interface
megamorphic call
slow dispatching
unused
implementation
unused
implementation
the true cost of loaded classes isn’t just memory + start time
method dispatching:
Slide 142
Slide 142 text
@holly-cummins.bsky.social
the true cost of loaded classes isn’t just memory + start time
the one we
want
interface
Slide 143
Slide 143 text
@holly-cummins.bsky.social
the true cost of loaded classes isn’t just memory + start time
the one we
want
monomorphic call
fast dispatching
interface
Slide 144
Slide 144 text
@holly-cummins.bsky.social
how do we fix all this?
Slide 145
Slide 145 text
@holly_cummins
@
@
>
build time runtime
what if we initialize at build time?
Slide 146
Slide 146 text
@holly_cummins
@
@
>
build time runtime
what if we initialize at build time?
Slide 147
Slide 147 text
@holly_cummins
@
@
>
build time runtime
start
• thread pools
• I/O
• etc.
what if we initialize at build time?
Slide 148
Slide 148 text
@holly_cummins
@
@
>
build time runtime
ready to
do work!
start
• thread pools
• I/O
• etc.
what if we initialize at build time?
Slide 149
Slide 149 text
@holly_cummins
@
@
>
repeated starts are now efficient
Slide 150
Slide 150 text
@holly_cummins
@
@
>
repeated starts are now efficient
Slide 151
Slide 151 text
@holly_cummins
@
@
>
repeated starts are now efficient
Slide 152
Slide 152 text
@holly_cummins
@
@
>
repeated starts are now efficient
Slide 153
Slide 153 text
@holly_cummins
@
@
>
repeated starts are now efficient
Slide 154
Slide 154 text
@holly_cummins
@
@
>
repeated starts are now efficient
Slide 155
Slide 155 text
@holly_cummins
@
@
>
repeated starts are now efficient
less wasted work
Slide 156
Slide 156 text
@holly_cummins
the quarkus way enables native compilation
native
(graalvm)
@
@
>
jvm
build time
Slide 157
Slide 157 text
@holly_cummins
the quarkus way enables native compilation
native
(graalvm)
@
@
>
jvm
build time
Slide 158
Slide 158 text
@holly-cummins.bsky.social
Slide 159
Slide 159 text
@holly-cummins.bsky.social
doing more up-front
Slide 160
Slide 160 text
@holly-cummins.bsky.social
doing more up-front
- speeds up start
Slide 161
Slide 161 text
@holly-cummins.bsky.social
doing more up-front
- speeds up start
- shrinks memory footprint
Slide 162
Slide 162 text
@holly-cummins.bsky.social
doing more up-front
- speeds up start
- shrinks memory footprint
- improves throughput (!)
Slide 163
Slide 163 text
@holly-cummins.bsky.social
implementation corollary:
libraries must participate in the build process,
not just the runtime process
Slide 164
Slide 164 text
@holly-cummins.bsky.social
implementation corollary:
libraries must participate in the build process,
not just the runtime process
you need an extensible build process
Slide 165
Slide 165 text
@holly_cummins
have the right plug-points,
so the whole ecosystem
can become faster
Slide 166
Slide 166 text
@holly-cummins.bsky.social
build steps
+
build items
extensible builds
Slide 167
Slide 167 text
@holly-cummins.bsky.social
build steps
+
build items
extensible builds
any extension can make
Slide 168
Slide 168 text
@holly_cummins
#Quarkus #RedHat
Slide 169
Slide 169 text
@holly_cummins
#Quarkus #RedHat
build steps
Slide 170
Slide 170 text
@holly_cummins
#Quarkus #RedHat
Slide 171
Slide 171 text
@holly_cummins
#Quarkus #RedHat
build items
Slide 172
Slide 172 text
@holly_cummins
#Quarkus #RedHat
Slide 173
Slide 173 text
@holly_cummins
#Quarkus #RedHat
framework automatically
determines correct execution
order and injects parameters
Slide 174
Slide 174 text
@holly_cummins
#Quarkus #RedHat
build items are communication
mechanism between build steps
framework automatically
determines correct execution
order and injects parameters
Slide 175
Slide 175 text
No content
Slide 176
Slide 176 text
@holly_cummins
isn’t long compilation kind of
terrible for developers?
@
@
>
Slide 177
Slide 177 text
@holly_cummins
isn’t long compilation kind of
terrible for developers?
not if you have live coding :)
@
@
>
Slide 178
Slide 178 text
@holly_cummins
framework works out required level of reload
live coding
Slide 179
Slide 179 text
@holly_cummins
file reload
live coding
Slide 180
Slide 180 text
@holly_cummins
file reload
live coding
SASS changes detected, will rebuild: [META-INF/
resources/public/stylesheets/live.scss]
Files changed but restart not needed - notified
extensions in: 0.043s
Slide 181
Slide 181 text
@holly_cummins
file reload JVM agent reload
live coding
Slide 182
Slide 182 text
@holly_cummins
file reload JVM agent reload
live coding
Application restart not required, replacing classes via
instrumentation
Live reload performed via instrumentation, no restart
needed, total time: 0.180s
Slide 183
Slide 183 text
@holly_cummins
full restart
file reload JVM agent reload
live coding
Slide 184
Slide 184 text
@holly_cummins
full restart
file reload JVM agent reload
live coding
Restarting quarkus due to changes in
Application$RenardeRequest.class, Application.class,
Application$ApplicationGlobals.class,
Application$Templates.class.
Live reload total time: 1.415s
Slide 185
Slide 185 text
@holly_cummins
full restart
file reload JVM agent reload
not noticeable
(quarkus starts fast)
live coding
Restarting quarkus due to changes in
Application$RenardeRequest.class, Application.class,
Application$ApplicationGlobals.class,
Application$Templates.class.
Live reload total time: 1.415s
Slide 186
Slide 186 text
@holly_cummins
fast start enables live coding
Slide 187
Slide 187 text
@holly-cummins.bsky.social
how to make people efficient
Slide 188
Slide 188 text
@holly-cummins.bsky.social
how to make people efficient
- make it hard to get wrong
Slide 189
Slide 189 text
@holly-cummins.bsky.social
how to make people efficient
- make it hard to get wrong
- strong typing
Slide 190
Slide 190 text
@holly-cummins.bsky.social
how to make people efficient
- make it hard to get wrong
- strong typing
- garbage collection
Slide 191
Slide 191 text
@holly-cummins.bsky.social
how to make people efficient
- make it hard to get wrong
- strong typing
- garbage collection
- give them a tight feedback loop
Slide 192
Slide 192 text
@holly-cummins.bsky.social
how to make people efficient
- make it hard to get wrong
- strong typing
- garbage collection
- give them a tight feedback loop
- for manual testing
- for automated testing
Slide 193
Slide 193 text
@holly-cummins.bsky.social
how to make people efficient
- make it hard to get wrong
- strong typing
- garbage collection
- give them a tight feedback loop
- for manual testing
- for automated testing
- allow less typing
Slide 194
Slide 194 text
@holly-cummins.bsky.social
how to make people efficient
- make it hard to get wrong
- strong typing
- garbage collection
- give them a tight feedback loop
- for manual testing
- for automated testing
- allow less typing
thank you, Java
- strong typing
- garbage collection
Slide 195
Slide 195 text
@holly-cummins.bsky.social
how to make people efficient
- make it hard to get wrong
- strong typing
- garbage collection
- give them a tight feedback loop
- for manual testing
- for automated testing
- allow less typing
thank you, Java
- strong typing
- garbage collection
we just covered this
Slide 196
Slide 196 text
@holly_cummins
index, index, index
(and let your ecosystem use the index)
Slide 197
Slide 197 text
@holly-cummins.bsky.social
most frameworks need to…
Slide 198
Slide 198 text
@holly-cummins.bsky.social
most frameworks need to…
-find all classes + interfaces +
methods + fields annotated with @a
Slide 199
Slide 199 text
@holly-cummins.bsky.social
most frameworks need to…
-find all classes + interfaces +
methods + fields annotated with @a
-find all classes implementing or
extending b
Slide 200
Slide 200 text
@holly-cummins.bsky.social
most frameworks need to…
-find all classes + interfaces +
methods + fields annotated with @a
-find all classes implementing or
extending b
java doesn’t help us
nothing in the reflection package does this
Slide 201
Slide 201 text
@holly-cummins.bsky.social
jandex
Slide 202
Slide 202 text
@holly-cummins.bsky.social
jandex
“offline reflection”
Slide 203
Slide 203 text
@holly-cummins.bsky.social
we have an index
now what?
Slide 204
Slide 204 text
@holly-cummins.bsky.social
challenge assumptions.
Slide 205
Slide 205 text
@holly-cummins.bsky.social
challenge assumptions.
what if developers
didn’t have to … ?
Slide 206
Slide 206 text
@holly-cummins.bsky.social
#RedHat
package com.example;
import org.jboss.logging.Logger;
public class Thing {
private static final Logger log = Logger.getLogger(Thing.class);
public void doSomething() {
log.info("It works!");
}
}
example: logging
Slide 207
Slide 207 text
@holly-cummins.bsky.social
#RedHat
package com.example;
import org.jboss.logging.Logger;
public class Thing {
private static final Logger log = Logger.getLogger(Thing.class);
public void doSomething() {
log.info("It works!");
}
}
example: logging
import io.quarkus.logging.Log;
Log
Slide 208
Slide 208 text
@holly_cummins
Slide 209
Slide 209 text
@holly_cummins
don’t make humans tell
the computer what the
computer already knows
Slide 210
Slide 210 text
@holly-cummins.bsky.social
“but isn’t that dynamism expensive?”
Slide 211
Slide 211 text
@holly-cummins.bsky.social
no.
“but isn’t that dynamism expensive?”
Slide 212
Slide 212 text
@holly-cummins.bsky.social
no.
- use Jandex to find use-sites of the Log class
“but isn’t that dynamism expensive?”
Slide 213
Slide 213 text
@holly-cummins.bsky.social
no.
- use Jandex to find use-sites of the Log class
- inject a static logger field $logger
“but isn’t that dynamism expensive?”
Slide 214
Slide 214 text
@holly-cummins.bsky.social
no.
- use Jandex to find use-sites of the Log class
- inject a static logger field $logger
- replace all calls of Log.method with calls to $logger.method
“but isn’t that dynamism expensive?”
Slide 215
Slide 215 text
@holly-cummins.bsky.social
no.
- use Jandex to find use-sites of the Log class
- inject a static logger field $logger
- replace all calls of Log.method with calls to $logger.method
… all at build time
“but isn’t that dynamism expensive?”
Slide 216
Slide 216 text
@holly-cummins.bsky.social
logging: compiled version
public class MyService {
// injected
private static final Logger $logger = Logger.getLogger(Thing.class)
public void doSomething() {
$logger.info(“It works!”);
}
}
Slide 217
Slide 217 text
@holly-cummins.bsky.social
what if…
you could inherit boilerplate
Hibernate queries from a superclass,
instead of having to write them all?
example: hibernate
Slide 218
Slide 218 text
@holly-cummins.bsky.social
#RedHat
@ApplicationScoped
public class GreetingRepository {
public Entity findByName(int name) {
return find("name", name).firstResult();
}
void persist(Entity entity) {}
void delete(Entity entity) {}
Entity findById(Id id) {}
List list(String query, Sort sort, Object... params) {
return null;
}
Stream stream(String query, Object... params) {
return null;
}
long count() {
return 0;
}
long count(String query, Object... params) {
return 0;
}
}
example: hibernate with panache
Slide 219
Slide 219 text
@holly-cummins.bsky.social
#RedHat
example: hibernate with panache
@ApplicationScoped
public class GreetingRepository implements PanacheRepository {
public Entity findByName(int name) {
return find("name", name).firstResult();
}
}
Slide 220
Slide 220 text
@holly-cummins.bsky.social
#RedHat
DAO
example: hibernate with panache
@ApplicationScoped
public class GreetingRepository implements PanacheRepository {
public Entity findByName(int name) {
return find("name", name).firstResult();
}
}
repository pattern
Slide 221
Slide 221 text
@holly-cummins.bsky.social
#RedHat
example: hibernate with panache
Slide 222
Slide 222 text
@holly-cummins.bsky.social
#RedHat
example: hibernate with panache
active record pattern
@Entity
public class Greeting extends PanacheEntity {
public String name;
public LocalDate issued;
@Version
public int version;
public static List getTodaysGreetings() {
return list("date", LocalDate.now());
}
}
Slide 223
Slide 223 text
@holly-cummins.bsky.social
why was this even hard?
public class PanacheEntity {
public static List listAll()
{
// but… how do we know which entity to query?
throw new UnobtainiumException();
}
}
Slide 224
Slide 224 text
@holly-cummins.bsky.social
why was this even hard?
public class PanacheEntity {
public static List listAll()
{
// but… how do we know which entity to query?
throw new UnobtainiumException();
}
}
signature can be generic
Slide 225
Slide 225 text
@holly-cummins.bsky.social
why was this even hard?
public class PanacheEntity {
public static List listAll()
{
// but… how do we know which entity to query?
throw new UnobtainiumException();
}
}
implementation cannot be generic
signature can be generic
Slide 226
Slide 226 text
@holly-cummins.bsky.social
here’s what we do
@Entity
public class Order extends PanacheEntity {
// … original class
// injected in the bytecode
// we add a Order.listAll method
public static List listAll() {
return DbOperations.listAll(Order.class);
}
}
Slide 227
Slide 227 text
@holly-cummins.bsky.social
machine efficiency unlocked human efficiency
we broke the tradeoff
Slide 228
Slide 228 text
@holly_cummins #RedHat
but magic
Slide 229
Slide 229 text
@holly_cummins #RedHat
magic
should always be optional + debuggable
Slide 230
Slide 230 text
@holly-cummins.bsky.social
why
#RedHat
Slide 231
Slide 231 text
@holly-cummins.bsky.social
happiness
is the most
important thing
Slide 232
Slide 232 text
@holly-cummins.bsky.social
happiness
is the most
important thing
Slide 233
Slide 233 text
@holly-cummins.bsky.social
happiness
is the most
important thing
developer
Slide 234
Slide 234 text
@holly-cummins.bsky.social
lower resource usage saves money
Slide 235
Slide 235 text
@holly-cummins.bsky.social
lower resource usage saves money
Slide 236
Slide 236 text
@holly-cummins.bsky.social
lower resource usage saves money
smaller memory footprint
Slide 237
Slide 237 text
@holly-cummins.bsky.social
lower resource usage saves money
smaller memory footprint
less cpu usage
Slide 238
Slide 238 text
@holly_cummins #RedHat
does being faster
and lighter mean
quarkus is greener?
Slide 239
Slide 239 text
@holly_cummins #RedHat
waste is killing the planet
Slide 240
Slide 240 text
@holly_cummins #RedHat
e-waste is killing the planet
Slide 241
Slide 241 text
@holly_cummins #RedHat
energy waste is
killing the planet
Slide 242
Slide 242 text
@holly_cummins #RedHat
inefficient software
is killing the planet
Slide 243
Slide 243 text
@holly-cummins.bsky.social
sustainability
saving planet
doing stuff we
want to do
Slide 244
Slide 244 text
@holly-cummins.bsky.social
sustainability
saving planet
doing stuff we
want to do
Slide 245
Slide 245 text
@holly-cummins.bsky.social
the vrroooom model gives hope
but …
Slide 246
Slide 246 text
@holly-cummins.bsky.social
Slide 247
Slide 247 text
@holly-cummins.bsky.social
invented by Dr. Vroom
(really!)
Slide 248
Slide 248 text
@holly-cummins.bsky.social
naming is the hardest problem in computer science
invented by Dr. Vroom
(really!)
Slide 249
Slide 249 text
@holly-cummins.bsky.social
naming is the hardest problem in computer science
Slide 250
Slide 250 text
@holly-cummins.bsky.social
my vrroooom model
Slide 251
Slide 251 text
No content
Slide 252
Slide 252 text
@holly-cummins.bsky.social
#RedHat
Slide 253
Slide 253 text
@holly-cummins.bsky.social
#RedHat
these two columns
are almost the same
Slide 254
Slide 254 text
@holly-cummins.bsky.social
#RedHat
Energy
1
10
100
Time
1 10 100
energy efficiency across programming languages
Python
Rust
Java
Go
Slide 255
Slide 255 text
@holly-cummins.bsky.social
#RedHat
Energy
1
10
100
Time
1 10 100
the trend line is more
or less straight
energy efficiency across programming languages
Python
Rust
Java
Go
Slide 256
Slide 256 text
@holly-cummins.bsky.social
#RedHat
vrrrooom model:
speed is a useful proxy
for energy efficiency
Slide 257
Slide 257 text
@holly_cummins #RedHat
capacity
Source: John O’Hara
Setup:
• REST + CRUD
• large heap
• RAPL energy measurement
Assumptions:
• US energy mix
climate impact at low load (single instance)
lower is better
Slide 258
Slide 258 text
@holly_cummins #RedHat
capacity
Source: John O’Hara
Setup:
• REST + CRUD
• large heap
• RAPL energy measurement
Assumptions:
• US energy mix
climate impact at low load (single instance)
line length shows max throughput
lower is better
Slide 259
Slide 259 text
@holly_cummins #RedHat
capacity
Source: John O’Hara
Setup:
• REST + CRUD
• large heap
• RAPL energy measurement
Assumptions:
• US energy mix
climate impact at low load (single instance)
quarkus on JVM has the lowest carbon …
because it has the highest throughput
line length shows max throughput
lower is better
Slide 260
Slide 260 text
@holly_cummins #RedHat
saving planet
coding how we
wanna code
we beat the tradeoff
Slide 261
Slide 261 text
@holly_cummins #RedHat
saving planet
coding how we
wanna code
we beat the tradeoff
Slide 262
Slide 262 text
@holly_cummins #RedHat
tl;dpa
(too long didn’t pay attention)
deployment density
lower cloud bill
frictionless development experience
Medium Nano
auto-provision services
live coding
continuous testing
don’t make people tell the
computer what the computer
already knows
greener
reducing waste
indexes
moving work so it hurts less
only doing work once
Slide 263
Slide 263 text
slides
Holly Cummins
Red Hat
https://hollycummins.com/quarkus-joy-oredev
@holly-cummins.bsky.social