Choosing the right Java Garbage Collector (GC) is critical to your application’s performance and stability. In this presentation, we explore the strengths and trade-offs of popular GC algorithms including G1, ZGC, Shenandoah, CMS, Serial, and Parallel—and show you how to decide which one is the best fit for your JVM. What you’ll find inside this deck: A clear breakdown of major Java Garbage Collector algorithms How to align GC strategies with application memory and SLA goals Performance metrics that highlight when a GC may be falling short Real-world examples of JVM performance impacts caused by GC choices Practical tuning tips and diagnostics to reduce pauses and boost efficiency Whether you’re aiming for low latency, high throughput, or balanced performance, this deck will help you choose the GC “partner” that makes your JVM thrive. Explore more on JVM performance: https://blog.gceasy.io

2. Choosing a GC Algorithm = Choosing a Life
Partner
Coexist In The Same Roof (JVM) GC Cleans Up The Mess Your App Creates
Aggressive App, Aggressive GC Competing For The Same Resources (CPU Cycles)

3. What is Garbage?
HTTP Request
Objects
Memory
Garbage

4. 3 - 4 Decades ago
Developer
Writes code to Manually evict Garbage
JVM
Automatically evicts Garbage
Now
How are objects Garbage Collected?
Evolution: Manual -> Automatic

5. Automatic GC sounds good right?
Yes, but for
GC pauses CPU consumption

6. Key Performance Indicators
GC Event’s Pause Time
1. Latency
Percentage of time spent in processing
customer transactions vs time spent in GC
activity. i.e. productive work vs non-productive
work
2. Throughput
Memory, CPU, Response Time
3. Foot Print
99.994
%
Memory: 2GB
CPU Time: 1 hr
9min
You can't optimize, what you can't measure
1.2 secs
4. Key Metric
Key Metrics in any
Performance Exercise

7. What is your Goal?
2. High Throughput
1. Low Pause Time
3. Footprint
Example: Online trading application - A 100 ms GC pause could delay stock trades, leading to
financial loss. Here the priority is predictability & ultra-low latency.
Example: Batch data processing system – A Hadoop/Spark job crunching terabytes overnight.
Occasional pauses don’t matter, but finishing the job faster (higher throughput) reduces infrastructure
cost.
Example: Microservice running in Kubernetes – Each pod is allocated a tight memory budget
(say 512 MB). Optimizing GC for low memory footprint allows higher pod density and lowers the
overall cloud bill.

8. CMS
Parallel
Serial
Shenandoa
h
G1
03
02
01
05
04
Z
GC
06
Epsilo
n
0
7
How many GC algorithms in Java 11?

9. GC Algorithm Performance Comparison
Stefan Johanson, Oracle
Stefan’s Blog: https://kstefanj.github.io/

10. Test Setup
2. Heap Size 16GB
1. SPECjbb® 2015 – Well-Known Standard Benchmark
3. Enable Large Pages

15. How to find Right GC Algorithm for your Application?
Tools, Tips & Tricks

16. Key Performance Indicators
GC Event’s Pause Time
1. Latency
Percentage of time spent in processing
customer transactions vs time spent in GC
activity. i.e. productive work vs non-productive
work
2. Throughput
Memory, CPU, Response Time
3. Foot Print
99.994
%
Memory: 2GB
CPU Time: 1 hr
9min
You can't optimize, what you can't measure
1.2 secs
4. Key Metric
Key Metrics in any
Performance Exercise

17. 17
Step 1: Enable GC Logging
-XX:+PrintGCDetails -XX:+PrintGCDateStamps -Xloggc:<file-path>
Till Java 8
-Xlog:gc*:file=<file-path>
From Java 9
Enable GC logs (always)
Almost zero overhead.
https://blog.gceasy.io/overhead-added-by-garbage-collection
ging/

18. 2019-08-31T01:09:19.397+0000: 1.606: [GC (Metadata GC Threshold) [PSYoungGen: 545393K->18495K(2446848K)] 545393K->18519K(8039424K),
0.0189376 secs] [Times: user=0.15 sys=0.01, real=0.02 secs]
2019-08-31T01:09:19.416+0000: 1.625: [Full GC (Metadata GC Threshold) [PSYoungGen: 18495K->0K(2446848K)] [ParOldGen: 24K->17366K(5592576K)]
18519K->17366K(8039424K), [Metaspace: 20781K->20781K(1067008K)], 0.0416162 secs] [Times: user=0.38 sys=0.03, real=0.04 secs]
2019-08-31T01:18:19.288+0000: 541.497: [GC (Metadata GC Threshold) [PSYoungGen: 1391495K->18847K(2446848K)] 1408861K->36230K(8039424K),
0.0568365 secs] [Times: user=0.31 sys=0.75, real=0.06 secs]
2019-08-31T01:18:19.345+0000: 541.554: [Full GC (Metadata GC Threshold) [PSYoungGen: 18847K->0K(2446848K)] [ParOldGen: 17382K-
>25397K(5592576K)] 36230K->25397K(8039424K), [Metaspace: 34865K->34865K(1079296K)], 0.0467640 secs] [Times: user=0.31 sys=0.08, real=0.04
secs]
2019-08-31T02:33:20.326+0000: 5042.536: [GC (Allocation Failure) [PSYoungGen: 2097664K->11337K(2446848K)] 2123061K->36742K(8039424K),
0.3298985 secs] [Times: user=0.00 sys=9.20, real=0.33 secs]
2019-08-31T03:40:11.749+0000: 9053.959: [GC (Allocation Failure) [PSYoungGen: 2109001K->15776K(2446848K)] 2134406K->41189K(8039424K),
0.0517517 secs] [Times: user=0.00 sys=1.22, real=0.05 secs]
2019-08-31T05:11:46.869+0000: 14549.079: [GC (Allocation Failure) [PSYoungGen: 2113440K->24832K(2446848K)] 2138853K->50253K(8039424K),
0.0392831 secs] [Times: user=0.02 sys=0.79, real=0.04 secs]
2019-08-31T06:26:10.376+0000: 19012.586: [GC (Allocation Failure) [PSYoungGen: 2122496K->25600K(2756096K)] 2147917K->58149K(8348672K),
0.0371416 secs] [Times: user=0.01 sys=0.75, real=0.04 secs]
2019-08-31T07:50:03.442+0000: 24045.652: [GC (Allocation Failure) [PSYoungGen: 2756096K->32768K(2763264K)] 2788645K->72397K(8355840K),
0.0709641 secs] [Times: user=0.16 sys=1.39, real=0.07 secs]
2019-08-31T09:04:21.406+0000: 28503.616: [GC (Allocation Failure) [PSYoungGen: 2763264K->32768K(2733568K)] 2802893K->83469K(8326144K),
0.0789178 secs] [Times: user=0.12 sys=1.59, real=0.08 secs]
Vanilla Format

19. GC Log analysis tools
https://developer.ibm.com/ja
vasdk/tools/
IBM GC & Memory visualizer
GC Viewer
https://github.com/chewiebu
g/GCViewer
GCeasy
https://gceasy.io/
Google Garbage cat (cms)
https://code.google.com/arch
ive/a/eclipselabs.org/p/garba
gecat
HP Jmeter
https://h20392.www2.hpe.co
m/portal/swdepot/displayPro
ductInfo.do?productNumber
=HPJMETER
03
02
01
05
04

20. https://tinyurl.com/mw4kak4a
GC Tuning Worksheet

21. 9 GC Tuning Tips & Tricks
https://blog.gceasy.io/reduce-long-garbage-collection-pauses/

22. https://blog.gceasy.io/automobile-company-optimizes-performance-using-gceasy/
Real case study: GC Performance of a one of the Largest Car Maker

23. Large Car Maker Reduces Response time by
50%
https://blog.gceasy.io/automobile-company-optimizes-performance-using-gceasy/

24. Real case study: Warehouse Robotics App Poor Performance
https://blog.gceasy.io/2024/02/21/optimizing-robotics-applications-performance/

25. GC Tuning
How We Saved 70K Cores Across 30 Mission-Critical Services?
https://www.uber.com/en-IN/blog/how-we-saved-70k-cores-across-30-mission-critical-services/

26. More GC Tuning case studies
Uber Saves Millions of $
https://blog.gceasy.io/2022/03/04/garbage-collection-tuning-success-story-reducing-young-gen-siz
e/
Large Automobile Manufacturer Improves Response Time
https://blog.gceasy.io/2022/03/04/garbage-collection-tuning-success-story-reducing-young-gen-siz
e/
CloudBees (Jenkins Parent company) optimizes
https://blog.gceasy.io/2019/08/01/cloudbees-gc-performance-optimized-with-gceasy/
Oracle optimizes App performance by tuning GC
https://blog.gceasy.io/2022/12/06/oracle-architect-optimizes-performance-using-gceasy/

27. 27

28. Thank you my friends!
Ram Lakshmanan
@ycrash_rca
This deck will be published in: https://blog.gceasy.io
https://www.linkedin.com/company/ycrash