Scaling Jakarta EE Applications Vertically and Horizontally with Jelastic PaaS
0 likes291 views
In this presentation, you'll find out what metrics should be tracked in order to meet the load requirements of application, how to finetune scaling triggers in order to efficiently handle different load levels, how to automate vertical and horizontal scaling of Jakarta EE applications running in the cloud. Also, we share how to integrate load performance testing tools for adjusting horizontal scaling and making sure that your application can cope with production workloads. Practical side is shown based on Jelastic PaaS https://jelastic.com/
![Timely Reduce Unused Committed Memory (JEP 346)
Make the G1 garbage collector automatically give back Java heap memory
to the operating system when idle
● -XX:G1PeriodicGCInterval=[milliseconds]
● -XX:G1PeriodicGCSystemLoadThreshold=[float]
● -XX:+G1PeriodicGCInvokesConcurrent
JEP 346: Promptly Return Unused Committed Memory from G1
java -Xmx2g -XX:+UseG1GC -XX:G1PeriodicGCInterval=900k
-XX:G1PeriodicGCSystemLoadThreshold=0.6 -jar app.jar
Available from Java 12](https://image.slidesharecdn.com/jakartaone-scalingjakartaeeapplicationsverticallyandhorizontallywithjelasticpaas-201208160101/85/Scaling-Jakarta-EE-Applications-Vertically-and-Horizontally-with-Jelastic-PaaS-29-320.jpg)
Ad
More Related Content
What's hot (20)
Similar to Scaling Jakarta EE Applications Vertically and Horizontally with Jelastic PaaS (20)
Ad
More from Jelastic Multi-Cloud PaaS (20)
Ad
Recently uploaded (20)
Scaling Jakarta EE Applications Vertically and Horizontally with Jelastic PaaS
- 1. Scaling Jakarta EE Applications Vertically and Horizontally with Jelastic PaaS
- 2. About Speaker ● Ruslan Synytsky ● CEO and co-founder of Jelastic PaaS ● Java Champion ● Two-times Duke’s Choice Award Winner ● Former lead of engineering team at National Data Center (NDC) at National Space Agency of Ukraine ● @siruslan
- 3. Agenda ● Vertical vs Horizontal Scaling ● Scaling Jakarta EE Applications in VMs and Containers ● Overcoming Problems of Java Vertical Scaling ● Testing Java Vertical and Horizontal Scaling with Jakarta EE Application Server
- 5. Vertical vs Horizontal Scaling Vertical Scaling Horizontal Scaling VS
- 6. Vertical Scaling Pros & Cons ● Easy to deploy and configure ● No code change required during scaling up ● No need in data synchronization ● No network connection and latency issues ● No complexity of multiple instances management ● Cost efficiency, minimal overhead comparing to running several nodes ● Instances with large memory limits available across different infrastructure providers ● The fastest way to boost performance ● No high availability - significant risk of downtime if the server fails or a load exceeds the available resource ● Rolling deployments without downtime is difficult and available only if supported at the runtime stack level ● Hard to migrate to a bigger physical box, live migration support is required in virtualized environments for keeping uptime The ability to grow or shrink the resources available to a particular instance in order to react on load changes in fast and flexible way Pros Cons
- 7. Horizontal Scaling Pros & Cons ● Easier to achieve fault-tolerance and high availability ● Better resilience of data and services ● Rolling deployments without going offline ● Better performance at highly loaded projects ● Available tools to configure scaling without extensive tech knowledge required ● Must have for business and mission critical applications ● Designing scalable applications can be a challenge ● Complexity of data distribution and parallel processing, less consistency, large number of cross-server communications ● Higher overallocation of resources ● Often purchase of additional licenses required ● Increased load and management complexity on the network level Pros Cons The process of changing the number of nodes in order to enable high availability and improve performance
- 8. Scaling Jakarta EE in VMs & Containers
- 9. Resource Limit vs Real Usage in VM and Container
- 10. VMs vs Container Vertical Scaling Resizing of the same container on the fly is easier, cheaper and faster than moving to a larger VM
- 11. Automatic Vertical Scaling Every container hosted with Jelastic PaaS is divided into granular units – cloudlets (128MiB of RAM and 400MHz of CPU)
- 12. Automatic Vertical Scaling You can set up a maximum Scaling Limit for each container, so the resources will be always available in case of load spikes or other consumption changes
- 13. Jelastic PaaS adjusts configs of certified Java software stacks based on the vertical scaling limits Automatic Stacks Adjusting based on Scaling Limits Environment variable XMX_DEF_PERCENT=80
- 14. UNDERALLOCATION JVM Vertical Scaling OK OVERALLOCATION
- 15. Using automatic vertical scaling, cloud provides can offer economically advantageous pricing based on the actual resource consumption Forbes - Deceptive Cloud Efficiency: Do You Really Pay As You Use? Pay-As-You-Go Pay-per-Use Pay-As-You-Go vs Pay-per-Use
- 16. Real Statistics of Resource Consumption with Containers
- 18. Stateless (Create New) vs Stateful (Clone) MasterMaster Worker Worker Stateless Stateful Stateless mode creates an empty node from a base container image template. Works faster than stateful and easy to parallelize the scaling process. Stateful mode creates a new node as a full copy (clone) from the master. Usually takes longer than stateless, but data is replicated automatically. Empty Clone
- 19. Horizontal Scaling of Jakarta EE Containers
- 20. Automatic Horizontal Scaling of Jakarta EE Containers
- 21. Anti-Affinity Rules All newly added containers of the single layer are created at the different hosts, providing advanced high-availability and failover protection.
- 22. Automation of Scaling with Cloud Scripting Possibility to perform custom automation actions at the scaling events using Cloud Scripting ● onBeforeScaleOut ● onAfterScaleOut ● onBeforeScaleIn ● onAfterScaleIn ● onBeforeServiceScaleOut ● onAfterServiceScaleOut ● onBeforeAddNode ● onAfterAddNode ● onBeforeRemoveNode ● onAfterRemoveNode ● onBeforeSetCloudletCount ● onAfterSetCloudletCount
- 23. Automatic Clusterization of Jakarta EE Application Servers Clustering mode can be automatically enabled from the topology wizard. Interconnected servers with pre-configured session replication and load balancing provides high availability across nodes and guarantees continuous performance of deployed Java applications.
- 24. Automatic Clusterization of Jakarta EE Containers
- 25. Jakarta EE 9 Support ● Tomcat ● TomEE ● GlassFish ● WildFly ● Jetty ● Payara is coming ● OpenLiberty is coming
- 27. Most Challenging Aspect of Working with Jakarta EE Jakarta EE Developer Survey 2018 “The most widely acknowledged issue when employing with Java EE is large memory requirements (40%)”
- 28. Unreleased Heap Memory Over-Allocation and Underutilization The Problem Symptoms
- 29. Timely Reduce Unused Committed Memory (JEP 346) Make the G1 garbage collector automatically give back Java heap memory to the operating system when idle ● -XX:G1PeriodicGCInterval=[milliseconds] ● -XX:G1PeriodicGCSystemLoadThreshold=[float] ● -XX:+G1PeriodicGCInvokesConcurrent JEP 346: Promptly Return Unused Committed Memory from G1 java -Xmx2g -XX:+UseG1GC -XX:G1PeriodicGCInterval=900k -XX:G1PeriodicGCSystemLoadThreshold=0.6 -jar app.jar Available from Java 12
- 30. Immediately Improved Heap Elasticity Automatically Released Heap
- 32. Workaround for Java < 12: Calling Full GC Periodically https://github.com/jelastic- jps/java-memory-agent As compacting GC cycles are not triggered automatically, we execute them explicitly by injecting an agent which monitors the memory usage and calls System.gc() periodically: -javaagent:jelastic-gc-agent.jar=period=300,debug=true
- 33. G1 and Full GC java -XX:+UseG1GC -Xmx2g -jar app.jar https://github.com/jelastic/java-vertical-scaling-test
- 34. Running GC Tests with Jakarta EE
- 35. ● simple .war artifact deployed with GlassFish 5 ● JSP that sets 1MB attribute in session ● 1 min session timeout ● https://github.com/jelastic/java-vertical-scaling-test/tree/jakartaee/webapp Load Testing of Vertical Scaling with GlassFish 5 Load test GETs webapp endpoint n times for i in {1..n}; do curl -s localhost:8080 > /dev/null; done
- 36. Parallel -Xmx3g -XX:+UseCompressedOops -XX:+UseParallelGC for i in {1..1000}; do curl -s localhost:8080 > /dev/null; done
- 37. G1 -Xmx3g -XX:+UseCompressedOops -XX:+UseG1GC for i in {1..1000}; do curl -s localhost:8080 > /dev/null; done
- 38. Shenandoah -Xmx3g -XX:+UseCompressedOops -XX:+UseShenandoahGC -XX:ShenandoahGCHeuristics=compact for i in {1..1000}; do curl -s localhost:8080 > /dev/null; done
- 39. Main Points of Elastic Vertical Scaling A - initial usage B - maximum usage C - growth speed D - duration before release E - release speed F - minimum usage after release Different GCs provide different results and fine tuning options
- 40. Vertical and Horizontal Scaling in Jelastic PaaS Vertical Scaling ● Solving right-sizing problem - no need to guess in advance how much resources to allocate ● RAM and CPU are available accordingly to the runtime needs and specified container scaling limits ● The actual consumption is automatically monitored and billed according to the real use ● Ability to adjust maximum scaling limits without migration required Horizontal Scaling ● Solving high-availability problem - easy to start with production ready environment ● Automated clusterization (aka auto-clustering) reduces complexity of scaling management ● Automatic scaling based on triggers ● Stateful and stateless scaling mode ● Anti-affinity distribution rules ● Flexible node group management options via UI, API and JPS
- 41. Request Jakarta EE Scaling Optimization If you have scaling or memory usage issues with Jakarta EE applications running in containers or VMs request an assistance: [email protected]