Learn improved performance testing for Cloud-native applications by integrating Elasticsearch with Riverbed application performance monitoring (APM). The objective was to create realistic performance testing that was representative of real-world usage of the application.
3. § Richard Juknavorian
– Principal
– IT Analytics and Performance Management
– IT Squared LLC
§ Alex Kozlov
– IT Architect and Technical Evangelist
– Riverbed
Presenters
4. 4
Agenda
Observability, Monitoring Domains, Application
Performance Management
Use Case 1: Using ElasticSearch to Identify and
Troubleshoot Performance Anomalies in Application
Performance Testing
Use Case 2: Implementing Fail-Safe CI/CD Pipeline
with ElasticSearch and DevOps Automation Toolset
15. .so
Host
.so
.so
AgentAnalysis Server
EUE Transactions (no sampling)
Transaction Traces (no sampling)
1-Second Metrics
Stitched EUE + Multi-Tier Traces
(no sampling)
1-Second Metrics
.js
SteelCentral APM: Deployment Architecture
16.
17. 1. Deploy Application in PV Lab
2. Execute Load Test
3. Collect Metrics (Load, RT, Error Rate)
4. Monitor Host Resources and Metrics
(CPU, Mem, I/O, Database Wait, Garbage Collection)
5. Monitor SQL Query Executions
§ Response Time (RT) is Too Crude to Identify Anomalies
§ Need to understand composition of delays across tiers down to the CODE-level
Traditional Approach
Performance Testing
Challenges:
18. • Classify Every Transaction by Response Time
• Watch for Spikes in the Number of Slow Transactions
Improved Anomaly Detection:
Anomalies
Anomaly
20. 1. Logs &
Events
Improved RCA:
RCA with Distributed Tracing by SteelCentral APM
FileBeat
Logstash
2. Metrics 3. Tracing
Big Data
Analytics
MetricBeat SteelCentral
APM
21. 1. Deploy Application in PV Lab
2. Execute Load Test
3. Collect Metrics (Load, RT, Error Rate)
4. Monitor Host and Middleware Metrics with 1s resolution
5. Use always-on Tracing with Deep In-flight Code Analysis
6. Identify Anomalies by Transaction Within Narrow Time Range
7. Run Trace Analytics on Method Calls, SQL Calls, Host and Resource Metrics
to Identify Common Root Cause
8. Drill-down into Specific Traces as Needed
Optimized
Performance Testing
30. Would you like to know more?
Monitoring with Elastic Stack
SteelCentral APM and Application Tracing
DevOps and CI/CD analytics with Elastic Stack
Anomaly Detection and ML
Automation of CI/CD
Observability in Cloud-Native Architectures
31. Q&A - Cloud-Native Application Monitoring
Richard Juknavorian
Principal, IT Squared
rjuknavorian@it-sq.net
@ItSquared
Alex Kozlov
Technical Evangelist, Riverbed
alex.kozlov@riverbed.com
@YK02464
Try for free in our sandbox
www.riverbed.com/try-appinternals