microservice monitoring and quality management for modern apps and infrastructures
TRANSCRIPT
Microservice Monitoring and Quality Management for Modern Apps and
Infrastructures
Speakers
Pavlo Baron, CTO, Instana
Ravi Yadav, Partnerships Architect, Mesosphere
Pedro Pacheco, Solutions Architect, Instana
Why Instana & DC/OS?
Pavlo Baron, CTO, Instana
Architectures Have Evolved
Host
Container Middlewa
reMicroservices
Business Service
Complexity + Scale Never Cease
Modern Architectures Have Evolved to a new Level of Distribution and
Scale
Mesosphere DC/OS
DC/OS is the Foundation for Modern Architecture Implementation
• Scheduling• Container lifecycle
management • Location transparent resource
management• Continuous deployment
DC/OS engineers and automates change
•Containers facilitating rapid development and deployment
•A container allocation from DC/OS change code, dependency, messaging structure
Feedback! What just happened!
• What changed?• Where did change occur?• Was there any quality impact?• What is the root cause?
In a system of constant change, operators need feedback!
Polling Question #1
© 2017 Mesosphere, Inc. All Rights Reserved. 9
MESOSPHERE DC/OS
Ravi Yadav, Partnerships Architect
PLATFORM SERVICES
CONTAINERORCHESTRATI
ONCI/CD BIG DATA
ANALYTICSMESSAGE
QUEUEDISTRIBUTED
DATABASE SEARCH
BUSINESS APPS
Distributed computing expertise.DevOps process & culture.
Cloud-native technologies.
OPERATIONAL EXPERTISE
INFRASTRUCTURE
PHYSICAL VIRTUAL CLOUD
BUILDING THE MODERN ENTERPRISE STACK
10
BUSINESS APPS
$
$ $
$$ $
$$$
$$
$$$ $$$ $$
$ $
$$$
$
$$
CLOUD LIMITATIONS
- LOCK-IN- HIGH COST- NO CONTROL
AWSPLATFORM SERVICES
OPERATIONS & TOOLS
ADMINISTRATORS OPERATIONAL PROCESSES PROPRIETARY TECHNOLOGIES
CONTAINERORCHESTRATI
ONCI/CD BIG DATA
ANALYTICSMESSAGE
QUEUEDISTRIBUTED
DATABASE SEARCH
AWS INFRASTRUCTURE (EC2)
11
BUSINESS APPS
MESOSPHERE ENTERPRISE DC/OS
CONTAINERORCHESTRATI
ONCI/CD BIG DATA
ANALYTICSMESSAGE
QUEUEDISTRIBUTED
DATABASE SEARCH
PLATFORM SERVICES
RESOURCEMANAGEMENT OPERATIONS SECURITY MULTI-TENANCY
DATACENTER-CLOUD OPERATING SYSTEM
APACHE MESOS
PRIVATE CLOUD AWS INFRASTRUCTURE (EC2)
GOOGLE CLOUD MICROSOFT AZURE
MESOSPHERE ENTERPRISE DC/OS
More than 100 services
CLOUD INDEPENDENCE
12
© 2017 Mesosphere, Inc. All Rights Reserved.
MESOSPHERE DC/OS: RUN YOUR DATACENTER AS A GIANT COMPUTER
Deploy & run complex services with an app store like experience
Manage your infrastructure as a single giant computing pool
© 2017 Mesosphere, Inc. All Rights Reserved. 14
NEW UIDC/OS
DC/OS 1.9
© 2017 Mesosphere, Inc. All Rights Reserved. 15
KEY THEMES - DC/OS 1.9
WORKLOADS
● Pods ● GPU based
scheduling
COMPLIANCEREPORTS
● NIST-800/53● ISO 27001● FISMA-Mod.● SOC-2● HIPAA
● Alluxio ● Couchbase● Datastax DSE ● Elastic (ELK)● Redis● Apache Flink
DATA SERVICES ECOSYSTEM
OPERATIONS
● Remote Container Shell
● Unified Metrics● Unified Logging● Deployment
Failure Debugging● Upgrades &
Configuration updates
© 2015 Mesosphere, Inc. All Rights Reserved. 16
WHY MESOSPHERE?
Unmatched Technology: DC/OS
● Containers in production at scale
● Stateful data services
● Maturity of technology
● Easy to install & run - anywhere (cloud, physical VM)
● Evergreen platform with DC/OS services
Experts on Distributed Systems● Experts & committers of open source
software (e.g., Mesos)
● Experience building & operating distributed systems at scale
● Experience helping enterprises solve real modern apps challenges
e
Polling Question #2
Host
Container
Middleware
Modern Architecture
MicroServices
Business Service
Modern Technology Stack
Business Service
Business Service
•Interconnected with multiple dependencies
•Many programming languages and frameworks
• Many container vendors• Variety of service types, micro
to macro• Legacy components
Challenge with Monitoring Modern Architectures
Host Middlewa
re
Microservice
Container
Custom Tracing or APM
Logging
Container monitoring
System monitoring
• Alerts come from technical silos
• No correlation between alerts and business services
• Alarms don’t map to problems• No context of situation• Sampling/statistical based
monitoring• Too much specialized
expertise needed to assure quality
No Alignment Between Tools and the Modern Stack
Business Service
Misalignment = Zero Visibility
Instana’s Dynamic APM: Bringing Order to Complexity and Chaos
Host
Container
Middleware
Code Exceptions/Errors
MicroService
Business Service
Business KPIs & SLAs
Automatic discovery of architecture with alignment of context:
• Vertical technical stack• Service dependency map• Measurement of service quality
Automatic Root Cause Analysis in 3 Seconds
Immediate exact problem identification with full context
Data Ingestion &Health
Calculation
Sensor Data
Real-time Stream Processing
Incident Detection
Alerting
Dependency
Health
Metrics
Dynamic Graph
API & CLI
Configuration
Instana Processing Pipeline
3 seconds from sensing to root cause identification
Quality of Service
3D Map
One Agent per Host
One Sensor per active component
Continuous discovery up the stack
Trace messages between microservices Sensor Repository
Agent
Knowledge Engine
Elasticsearch sensor
Tomcat sensor
JVM sensor
Linux sensor
Auto Discovery
Communication
LocalSensor Memory
&Contextual
Compression
Immediate, Automatic and Continuous Discovery of Components and Dependencies
Demo Use Cases
Continuous and Automatic Discovery
Application Mapping
Changes Detection and Marathon Events Correlation
Root Cause Understanding
Polling Question #3
Demo