The Logentries and Hosted Graphite integration allows you to connect two of your favorite Ops tools to easily extract important data from your log files, visualize them as metrics, and share them in Hosted Graphite dashboards.
• Integrate your systems to extract the metrics you need, from both your applications and log data.
• Set-up log metric dashboards based on common use cases (e.g. error tracking, performance, app usage).
• Get off the "complexity elevator" of hosting your own in-house logging or graphite solutions.
• Delight your team and organization with valuable metrics and performance insights.
Reassessing the Bedrock of Clinical Function Models: An Examination of Large ...
Correlate Log Data with Business Metrics Like a Jedi
1. by RevelOps
How to Correlate Log Data with Business Metrics Like a Jedi
Oct 22nd, 2014
2. by RevelOps Housekeeping
• Webinar is being recorded, the archive will be emailed to you
• You can ask questions via the chat pane
• Follow us @Logentries and @HostedGraphite
3. by RevelOps Meet the Hosts
Dave Concannon
Hosted Graphite CEO & Co-founder
Trevor Parsons
Logentries Co-founder and Chief Scientist
4. by RevelOps Outline
• Introduction: Logentries & Hosted Graphite
• Logs and Metrics: Correlation is King!
• What vs. Why
• Collecting data from every layer in your software stack
• Turning Logs into Metrics
• Sending Metrics to Graphite
• Real Time Metrics with Graphite
• Use Case: Application Performance Monitoring w/ Logentries & Hosted Graphite
• Rolling your own Logging and Graphite Solutions: Pros and Cons
• Q & A
5. by RevelOps
A Heritage in Advanced Research
•Founders hold PhDs in Distributed Systems
•Former Scientists at the IBM Center for Advanced Studies
•Built Log Management Solution for IBM (RTCE)
Global SaaS Log Management Service
•More than 1,000 customers in over 100 countries, processing over 10B events
daily
•Locations in Boston, Dublin, Prague
•Raised more than $11M in venture funding
About Logentries
Leadership Team
President & CEO
Andrew Burton
Proven software executive, with over 16 years of experience in start-ups
& enterprise companies.
Co-founder, Chief Scientist
Trevor Parsons
Formerly a scientist at IBM's Center for advanced studies with a
PhD in Computer Science.
VP, Engineering
Brendan Dillon
Experienced software engineering & security executive,
formerly CTO at Adaptive Mobile.
Viliam Holub
Over a decade developing enterprise software, with a PhD in
Computer Science.
Co-founder, CTO
VP, Marketing
Leigh Merrigan
SaaS marketing, communications and branding executive,
formerly with AppNeta.
6. by RevelOps About Hosted Graphite
Team
Charlie Von Metzradt
Hacker and tinkerer with experience in running very large metric
collection systems
Rob McAdoo
PhD. Computer science
Dan Fox Ops Engineer
Ex-Amazon ops engineer
Cian Synnott
Ex-google data wrangler
Co-founder
Years of measuring things in a variety of companies
Co-founder
Engineer
Data Scientist
Dave Concannon
Emmanuel Stone
Experienced startup engineer
Engineer
Years of dealing with metric data
•Founders worked with high-volume data that measured games, like “Call of
Duty”
•Broad experience across many industries and company sizes
SaaS Metrics Service
•Hundreds of customers, hundreds of countries, billions of data points daily
•Locations in Dublin and Indiana
•Bootstrapped with blood, sweat, and tears :)
Hugh Nowlan Security Engineer
8. by RevelOps Logs & Metrics: Correlation is King
Question
•What Happened?
Answer
•Things got slow!
9. by RevelOps Logs & Metrics: Correlation is King
Question
•Why?
Answer
•Correlation is King!
10. by RevelOps Logs & Metrics: Correlation is King
• Response Time Increasing AND CPU Load Increasing
11. by RevelOps Logs & Metrics: Correlation is King
Data point 1
•Number of Users
CPU
and
Response Time
Solution
•Need more servers
12. by RevelOps Logs & Metrics: Correlation is King
Data point 2
•Java Exceptions
CPU
and
Response Time
Solution
•Fix the Code
13. by RevelOps Logs & Metrics: Correlation is King
Data point 3
•Slow Queries
CPU
and
Response Time
Solution
•Tune the DB
Logs maintain the evidence
14. by RevelOps Data Sources
What you can typically get from logs
• OS Level
• Syslog
• Windows Event Logs
• Server Resource Usage
• Middleware:
• Web Servers
• App Servers
• Databases
• Message Queues
• Application Logs
• Routers, Firewalls
• APIs:
• E.g. Cloudwatch, CloudTrail
• Client Side:
• Web Browser
• Mobile Apps
15. by RevelOps Data Sources
What you will typically send to Graphite:
• Server Level:
• CPU Load
• Network IO
• Disk IO
• Memory usage
• Application data:
• Timing / Measurement of specific functions or functional
areas
• Overall request time and frequency
• How often the out-of-memory killer kills a process
• Security information:
• Number of SSH login attempts - failed and succeeded
• Sensor Data:
• Temperature / Pressure / Flow rates
• Transmission rates
• APIs:
• E.g. Cloudwatch
• Client Side:
• Web Browser
20. by RevelOps How to work with Metrics in Hosted Graphite?
StatsD - Pre-aggregator for metric data
• We also provide Hosted StatsD!
• Support for counts / gauges / percentile data
Diamond
• A useful server agent to collect system data (CPU/Memory/etc)
Aggregating hundreds or thousands of sources into a single metric
• Power in percentile data
• 90th percentile as an indicator of general health - Averages lie!
• Dig in to specific problems after seeing a quick overview
• Avoids “twitchy” monitoring
Data views
• Sums / rates / observations, no need to define views beforehand
21. by RevelOps Typical Use Case: DevOps Dashboard
Logs Metrics Graphite
OS logs & Server Metrics:
• SSH Acccess,
• CPU, Mem…
Web Server Logs:
• Response time
App Server Logs:
• App Exceptions
Client Side Logs:
• Response time (client side)
• Url & Usage tracking
22. by RevelOps
Typical Use Case: DevOps Dashboard
Metrics Graphite
Application Metrics:
• Code Latency
• Query Time to DB
• Throughput
Diamond:
• Collectors… Hadoop, Mongo, Kafka, MySQL,
NetApp, RabbitMQ, Redis, AWS S3…
• Check
out:https://github.com/BrightcoveOS/Diamond/wiki
/Collectors
23. by RevelOps Typical Use Case: DevOps Dashboard
Some questions you can answer:
Application Performance Monitoring:
•Is User experience ok?
• Client side response time
• HTTP 404s/500s?
• Query time
Server Capacity:
•CPU vs user load
Security:
•Failed Logins vs. unusual system behavior
24. by RevelOps Rolling your own Logging/Metrics
Great open source tools e.g. Logstash & Graphite
•They are free…
• No license/monthly billing etc.
•They are not free…
• 250 GB per day, 8 servers on AWS, ~$4k per month
• Maintenance, few hours per week, especially when you have a clustered server environment
• Downtime can occur as volumes grow, if no autoscaling in place
• Log volumes ALWAYS grow…
•Some more reading:
• https://blog.logentries.com/2014/09/the-pros-and-cons-of-open-source-logging/
25. by RevelOps Give it a Try for Free
• Get started in minutes with the free Logentries service!
• https://logentries.com/quick-start
• Free 14 Day trial of Hosted Graphite
• https://www.hostedgraphite.com/accounts/register/
• Logentries & Hosted Graphite: https://blog.logentries.com/2014/10/connecting-logs-and-metrics-logentries-and-hosted-
graphite/
• Contact us directly!
• Trevor.parsons@logentries.com
• support@logentries.com
• dave@hostedgraphite.com, help@hostedgraphite.com
• Let’s take a few questions!