Technical Track presented by Wayne E. Seguin at Stark & Wayne.
When working with Cloud Foundry, AppFirst’s unique data collection provides unparalleled visibility over traditional tools. This full stack visibility spans the entire PaaS and respective services.
In this talk, Wayne E. Seguin of Stark & Wayne will discuss how his team found and utilized AppFirst for PaaS to gain unprecedented visibility. Through years of experience working with PaaS operations, they combined lessons learned in operations workflows yielding a deep understanding of both the space and needs. This year we will roll out an integration enabling Cloud Foundry operators to have full scale visibility across all resources.
7. Our Backgrounds
• Years in the PaaS space focused on Operations
– Metrics & Data Collection
– Aggregation
– Correlation
– Alerting
– Working for/with Dr. Nic
17. Data Collection Requirements
Process Resource Metrics
CPU & RAM utilization
Files & Socket read/write tps
Threads
Stolen Time (Living in a virtual world)
etc…
34. Have Cake & Eat It
• Minimize CapEX & OpEX
– Through consolidation of Systems
• Allow Ops to run lean & mean
– This is one reason we love BOSH+CF after all!!!
38. Phase I – BOSH Integration
We are currently working with Pivotal Web
Services and AppFirst on the BOSH layer
integration.
39. Phase I – BOSH Integration
This gives us the basic essential data blocks
System Metrics
Process Metrics
Log Metrics
40. Phase II – Targeted Alerting
Alerting based on detected failure cases
Zombie containers
File Systems remounted Read Only
File Systems remounted incorrectly
Network saturation
DEA Pool Saturation
etc…
41. Phase III – Subsystems Detection
Detection of BOSH/CF Subsystems for Automatic
Targeted metrics/logs collection
Operations Dashboards per Subsystem
DEAs
Message Bus
Router
Cloud Controller
etc…
42. Phase IV – Applications View
Aggregated process health and log view for any
given application running across all DEAs
We were excited to come and meet everyone here at CF Summit!
The flight here was a lot of fun…
We want all of this across BOSH, Services deployed on BOSH, CF, Applications running on CF and oh yeah we want our Legacy Systems to also be able to be monitored by the same methods.
We also want this to be able to be deployed via SaaS for smaller environments as well as installable on-prem for our larger more strict environments.