Stage 2 Management Systems Building a foundation for flexible service delivery.
-
Upload
dina-matthews -
Category
Documents
-
view
214 -
download
0
Transcript of Stage 2 Management Systems Building a foundation for flexible service delivery.
The Concept Empower infrastructure to perform its own administration
Free up IT resources from maintaining delivered projects
Automate the post-mortem
Intelligence assisted managed services
Closing The Loop Open Loop Processing
- The main purpose behind Intelligence Assisted Systems/Operations is to leverage the theory of closed feedback loop processing.
- When heating your house with an open loop control system, the parts look like the below diagram:
Closing The Loop Intelligence Assisted Systems/Operations
- To take this a step further and enter into intelligence assisted systems/operations we would make a slight tweak to our open loop system:
Prototyping Intelligence Assisted Systems/Operations
Orange Systems:
Controller
Sensor
Mass Orchestrator
Control Logic
Healing Logic
Broker
Notifications
Sync/Async Tasks
Green Systems:
Report API
Service/Health Scheduler API
Solving Problems – Running in production Intelligence Assisted Systems/Operations
- Initial test runs on production systems have been quite promising
- Most current Heal-Ops are completed in under 2 seconds
- Iterative algorithms allow for state update after each operation pass to mitigate the chance of attempting a Heal-Op on a non-existent error.
- Have been able to heal CRM cluster failures before Nagios can scream
- Can provide data streams for system health and usage
- Fully agent-less design
Why Stage 2? Stage 2 Management Systems – Automated Troubleshooting
- Not re-inventing the wheel, instead managing the failure of common service management tools
- Programmed to heal with IT methodologies
- Not just automating the system, but the system administration
- Works in conjunction with most current infrastructure management tools
How Much Further Can We Go? Stage 3 Management Systems – Predictive Analytics
- Working closely with Metafor Software as an upstream analytics engine
- Providing system resource streams
- State and drift measurement over large clusters
- Pro-active healing actions
- Predict component failure
- Isolate causes of system turbulence or bad cluster cells
- Provide scale and de-scale analysis
- Predict optimum load resource commitments
IASO and IaaS Stage 2/3 Management Systems – Cloud Enhancements
- Application infrastructure auto-scaling
- Using mass-orchestration and CFL processing data
- Predictive scaling for fault and load scenarios
- Currently supporting Apache CloudStack
- Automated guest administration
- Working on agent-less guest tools beyond auto-scaling
- Contemplating adding the code for self user management
Managed Production System Intelligence Assisted Systems/Operations
Example of a multi-tiered application infrastructure running on the cloud and managed by closed feedback loop systems
Application Server
Application Server
Load-Balancer{Nginx}
Load-Balancer{Nginx}
Master Database{MySQL}
Firewall
Memcache: DataMemcache: SessionsMySQL Read
Replication: Slaves on D-RAAS
Inbound Web
Shared Storage
{NetApp}
.
.
Caching Server
{Varnish}
Backbone Technology | Distributed RAAS Cluster v5 rev2
Caching Server
{Memcache}
Caching Server
{Memcache}
Clustered Pool
Firewall
Backup Services GW
Logging{Graylog}
External Monitoring
Internal Monitoring | Self healingAML
Services
Thank You Contact Details
- Kelcey Damage
- irc.freenode.net #cloudstack id:kdamage