Notes

2 types of RDS monitoring
In cloudwatch you can monitor RDS by metrics

In rds itself you can monitor RDS by events

RDS Metrics to know(see AWS documentation for specifics):
DatabaseConnections
DiskQueueDepth
FreeStorageSpace
ReplicaLag(seconds)
ReadIOPS
WriteIOPS
ReadLatency
WriteLatency


Monitoring ELB
-Only reports when requests are flowing through the load balancer.
-ELB send metrics in 60 second intervals(only if there is traffic)
- See AWS documentation for ELB metrics(especially SurgeQueueLength & SpilloverCount)
SurgeQueueLength: count of the number of total requests that are pending submission to the registered instance
SpilloverCount: measures the number of requests that are rejected due to the queue being full
- It is common practice to build the Autoscaling around the SurgeQueueLength so requests don't get rejected.

Monitoring Elasticache
4 Important factors for monitoring elasticache:
- CPU Utilization
-Swap Usage
- Evictions
- Concurrent Connections

CPU Utilization
Memecached:
- Multi-threaded
- loads of up to 90%. If > 90 % add cluster.
Redis
-not mutithreaded
- to determine when to scale 90 / # of cores
example: cache.m1.xlarge node has 4 cores. Threshold for utilization would be 90 / 4 = 22.5%

SwapUsage
-the amount of swap being used
- typically the size of swap == size of ram

Memcached
- SwapUsage should be close to 0 and should not exceed 50mb
- if > 50mb increase memecached_connections_overhead parameter
- memecached_connections_overhead defines the amount of memory reserved for memecached connections and other misc. overhead
Redis
- no SwapUsage metric instead use reserved-memory

Evictions
Evictions occur when a new item is added and an old item must be removed due to lack of free space in the system.

Memecached
-No recommended setting. Choose threshold based of your application.
-Either scale up or (increase memory of existing nodes)
-scale out(add more nodes)

Redis
- no recommended setting
-Only scale out(add read replicas)

Concurrent Connections

Memecached & Redis
-No recommended setting
-if large and sustained spike in concurrent conns it could be large traffic spike or application is not releasing connections as it should be.

AWS Organizations

Account management service that enables you to consolidate multiple AWS accounts into an organization
- Consolidated Billing
-All features
Advantages
-1 bill per AWS account
- very easy to track charges and allocate costs
-volume pricing discount