What can be monitored with CloudWatch?
- AWS services
- Custom metrics generated by your applications/services
- Any log file your application generates
Can you assign a role to an instance after its creation?
No
Minimum granularity for CW custom metrics?
1 minute
Use case for every EBS volume type
- gp2: System boot volumes, virtual desktops, small-medium DBs, dev and test
- io1: Critical apps, large databases
- standard: data infrequently accessed, scenarios where cost is important
Volume size for every EBS volume type
- gp2: 1GB-16TB
- io1: 4GB-16TB
- standard: 1GB-1TB
Max throughput for every EBS volume type
- gp2: 160MB/s
- io1: 320MB/s
- standard: 40-90MB/s
IOPS performance for every EBS volume type
- gp2: baseline-3IOPS/GB(up to 10,000 IOPS) with the ability to burst to 3,000 IOPS for volumes under 1,000GB
- io1: Performs at provisioned level. Max 20,000
- standard: averages 100IOPS with the ability to burst to hundreds
Metric to check if you have the correct amount of IOPS (exemple: if your DB is slow)
VolumeQueueLenght: Number of read and write operation requests waiting to be completed in an specific period of time.
EBS posible volume status
ok: normal
warning: degraded/severely degraded
impaired: stalled/not available
insufficient-data:Insufficient data
RDS metric for number of DB connections in use
DatabaseConnections
RDS metric for number of IOs waiting to access the disk.
DiskQueueDepth
RDS metric for Amount of available storage space
FreeStorageSpace
RDS metric for Amount of time a read replica DB instance lays behind source DB
instance.
ReplicaLag
RDS metric for Average number of disk read and write IOPs/second
ReadIOPs, WriteIOPs
RDS metric for Average amount of time for reading and writing per IOP
ReadLatency, WriteLatency
ELB metric for total number of requests that are pending submission to a regitered instance
SurgeQueuelength
ELB metric for count of total number of requests that were rejected due to the queue being full
SpilloverCount
Important things related with monitoring ElastiCache
- CPU utilization
- SWAP usage
- Evictions
- Concurrent connections
Cause for your monitoring not reaching EC2 instances
You should allow ICMP or specific port
EC2 instance types
- On demand
- Reserved
- Spot price
How much you can save using reserved instances
Up to 75%
Average savings when moving down between EC2 instance types
25-50%
What is ASG’s warm up period
Period for the instance to start participating to the metrics
What is ASG’s cooldown period
Amount of time between autoscaling events
