Deck 1 Flashcards
Default time period that CloudWatch metrics are stored
2 weeks
Metrics monitored on hosts
CPU
Network
Disk
Status
How to extent the retention of metrics past 2 weeks
GetMetricsStatistics API or with 3rd party tools
How long are terminated EC2 and ELB metrics stored for
2 weeks
System status checks versus Instance checks
System checks are the physical layer and instance is at the EC2 layer
Systems check information includes what?
Loss of network
Power
Hypervisor issues
Hardware issues
How to resolve system issues
Stop and start the instance
Instance status checks include what?
Failed system checks Misconfigured networking or startup config Exhausted memory Corrupt File System Incompatible kernel
Best way to resolve instance issues
reboot
How many IO credits do you get on a GP2 volume
5,400,000
How many minutes will it take to use 5,400,00 IO credits at max burst performance of 3000 IOPS on a GP2 volume
30 minutes
What is happening to your credit balance when you are not bursting IOPS on a GP2 volume
You will earn credits
What is the best way to prewarm a normal (new) EBS volume
Write to all blocks
What is the best way to prewar an EBS volume based on a snapshot
Read all blocks
For Memcached at what CPU utilization is recommended to autoscale
> 90%
Is Reddis or Memcached multithreaded?
Yes Memcached is.
How to determine when you should scale a Reddis for CPU utilization
90% / # of cored
I.E. 90% / 4 = 22.5%
What parameter should you increase if Memcached exceeds 50MB of swap usage
Memcache_Connections_Overhead parameter
What does the Memcache_Connections_Overhead parameter specify?
The amount of reserved memory for connections and other misc overhead
What is the Reddis parameter for reserving memory
Reserved_Memory
What can be done to to reduce the amount of evictions for Memcached?
Scale up (increase memory) or Scale out (add more instances)
How to reduce the amount of evictions on Reddis?
Scale out (Adding more replicas)
What would a large spike with sustained concurrent connections mean for Memcached and Reddis?
Maybe not releasing connections as it should be
What region must be used for billing alarms
US-East
Elasticity on EC2
Instance size
Scalability on EC2
Increase of # of instances bases on autoscale
Scalability on DynamoDB
Unlimited storage
Elasticity on DynamoDB
Increasing IOPS for spikes and then decreasing
Scalability on RDS
Increasing instance size from small to medium or similar
RDS Elasticity
RDS is not elastic in nature since it cannot scale on demand
How many IGWs can there be per VPC
1
How to increase network throughput on a instance
Create a larger instance - scale up
How to scale networking if instance type is maxed
Add more instances - Scale out
How to provide more NAT throughput on a network
Add more instances - Scale out
How much of a performance hit will you take by not reading all blocks first (initializing) an EBS volume created from a snapshot?
5-50%
EBS Cloudwatch metric that indicates the operations waiting to be completed
Volume Queue Length
Volume status check that indicate the EBS volume is stalled; not available (severe)
Impaired
Volume status check that indicates that EBS volume is degraded (performance impacting)
Warning
