[EC2 ICE] Is c5a series more prone to capacity issues? Is there an EC2 capacity heatmap?
Today I encountered Insufficient Capacity Error whilst launching 3x c5a.8xlarge in Sydney region (AZ apse2-az2). After some re-trying I managed to launch c5a.12xlarge in the same AZ. A month or two ago I encountered the same issue with c5a.8xlarge in Mumbai region. Managed to switch some instance to c5 (without the a).
My usecase is online events. I need to spin up a number of EC2 instances on short notice. Max so far has been about 10 EC2 instances (I don't think that a big number for AWS). I am using different regions depending on location of the end user. I am now using c5a series (usually c5a.8xlarge) instead of c5 to save on cost. However if another instance series is more reliable (i.e. not getting ICE'd) I would consider changing to a different series. Is the c5a series more prone to ICE?
Is there a 'capacity heatmap' available, where I can see the EC2 availability across instance series, regions, and AZs? Ideally I would have access to a 'weather forecast' where I can see the current & expected capacity forecast. In my current architecture I can make changes to instance series / AZ a few days in advance of an event, but it's stressful to make changes within the hour before my customer's event starts.
Hello,
Thank you for reaching out. As you properly stated, the Insufficient Capacity errors mean that at a given moment, AWS lacks capacity for on-demand Instances in an specific Availability Zone when you tried to start or launch an Instance (1).
On-Demand capacity fluctuates overtime, so waiting few minutes, changing Instance type temporary or (when launching an Instance) select a new Availability Zone where capacity is available. You can use the AWS CLI command describe-instance-type-offerings to validate the Instance types you can launch in an Availability Zone (2-3).
If your operation is critical, your team may consider also take a look to On-Demand Capacity Reservations so you have specific compute capacity allocated in specifics Availability Zones. Note that there is a cost associated with the use of Capacity Reservations (4-5).
Unfortunately, there is no roadmap as to how more resources are made available over time. As to which Instances are prone to be more likely to encounter this issue, it all depends on the resources available c5a.8xlarge does require 32 vCPUs. If you take that multiplied by x number of Instances, they add up for sure.
I understand you were looking for a more suitable answer, in AWS our customers are our priority and we constantly keep on adding more resources to alleviate this kind of issue from happening.
Have a nice day!!
References
(2) describe-instance-type-offerings
(3) Find an Amazon EC2 instance type - (Find an instance type using the AWS CLI
(4) On-Demand Capacity Reservations
Relevant questions
Experiencing InsufficientInstanceCapacity despite using Capacity Reservation
asked 6 months agoCheck instance capacity
asked 18 days ago"Failed to start the instance. Insufficient Capacity" error
asked 4 months agoInsufficient Capacity
asked 10 months agounused reserved capacity with regional reserved Instance
Accepted Answerasked 6 months ago[EC2 ICE] Is c5a series more prone to capacity issues? Is there an EC2 capacity heatmap?
asked 19 days agounable to delete RDS database cluster due to insufficient storage capacity issue
asked a month agoInsufficientDBInstanceCapacity Sydney Region
asked 3 years agoRDS t3.small multi-az capacity in us-east-1
asked 3 years agoAre there capacity issues for g4dn instances in eu-west-1?
asked 8 months ago