AWS Atlanta Meetup -AWS Spot Blocks and Spot Fleet
Download as PPTX, PDF1 like443 views
The document discusses various pricing models for AWS EC2 instances, including on-demand, reserved, and spot instances, highlighting the significant cost savings associated with spot instances that can be as much as 90% off on-demand prices. It explains the flexible nature of spot instances, their management through spot fleets, and best practices for maximizing their use while minimizing interruptions. Additionally, it provides step-by-step instructions for creating and managing a spot fleet within the AWS environment.
AWS Atlanta Meetup -AWS Spot Blocks and Spot Fleet
- 4. Significant discount compared to On- Demand instance pricing. Pay for compute capacity by the second or hour with no long-term commitments On-Demand Spare EC2 capacity for up to 90% off the On-Demand price Reserved Spot Steady state applications or predictable usage, databases. For fault tolerant, instance flexible or time-insensitive workloads. For spiky workloads or to define needs initially.
- 5. You can group your accounts into organizational units (OUs) and attach different policies to each OU. On-Demand (2006) Reserved Instance (2009) Spot Block (2015) Spot (2009) Spot Fleet (2015) No Upfront RI (2014) All Upfront RI (2014) Partial Upfront RI (2014)
- 6. The prices changes infrequently based on supply and demand of spare capacity Just request capacity and pay the current rate.When AWS needs the capacity back they give a 2 minute warning.
- 7. Each Instance Family Each Instance Size Each Availability Zone C3 8XL 4XL 2XL XL L 1A 1B 1C On Demand $1.76 $0.88 $0.44 $0.22 $0.11 In Every Region Is a separate Spot Market $.50 $.21 $.08 $.27 $.29 $.04 $.01 $.30 $.16 $.07 $.05 $.01 $.08 $.04 $.04 * Pricing shown is for example purposes only
- 8. Example CLI command Spot instances with a specified duration are known as Spot Blocks. They are designed not to be interrupted and will run continuously for the duration you select You can specify a duration of 1, 2, 3, 4, 5, or 6 hours. The price you pay depends on the specified duration. aws ec2 request-spot-instances --instance-count 5 --block-duration-minutes 120 --type "one-time" --launch-specification file://specification.json
- 9. Get Diversified Resources Diversify your fleet. Grow your availability AND Get the Best Price Find the lowest priced instance power that works for you or LaunchThousands of Spot Instances With one RequestSpotFleet call Apply Custom Weighting Create your own capacity unit based on your application needs
- 10. • Multiple EC2 Spot Instances are Selected • Multiple Availability Zones selected • Pick the instances with similar performance characteristics: eg c3.large, m3 large, m4.large, r3.large, c4.large Spot Fleet
- 11. capacityOptimized The Spot Instances come from the pool with optimal capacity for the number of instances that are launching. diversified The Spot Instances are distributed across all pools. lowestPrice The Spot Instances come from the pool with the lowest price. This is the default strategy . InstancePoolsToUseCount The Spot Instances are distributed across the number of Spot pools that you specify.This parameter is valid only when used in combination with lowestPrice
- 12. https://aws.amazon.com/ec2/spot/instance-advisor/ The Spot Instance advisor helps you determine pools with the least chance of interruption and provides the savings you get over on-demand rates.You should weigh your application’s tolerance for interruption and your cost saving goals when selecting a Spot instance.The lower your interruption rate, the longer your Spot instances are likely to run.
- 13. • When the Spot price exceeds your bid price, the instance will receive a two- minute warning • Check for the 2 minute spot instance termination notification every 5 seconds leveraging a script invoked at instance launch.
- 15. Cannot be stateless and fault tolerant? Running a Monolith or in-mem cache on the instance? Running Dev/Test on Spot could still be successful as an interim step, risk vs cost calculation, choose low interruption rate instance types Running periodic benchmarks or load tests - Use Spot Blocks To ensure that load does not decrease against app under test when Spot is interrupted Jenkins Build Slaves - use Spot Fleet Plugin Jenkins will automatically increase Spot Fleet target capacity upon pending jobs and scale in when no queue
- 16. Your instance type requirements, budget requirements, and application design will determine how to apply the following best practices: Be Flexible about InstanceTypes Test your application on different instance types when possible. Because prices fluctuate independently for each instances type in an Availability Zone, you can often get more compute capacity for the same price when you have instance type flexibility. Request all instance types that meet your requirements to further reduce costs and improve application performance. Spot Fleets enable you to request multiple instance types simultaneously. Minimize the impact of interruptions . Amazon’s EC2 Spot’s Hibernate feature allows you to pause and then resume Amazon EBS backed instances when capacity is available. Hibernate is just like closing and opening your laptop lid, with your application starting up right where it left off.
- 17. CONTINUED Choose pools where prices haven’t changed much Because prices adjust based on long-term demand, popular instance types (such as recently launched instance families), tend to have more price adjustments. Therefore, picking older-generation instance types that are less popular tends to result in lower costs and fewer interruption. Similarly, the same instance type can have different prices in different Availability Zones. Start with LaunchTemplates sooner then later These will be required for new features.
- 18. https://www.meetup.com/AWS-Atlanta/contribute/ Photo by Buro Millennial from Pexels
- 19. 1. Open up the ElasticContainer Service Console using the link https://console.aws.amazon.com/ecs/ . 2. On the left hand menu under Amazon ECS click on Clusters 3. Now click on the blue Create Cluster Button. 4. Select the EC2 Linux + Networking template then click on the Next Step button at the bottom. 5. Enter the cluster name as SpotECScluster 6. Under the provisioning model - Select Spot 7. For the Spot Instance allocation strategy - choose Diversified 8. Choose an instance type(s) from the dropdown (suggested r3.large) and put in number of instances as 3 9. Under maximum price, check table - try to go for $0.02 per hour if you set (optional) 10. Leave the other values as default and then choose a keypair to launch the instances with 11. Choose create a new role for the ECS instances if you don’t already have a specified ECS role 12. Now click on the blue Create Button. 13. When the cluster completion creates you will be able to see the EcsSpotFleet logical ID 14. Copy this ID to your clipboard 15. Go to the EC2 services part of the Console https://console.aws.amazon.com/ec2/ Testing Spot Fleet
- 20. 15. Under Instances choose Spot Requests on the Left Hand menu 16. Paste the resource id from your clipboard into the search bar at the top 17. If you expand the request you can see the EC2 instances which are part of the request 18. Go back to the ECS console and view your cluster We will schedule a task 19. Click on the Services tab 20. Click the Blue Create tab under the Services tab 21. On the Configure Service screen - choose EC2 as the Launch type 22. Choose yourTask Definition and revision 23. Choose the SpotECScluster from the Cluster name Dropdown list 24. Change Minimum Healthy Percent to 50 (leave everything else defaults) 25. Click Next Step 26. Choose None for load balancer 27. Click Create Service 28. Go back to the EC2 Console -> Spot Requests and select the current Spot Fleet 29. At the top of the menu SelectActions -> Modify target Capacity 30. Set the new target capacity to 2 (from previous value of 3) , Click Submit Testing Spot Fleet
Editor's Notes
- #9: You simply submit a Spot instance request and use the new BlockDuration parameter to specify the number of hours your want your instance(s) to run, along with the maximum price that you are willing to pay. When Spot instance capacity is available for the the requested duration, your instances will launch and run continuously for a flat hourly price. They will be terminated automatically at the end of the time block (you can also terminate them manually). This model is a good for situations where you have jobs that need to run continuously for up to 6 hours.
- #17: Either do Whitelisting or Blacklisting - but don’t do both
- #18: Either do Whitelisting or Blacklisting - but don’t do both
- #20: Even though you access the hub console from Oregon, you can move into any region as long as the migration tool supports it.
- #21: Even though you access the hub console from Oregon, you can move into any region as long as the migration tool supports it.