When D2L first moved to the cloud, we were concerned about being locked-in to one cloud provider. We were compelled to explore the opportunities of the cloud, so we overcame our perceived risk, and turned it into an opportunity by self-rolling tools and avoiding AWS native services. In this session, you learn how D2L tried to bypass the lock but eventually embraced it and opened the cage. Avoiding AWS native tooling and pure lifts of enterprise architecture caused a drastic inflation of costs. Learn how we shifted away from a self-rolled "lift" into an efficient and effective "shift" while prioritizing cost, client safety, AND speed of development. Learn from D2L's successes and missteps, and convert your own enterprise systems into the cloud both through native cloud births and enterprise conversions. This session discusses D2L’s use of Amazon EC2 (with a guest appearance by Reserved Instances), Elastic Load Balancing, Amazon EBS, Amazon DynamoDB, Amazon S3, AWS CloudFormation, AWS CloudTrail, Amazon CloudFront, AWS Marketplace, Amazon Route 53, AWS Elastic Beanstalk, and Amazon ElastiCache.
12. The dreaded Vendor Lock-in
And here we see it, changing the code to handle S3
instead of CIFS shares. Think of the poor developers!
Yes, let’s think of them… Math time!
ROI = <Estimated Cost of Dev Effort>
---------------------------------------------------------------------
(<OpExold> - <OpExnew>)
(Watch your units)
15. DNS
“Transforming the way the
world learns”
Not: “implementing DNS”
Amazon
Route 53
$
Number of domains
16. DNS
“Transforming the way the
world learns”
Not: “implementing DNS”
Amazon
Route 53
$
Number of domains
17. Caching
“Transforming the way the
world learns”
Not: “re-implementing
Memcached”
Amazon
ElastiCache
$-
$1,000
$2,000
$3,000
$4,000
$5,000
$6,000
$7,000
$8,000
$9,000
Number of instances
18. Scaling and Opportunity Cost
If D2L were to have 1 DNS
specialist that would be like
Amazon having
Classic Build vs. Buy pattern
Better integration into the rest
of the ecosystem
Just superior feature sets
19. The Cloud is NOT a
Hosting Facility in the Sky
Logging Errors Amazon Kinesis
Firehose
Amazon
Kinesis
Amazon
Elasticsearch Service
20. SQL as a Logging Database – Legacy Design
• Unpleasant to search
• Only 30 days retained
• Hardcoded sharding
• Costly
50
100
150
200
250
300
350
400
450
500
550
600
650
700
750
800
850
900
950
1000
$
TB
21. Kinesis Firehose -> Elasticsearch
(with S3)
• Designed for search
• 12 months retained
• Could retain more
• Sharding by choice
• Cheaper (~1/4th the cost)
50
100
150
200
250
300
350
400
450
500
550
600
650
700
750
800
850
900
950
1000
$
TB
Amazon Kinesis
Firehose
Amazon
Elasticsearch Service
Amazon
S3
30. Recap
• Who’s afraid of EC2?
• Don’t forget to account for all costs
• Don’t compete outside your core competency
• The cloud is not a Hosting Facility in the Sky
• PaaS > IaaS
• This AWS thing might stick around for a bit
33. Amazon EFS
File
Amazon EBS
Amazon EC2
Instance Store
Block
Amazon S3 Amazon Glacier
Object
Data Transfer
AWS Direct
Connect
AWS
Snowball
ISV Connectors Amazon
Kinesis
Firehose
S3 Transfer
Acceleration
AWS Storage
Gateway
Storage is a platform: AWS Storage Maturity
34. AWS Database
Migration Service
Start your first migration in 10 minutes or less
Keep your apps running during the migration
Replicate within, to, or from Amazon EC2 or Amazon RDS
Move data to the same or different database engineAWS Schema
Conversion Tool
35. AWS Application Discovery Service
Identify application
Inventory
Map application
dependencies
Baseline system and
process performance
Automate data center application discovery
36. AWS Server Migration Service
• Support VMWare VMs migration
• Agentless VM Migration
• Capture incremental changes
• Migrate a group of VMs
• Management Console/API Access
• Launch EC2 instances from AMIs
37. Thank you!
Ben Snively (AWS)
snivelyb@amazon.com
Stephen S. Skrzydlo and Stan Przychodzki
Stephen.Skrzydlo@D2L.com Stan.Przychodzki@D2L.com