A Day in the Life(cycle Manager)

We are often asked why our product focuses on disaster recovery across regions and does not also manage traditional local backup. Three words: AWS Lifecycle Manager.

For small and medium businesses with modest cloud infrastructure and requirements, straightforward hourly and daily snapshots should suffice to protect mission-critical instances against corruption or accidental deletion, much like you might backup a Mac laptop with Time Machine.

This is precisely what Lifecycle Manager does, so it would be pointless to compete against a native AWS service. For more sophisticated requirements, such as providing database checkpointing scripts, requires a more sophisticated product that is outside our scope or that users can implement themselves with a quick Internet search.

With Lifecycle Manager if an instance becomes corrupted or infected with a virus, merely power it down, remove the volumes, and replace them with volumes created from a snapshot. No additional management software is necessary.

You can even create an entire AMI image regularly — in case of individual file restore just launch a separate instance and grab files from the AMI backup that you need.

Aha, you say, you can also remotely copy AMI images to a different region, so why the need for DR automation software. One word: testing.

Restoring a corrupted EC2 instance in the same region is straightforward as all of the information you need (e.g. device name, or instance size if restoring from an AMI) is at your fingertips in the form of the original instance.

For DR in case of a failover due to an outage at the primary region there is much more than just creating a new instance from the image. What is the size of the instance? What are the security group configurations? Is there a specific subnet to which it should be attached? Placement group? NAT Gateway?

None of that information is at your fingertips … it was present only at the production region, which is inaccessible. You could write it down somewhere, but of course we all know those documents quickly grow stale.

Even with a comprehensive failover runbook, how often do you test it to make sure you can recover your mission-critical workload in case of an outage. Maybe an annually DR drill, but do you QA your code once per year? With any testing, the more the better.

To do disaster recovery preparation correctly in the past cost a lot of resources: either your time, or a lot of money for complex automation software.

Thunder for EC2 Serverless is the best of both worlds. It automates the advance provisioning of duplicate EC2 instances with the proper settings; it performs differential, periodic snapshot replication across regions and attaches the up-to-date volumes to the duplicate instance so it can be merely powered on in case of a disaster. And it automates regular testing of the backup instances: at a minimum powering them on and off, with software for application-level testing such as connecting to databases or webservers in order to confirm the application hosted in the instance has recovered.

All of the functionality is 100% cloud-native: an ultra-inexpensive Lambda function drives the provisioning and replication, CloudFormation drivers the configuration, and CloudFront presents the current status on any device through a signed URL. CloudWatch Events and CloudWatch Logs manage the scheduling and logging.

Best of all is the price: a one time $399 fee with no ongoing subscription costs. We estimate the Lambda execution costs will be roughly $30 per year, less per year than a dedicated SaaS EC2 instance solution costs per month. And with a simple CloudFormation interface through the AWS console, there are no separate users or security policy to implement other than how you manage your AWS infrastructure in the console today.

The new way to do DR “right” in AWS:

• Purchase Thunder for EC2 Serverless for one-time $399 fee (contact us at info@thundertech.io for details, Lambda functions are not yet for sale on AWS Marketplace)
• Configure the product through CloudFormation in a few clicks and provision duplicate, offline instances in your DR region, as well as set periodic schedules, again through CloudFormation; the function will cost roughly $30 per year; you will be charged by AWS for data replication at $0.02 / GB but that is the same as any solution, including Lifecycle Manager
• Configure deep testing of common applications (again through CloudFormation, also implemented as Lambda functions) or write your own simple ones; these tests are called after every replication job, set and forget; security is granted only through an instance’s private network
• In case of the unthinkable, a true outage at the AWS region hosting your product, merely navigate to your DR site and power on your pre-configured, fully tested backup instances and continue your business, having spent a tiny fraction of your monthly AWS spend on this insurance and almost none of your precious time. Anyone can do this; no need for a sophisticated user to be available to walk through a run book.

Or you can try to write it all down somewhere and remember to take time out of your day regularly to make sure it works. Conversely you can pay at a minimum $199 / month plus EC2 hosting fees plus the time to configure and secure a new product and new user interface from our competitors, adding up to at least $10,000 over a 5-year period.

With Thunder for EC2 Serverless you get automation and a low price, arguably the best way to do DR. For more information contact us at info@thundertech.io or try it yourself with our hands-on demo at https://thunderdocs.s3.us-west-1.amazonaws.com/demo/index.html