The core of the Plan B Service is a unique automated system that controls the copying or ‘snapshotting’ of protected servers, the secure transfer of snapshot data to a Plan B Recovery Platform, and the immediate ‘pre-recovery’ and functional testing of systems.
Protected Server Snapshotting
- Dedicated Plan B Appliance manages copying or ‘snapshotting’ of live servers
- Data transfers are minimised and can use relatively low bandwidth links
- Appliances retain snapshots providing a local file retrieval function
Plan B uses an on-site Appliance to take snapshots of a customer’s server(s) and to efficiently and securely transfer those snapshots off-site to the Plan B recovery platform.
Once set up, the Appliance will talk directly to the operating systems of each protected server and will request a single-state incremental copy or snapshot using the Microsoft VSS (Volume Shadow copy Services) component on each protected server. The Appliance stores the incremental snapshots and then works out in collaboration with the Plan B Master systems the minimum amount of unique data that needs to be uploaded to the recovery platform using an encrypted session.
Once transferred to a Plan B Master the server snapshots will be immediately ‘pre-recovered’ into fully working and configured system images.
Find out more about the ‘pre-recovery’ process.
Continuous Pre-Recovery of Systems
- Automatic pre-recovery of system images every night
- Updated snapshot data immediately converted and configured into bootable system images
- Plan B’s Master system configured to replicate all dependencies and enabled to reproduce secure network connectivity and switch DNS and MX records
All the work to change snapshot data of protected servers into images that will run natively on Plan B’s virtual server environment (known as Rescue Images) is done on a Plan B ‘Silo’, which runs part of Plan B’s Disaster Recovery platform.
The Disaster Recovery platform is a complex control and processing environment running across multiple ‘Silos’. The DR platform automatically processes snapshot data for protected servers, turning them into runnable ‘Rescue Images’, which it then continually tests to prove they are working. At the same time it automatically controls the entire process and maintains an understanding of the interdependencies and configuration requirements of every set of systems in Recovery. Finally, the Disaster Recovery platform is able to automatically set up appropriate VM environments, boot Rescue Images, and set up and configure customer specific networking, security and DNS changes for service invocations.
Once an Appliance has transferred new snapshot data to a Plan B Master system, the new snapshot data is automatically tagged for immediate processing into new operational Rescue Images. This is done by combining the new snapshot increments with the existing snapshot data to produce an up-to-date system image. This is then automatically converted and configured according to the specific requirements of the original protected server to run natively within Plan B’s virtual machine environment.
At this stage all changes are made to the unbooted server images to fix everything required to make those images work and run as carbon copies of the originals. This would include things like a physical-to-virtual conversion, fix up to the specific operating system files, addressing, licensing, application changes and system configuration amendments. A system specific ‘Test‘ agent is also injected at this stage for use in the later testing process. All these changes are done automatically to match the server’s specific operating system and application configurations. The objective is to produce a set of Rescue Images that are ready to boot up in a completely error-free state.
The final part of the daily automatic process is for all new Rescue Images to be physically booted in Plan B virtual machines and then tested to prove they are working and ready if a customer needs them.
100% daily Testing
- Continuous disaster recovery testing ensures systems will work when needed
- Complete system solutions are automatically booted and tested every night
- Any errors are reported for engineer investigation
It is a cornerstone of Plan B’s approach that no matter how good your technology and processes are, you can’t ever really know something will work unless you test it. So that’s what we do for every newly converted Rescue Image, and our testing is automated and an integral part of the process which can’t get forgotten or put off.
Proactive Disaster Recovery Testing
Plan B tests all new Rescue Images every time they are updated, which for standard systems would be at least once every day. Testing is done automatically by a Tester process that boots up newly created Rescue Images, along with the Rescue Images of any interdependent servers, in specifically configured virtual servers on the Disaster Recovery Platform.
The Test agents report on everything from the status of the machine, the operating system and the domain, the status of processes, services and applications (which could include the results of customer-specific synthetic application tests), to the content of the event logs.
All new Rescue Images that pass testing are moved to the Rescue Image Repository and marked as the ‘Last-Known-Good’ and these are the images that will be booted for a customer in the event of a disaster. All images in the Repository are therefore images that are ready to run and have been proven to work against the customer’s agreed criteria. All that is needed to invoke fully running recovered systems from this point is for the Plan B system to simply boot them up and automatically set up networking, security and DNS from pre-defined invocation profiles. Fully working recovered systems can therefore be ready within minutes.
Any test failures are investigated each day by Plan B’s Engineering Support team and immediate action taken to resolve any issues. Plan B will proactively contact the customer for any issues that need their intervention.
We believe that continuous disaster recovery testing within an affordable service is unique to Plan B. From our point of view, if you haven’t tested an image then you don’t know it will work. Given that we process failed images every day, if you were to select a service which did not offer continuous testing you would be taking a definite risk that your DR systems would not work easily, or at all, when you needed them.