Ahh, CDP (Continuous Data Protection) – the feature that everyone has been waiting for. If you’re new to Veeam, or out of the loop, CDP was first announced way back at VeeamON 2017. There’s been a lot of hype about the feature, because in a nutshell it allows you to get RPOs down to a matter of seconds. It is also able to accomplish this without using snapshots. Sadly, it has taken a long time to come to market. The good news is that the Vanguards saw a functional demo of this at the Vanguard Summit recently.
WHAT IS VAIO?
CDP leverages VMware’s vSphere APIs for I/O Filtering (VAIO), which was first released in vSphere 6.0 Update 1. This API effectively gives 3rd party vendors (i.e. Veeam) the ability to get Guest OS data without a) impacting performance or b) needing to install any sort of guest agent. The way it works is the filter lives within the ESXi host, which allows it to intercept data from the Guest OS as it is being written to the VMDK. Data which is passed through this filter can then be used either for caching (thus providing better performance), or for replication. Needless to say, Veeam is using it for the latter. What is worth noting is the filter sits in user space, which means if something were to go wrong, it will not crash the ESXi host.
WHAT DOES CDP BRING TO THE TABLE?
Lot’s of organizations are already setup with regards to incorporating replication into their DR planning. Heck, Veeam’s flagship product, Veeam Backup & Replication supports it right out of the box (it’s in the name!). Along with most SANs supporting array-based replication, or products like VMware’s Site Recovery Manager, where does CDP fit in?
In my eyes, one of the biggest differentiators for CDP is its ability to get replication down to a few seconds. Although array-based replication can be quite versatile, it can involve expensive licenses, expensive hardware (i.e. another SAN to replicate to), potentially some bandwidth concerns, etc. Similarly, if you are using any form of snapshot-based replication, you are once again potentially dealing with issues such as performance or storage requirements. CDP eliminates these bottlenecks.
WHAT WILL CDP LOOK LIKE?
Workloads being protected with CDP will be setup as a continuously running job, similar to how SQL Server jobs can be configured in Veeam Backup & Recovery. They’ll continually monitor the VM and depending on your replication schedule, the data will be sent across the wire to the receiving host. You’ll have access to SLAs and other metrics to make sure you are meeting your requirements.
Visually you will also be able to see a granular slider indicating timestamps for each replication “queue”. Once again, this is similar to the slider bar you see in VBR when restoring a SQL server via the transaction logs. It is worth noting though that you will not be able to “fork” a copy of the VM from one of these replication queues. That is to say, the data is used for replication and not for backing up / restoring the VM.
COOL – WHEN CAN WE USE IT?
CDP has been facing delays for awhile now; if you subscribe to Anton Gostev’s weekly digest, you may recall awhile back he mentioned that there were issues with performance, particularly when trying to scale the solution. So, in true Veeam fashion, it will be released when it is ready, with no firm timeline. Although most of us don’t particularly care for that answer, I would counter that with the fact that generally speaking, Veeam’s QA has traditionally been some of the best in the business. So stay tuned, and hang tight.