I had an opportunity to get briefed on the product status for Druva ( Druva.com ) recently. They’ve taken the approach to DR/Backup in a completely different direction. I found myself, after seeing so many differing solutions, and so many different approaches to what is probably the hardest single aspect of business continuity in play today, completely compelled. Not only does the product seem very interesting, but the presentation by my friend, Mr. Backup, W. Curtis Preston ( @WCPreston ). Certainly, he wanted to present to a friendly audience, but also, he wanted to get my take on this product’s approach, which is not the most public of products.
Data Management, Copy Data Protection, etc., wherein the metadata is critical to that data is the most critical aspect of a distributed backup product is the new hotness in the space. Regardless of where that data resides, the metadata will point the infrastructure to that data, and thereby make the retrieval of that data a far less painful process. The methodology by which Druva tracks, stores, and allows for quick access to that metadata is key to this process. It’s key to most of this category’s space as well, but Druva takes a different approach, which I’ll outline a bit later.
In this space, there are many competitors, from the Data Domain, Actifio, Rubrik, and Cohesity model of data management, all good solid competitors in the space, but each requires a hardware investment in a product, usually a server-based engine, or a converged node of compute with storage, Druva takes a different approach.
The Druva model relies on a very lightweight client which runs on the endpoint, (server or workstation, Windows or Linux), which in-turn pushes data to an AWS S3 based infrastructure, which has the ability to archive off to Glacier that long-term untouched data. While an appliance could be implemented (virtual) on-premises to maintain the hottest data, it may not be entirely necessary. The solution is very flexible, in regards to the RPO/RTO requirements of the customer.
Some of the secret sauce built into the Druva solution is the capacity to scale the AWS based servers dynamically to accommodate for growth within the dataset, and to achieve the capacities required by that data.
Certain categories, like the unpredictable workload inherent to certain applications make the design of an infrastructure quite difficult to design, the growth of data, particularly in major infrastructural patterns, such as acquisition can create licensing nightmares for traditional infrastructures, think licensing… However, truly cloud-native/SaaS type applications, while these still encounter workload unpredictability and growth, are handled quite differently by the dynamic abilities of the scaling functionality.
The scaling relies on a secret sauce from Druva which allows for VM’s on AWS to be orchestrated upward, and also turned off as their need determines. Metadata, leaning on the DynamoDB database, also stored in highly scalable format on AWS, and even has the ability to backup full VM’s from the on-prem workloads.
One of the benefits of the Druva architecture is that it’s so comparatively low-touch. You’ll never need to see or manage the backup nodes. Costs are maintained, and well monitored automatically. Due to the parallelization there’s no need for large VM’s to be built to run the backup from the AWS side, but rather many tiny VM’s, which again will be shut down as their requirements drop off. And, from a financial perspective, it can be an all OpEx model, unless the on-prem component is determined to be a requirement. Another way in which Druva has the ability to keep costs down is due to their relationship with Amazon, as they are consistently amongst the top three partners on S3 storage. They’ve negotiated better storage pricing for their customers than just about anyone else.
Another problem we face on a regular basis in today’s cloud enabled approach, is the backup of SaaS applications and native cloud workloads such as EC2 and VMware on AWS (VMC). Druva can take care of all types of data sources, including mobile devices (Windows, Linux, MacOS, iOS and Android), SaaS like O365, G-Suite, Box, and even SFDC. It can handle virtualized machines from VMware, Nutanix, and Microsoft HyperV, physical machines from Linux, and Windows based VSS aware machines. It handles file shares that are CIFS based, or NFS based. It can even assist in the migration from one NAS to another. They also the first cloud platform to be certified to protect VMC, and they support native EC2 data protection..
Overall, I was impressed by how Druva has taken the challenges surrounding data management, and applied the most current approaches to these challenges not just for now but for the future concepts of cloudy infrastructures.