The industry is in the midst of a shift from physical data center to virtualization — a transition that favors lightweight, agentless backup solutions over difficult-to-manage agent-based solutions. While there are several technology-driven reasons behind this movement toward agentless backup, there are equally as many cost- and efficiency-driven factors that are pushing the industry toward this innovation.
Legacy agent-based backup systems rely on software running on the production machine alongside the production application. These agents have to reach and be reached by their central data protection server in order for backups to occur. The agent is responsible for reading production data through the OS layer, processing, and sending backup data over the wire to a media server or backup server, where it is finally written to a (typically proprietary) backup storage device. Because the production machines and network are impacted, the backup process itself becomes a risk, constraining businesses to work within “backup windows”.
This design burdens the system and administrators with immense overhead:
- Resource Consumption That Impacts Production
- Agents generate CPU load on the production machine for read, processing, and write operations
- Agents consume memory on the production machine
- Payload traffic from agents to the media server consumes virtual and physical networking resources
- Agent traffic requires another network (and at least one port) on each production machine
- Administrative Burden
- Administrators have one more network to manage
- Tenant admins are required to install and maintain agents on each and every production VM (both existing and new), mandating a change in the way tenants operate and slowing them down
- Agent updates and overall lifecycle need to be managed
- Security & Administrative Risk
- Additional points of failure are introduced: What happens if the media server is inaccessible, or the network is congested?
- Agents require access to production data, typically at the application level, which presents a risk to the production machine’s integrity and health
By consuming machine and network resources, the agent interferes with normal execution, and as a result the backup process is only allowed to run during “backup windows” when the application is supposed to be idle. Not only does that assumption invalid for modern always-on businesses, it also limits RPO to (typically) once a day.
Agentless backup solutions, by contrast, do not rely on any agent inside the machine. Agentless backup solutions perform centralized, network-wide backup via APIs, simplifying the management and maintenance of your data protection solutions. There are many benefits to an agentless backup solution, including:
- Cloud friendly, allowing a non disruptive service introduction to all tenants
- Reduced administrative cost and resource consumption
- Drastically simplified maintenance because tenants no longer need to reinstall agents with every change
- Faster backup and restore operations with tighter RPO
The Trilio Solution
TrilioVault is agentless. It utilizes a lightweight Data Mover process at the compute node level, which does not require additional servers or resources, and avoids wasteful sending of data over the wire.
Data access is done at the storage level directly and outside of the production VM, so there’s no need to read and write data through the production machine’s operating system. The Trilio Data Mover minimizes network traffic by reading snap diffs and sending only changed blocks to the target storage of choice, using industry-standard NFS or S3 protocols.
Trilio’s agentless design is cloud friendly. Tenants are not disrupted by the introduction of the Trilio service into the cloud, and can start using it right away on their own, with no need for any cloud admin assistance. With increasingly virtualized environments, agentless backup reduces disruption and makes it easier to maintain cloud infrastructure that, while in flux, contains critical workloads in need of protection.