What are the advantages of Agent vs Agentless backup?
Which type of backup is better – agent or agentless? There's plenty of debate.
It's worth saying straight off the bat that there's no binary yes/no answer. Depending on your requirements, agent or agentless backup can work for you. At Databarracks, we use both to help make businesses resilient.
There are pros and cons to both approaches, which we'll try to clear up.
First of all, what's the difference between the two?
An agent is a small application installed on a server to do a specific job. In the case of backup, an agent will support certain applications on that server. Each agent supports a specific function – so you have multiple agents installed on each machine across a network.
Agentless backup is a bit misleading. There is an agent, but only one. It works on one selected machine to back up the entire network.
So, who can agent and agentless backups help, and how? We could write a dissertation on the various intricacies – here are some of the most relevant points.
Greater host visibility, immediately - Agent-based backup is normally loaded in the OS stack. This means more control and visibility of the host system. This is not immediately available to an agentless backup, which needs time to review the file system and decide whether to do incremental or differential backups.
Doesn't use up network bandwidth – Agent-based backups use local resources to pre-process and compress data. They then transmit data across the network to the storage device (NAS/SAN, local USB drive, or remote host). If your bandwidth is near its limit, agent-based backup will not hamper it.
Expensive – Many environments now have multiple virtual machines (VMs) per server. Buying and managing an agent for each physical and virtual server is difficult for many businesses.
They can harm your environment - Certain backup agents use a kernel level space object and may cause a server to crash (if they have a bug). When applications fail, agents have shouldered a lot of the blame, falling out of favour in some data centres.
Extra work for IT admins - Backup agents need to be deployed and managed, resulting in additional work for the administrator in charge. Think about maintaining dozens, hundreds or even thousands of servers – effectiveness and efficiency can be challenging.
No installation downtime – In terms of convenience, agentless backup is easier to implement; you don't have to reboot your system when installing your backup. There's no issue of downtime.
Everything is centralised – Managing agentless backup is far easier as everything is in one place. Today, often a lot of businesses have a hybrid environment – physical, virtual, cloud all interweaved. Having your backup solution in one place makes this easy to manage.
Works in a fully virtualised environment – Virtualisation eliminates the disadvantages of the agentless approach. All major processes are performed by the hypervisor.
One solution means money saved – Without agent licensing fees, a business can save a lot of money with agentless. Plus, it's easy to scale up; you simply pay as you go and the cost increases proportionally to your growth – usually, models are based on data storage. As there's only one agent, there is also lower CPU usage (although it is consolidated on the backup server).
Needs a fully-virtualised environment – In some cases, you can only use agentless backup for virtual machines. Physical machines still need agent-based backup. Some software, such as Asigra, can work on virtual and physical servers.
Limited custom VM code execution - Some custom code, such as pre- and post-backup scripts and truncation of Exchange or SQL logs, can't execute during agentless backup in a VM. Again, this is not universal – some software can execute these.
As we said at the start, there's no clear winner. Everything depends on what fits you as an enterprise – which is how it should be. The key things to keep in mind are cost, ease of implementation and management, and obviously security.