While virtual servers offer a number of functions that set them above and beyond their physical counterparts, there are a number of things that virtualized environments must have in the same way as a physical machine. Even a virtual server does have a physical location somewhere; it is simply virtual to the company that is using it from a great distance away. As such, the data on the server needs to be backed up in order to be safe and secure, but traditional methods for backup have proven to be problematic.
When VPS options first emerged, standard practice was to stick a backup agent on every single virtual machine. This meant for an entire server, hundreds or even thousands of backup agents existed and when a massive, server-wide backup was done, the drain on the I/O would cause huge amounts of access slowdown and choking on other virtual machines on the server, whether they belonged to a company or not. In addition, an installation of so many agents necessitates their re-installation every time a new patch or better version comes out.
Some hypervisors allow the installation of backup agents on the host itself, but had the problem of being “crash consistent,” that is, they would cause a crash of the server every time they were run. This would produce a large scale backup but would mean any data cached in memory would be lost. Now, options such as VM snapshots and block-level incremental backups are being rolled out that access the storage medium of the VM itself to limit performance decreases and increase uptime.
No matter the type of backup chosen, however, a viable and reliable way to retrieve data is necessary, even on a virtual server.