Hyper-V replication across two sites: recovery in under 30 minutes

A service company with 20 to 49 employees now replicates its virtual machines to a second site every 5 minutes and has upgraded its servers and site-to-site connection from 1 to 10 Gbit. In the test, operations were running again at the second site after under 30 minutes, and the nightly backup now takes only about 1 hour instead of about 6 hours.

At a glance

Customer
Not named (anonymised case study)
Size
20 to 49 employees
Region
Berlin and Brandenburg
Period
2024 to 2025
Type
Project with subsequent ongoing support
Technology
Hyper-V with Hyper-V Replica, backup with an immutable copy and a cloud copy, network with 10 Gbit

Starting point

In the server area and between the company's two sites, the network ran at 1 Gbit. The nightly backup took about 6 hours. There was no fallback site: if a server failed, operations came to a standstill. Servers and operating systems were approaching the end of manufacturer support, the backup took too long, and there was no copy that cannot be altered to protect against ransomware.

Objective

In the event of a failure, the second site was to take over operations, with as little data loss as possible. The backup was to run faster and be stored in several locations, and the changeover was not to disrupt the employees' work.

Approach

  1. Network upgraded to 10 Gbit

    In the server area and between the two sites, the network now runs at 10 Gbit instead of 1 Gbit.

  2. Replication to the second site

    Up to 10 virtual machines are transferred to the second site every 5 minutes with Hyper-V Replica. In an emergency, this means that at most about the data of the last 5 minutes is lost.

  3. Backup in several locations

    One copy of the backup is stored at the second site, another copy is immutable, and an additional copy is stored in the cloud.

  4. Changeover during ongoing operations

    The systems were changed over through migration and upgrades during ongoing operations. Interruptions occurred only outside the company's core working hours.

  5. Failover test

    The failover to the second site was tested. Operations were running there again after under 30 minutes.

Result

If one site fails, the company can continue working at the second site. The nightly backup is completed in about an hour and is also stored at the second site, as an immutable copy and in the cloud.

  • under 30 minutes until operations were running again at the second site in the test. Period: failover test 2024. Source: measurement by conpro IT during the test.
  • at most about 5 minutes of possible data loss in an emergency, thanks to replication every 5 minutes. Period: 2024 to 2025. Source: replication setting.
  • from about 6 to about 1 hour duration of the nightly backup before and after the project. Period: 2024 to 2025. Source: measurement by conpro IT before and after the project.

View all project case studies