IT’s shift to the cloud: Veeam’s data protection report in detail

Workloads split between cloud and on-prem

In four years of the Veeam survey, while also asking about future plans, the clear trend has been towards a fairly even split between the cloud and on-premise as a location for server processing.

In 2020, 38% and 30% of physical and virtual servers respectively were run on-site, with about another third (32%) running in the cloud.

That has since dropped to just under half being run on-site (24% physical and 24% virtual) and 52% in a cloud hyperscaler. 

Backup a good fit for the cloud

Mirroring the trend towards some server capacity being hosted in the cloud, backup is also heading that way.

In 2020, the split was 60/40 between backup using on-prem tools only and use of the cloud. By the time respondents were asked to predict that split in 2025, those proportions come out as 74/26, but in favour of the cloud.

Heavy dependence on the cloud is also in evidence when it comes to disaster recovery and how organisations recover from outages.

Here, 48% recover to the cloud, either from cloud-resident backups (31%), to DR-as-a-service from a managed service provider (7%), or to cloud-hosted servers from cloud-based replicas.

Meanwhile, customers that recover to on-site servers from the cloud make up 38% of the total. Only 16% of restores have no cloud involvement, recovering from on-site backups to on-site servers.

Further cloud preference emerged when the Veeam survey asked respondents what they think is the most important attribute of an “enterprise backup” solution, protection of infrastructure as a service (IaaS) and software as a service (SaaS) came out on top.

The next most desirable attributes were to be reliable and manageable across multiple geographies, but fourth was the ability to protect enterprise applications such as SAP HANA and Oracle. It’s quite possible this reflects the split between on-prem and the cloud noted above.

And, to put a finer point on this in a backup context, it probably shows that for lots of data – effectively secondary data and rarely, if ever, accessed again – the cloud is a good fit if data is held in cheap, “cold” tiers. But there are some very important applications that must run on-site and customers would like unified solutions that can handle both.

Cyber attacks most impactful cause of outages, but not the only one

When the survey asked about the most common cause of outages, and the most impactful cause, ransomware came out on top. 

But it’s not the only fly in the IT ointment. Others rank fairly closely as the most impactful cause of outages. These were ranked as: networking and infrastructure failures; accidental deletions and data corruption; software failures; cloud service outages; and server hardware issues.

Storage hardware problems were the most impactful cause of outages for 8% of respondents in each year surveyed.

The frequency of outages is potentially quite staggering, with 28% of servers having had one outage in 2022. Most respondents (24%) said between 5% and 10% of their servers had gone down unexpectedly in the past 12 months.

Container workloads: responsibilities and backup uncertainty

As a harbinger of the cloud-native era, containers are now in production use – 65% in 2022 – by a significant proportion of respondents.

But when asked who is responsible for data protection of container workloads, it looks like the task hasn’t found a permanent home or go-to role. Here, we find responsibility split almost evenly between storage admins (26%), app admins (26%), Kubernetes admins (25%) and the backup team (23%).

Also, when the survey asked about how customers back up Kubernetes data, the picture shows a similar lack of certainty. Most back up the storage the data lives in (47%), while 22% back up the underlying database components. Only 26% use a third-party backup tool.