Many customers are faced with the decision of whether to outsource their backups – either to a geographically separate, self-operated location or to the cloud. Based on experience, the following factors play a crucial role in this decision:
- Cost
- Bandwidth (important for RTO – Recovery Time Objective)
- Data privacy
- Trust in the provider
- Future-proofing with the chosen cloud provider
- Security (access and encryption)
- Emotional factor („I can’t see the data and devices anymore.“)
One point that I consider particularly important and often bring up with customers is the need for a well-defined concept. Based on the company’s requirements, the decision to migrate to the cloud or not should be carefully evaluated. It is often sensible and legally feasible to outsource data, but many customers are hindered by a personal bias – they need to shift their perspective.
Here’s an example of a change in thinking when it comes to the cloud:
Historically, many customers in this segment replaced their local infrastructure every 5 to 6 years, requiring a recurring significant investment. When a customer (often not even the IT department but rather the management) wants to compare the costs of outsourcing „servers“ to the cloud, it is often estimated to be multiple times more expensive. This is a common mistake, as it involves taking an on-premise hardware offering and adding it up with cumulative cloud costs – a big mistake! The cloud encompasses not only the costs listed in the hardware offering, including potentially software (e.g., operating systems) and implementation services but also all operational expenses (Opex). This includes maintenance costs (hardware maintenance is not needed in the cloud), power and cooling expenses, access control costs, etc.
Unfortunately, many customers are still overwhelmed by the shift in thinking that suddenly introduces monthly recurring costs that cannot be easily cumulated. Instead of, for example, hardware maintenance, the cloud should be continuously optimized, and resources should be fine-tuned to optimize costs.
I also believe that the disaster recovery (DR) strategy should be designed and planned with a cloud perspective. Just because a cloud may seem more cost-effective than provisioning hardware, this path is not always sensible for all customers. The following arguments could be against outsourcing to the cloud:
- Single Point of Failure: Internet connection (many customers have only one high-performance internet connection).
- Costs could be a constraint if configuration errors are made.
- Costs could escalate if falling into the trap of „we have unlimited storage there.“
- RTO (the internet connection may not be sufficient to meet a desired RTO).
- Lack of trust in cloud providers (depending on the size).
The following arguments could be in favor of outsourcing to the cloud:
- Immutability is possible.
- Flexibility (as there is often no longer a need to adhere to storage limits).
- Geographic separation (especially in the case of building-related incidents).
- No maintenance and support of hardware.
- Compliance with the 3-2-1 rule by breaking protocols.
Important: I assume here that the primary repository is not being outsourced to the cloud but rather a secondary copy of the data is being considered.
In general, I recommend creating a concept first and defining all necessary values (e.g., RTO and RPO) within the company. Then, evaluate whether outsourcing a copy of the data to the cloud is possible and sensible or not. Deciding based on a single factor (e.g., cost) is not productive.
The following cloud providers are often considered for DR outsourcing:
- IONOS Cloud
- Wasabi Cloud (even with immutability, very attractive in terms of price-performance)
- Amazon S3
- Microsoft Azure BLOB
- Google Cloud Storage
A final thought:
A company with only one data center may benefit from the „cloud“ construct when initially outsourcing backup data to the cloud. Why? Consider a scenario where this company experiences a building-related incident, and the primary data center becomes permanently unusable. In such a situation, the customer can take advantage of the cloud by taking the backup data and recreating their infrastructure, previously operated on-premises, in the cloud. This allows them to become operational again relatively quickly, without the need for hardware procurement.
If this is taken a step further and the DR infrastructure in the cloud is prepared and incorporated into a well-defined concept, it can lead to a significant reduction in RTO and cost efficiency.