Considering a move from colocation to cloud? Here’s how to build a solid business case
Tue 21 Apr 2020 | Eltjo Hofstee
How to establish a solid rationale for migrating infrastructure from colocation to the cloud
The current crisis instigated by COVID-19 is putting immense pressure on organisations across the globe to maintain smooth operations while managing remote workforces and workloads. Therefore, many businesses that operate their own IT infrastructure – whether through racks located on the company’s premises, or via a data centre where they rent one or more colocation racks – are considering making a move to the cloud. In either situation, adopting a cloud strategy means that infrastructure needs to be moved from the office location to a data centre, or from one data centre to another that offers hosted services.
A recent conversation with a UK prospect operating in the retail sector led us to find out that the business had recently ordered new servers to host its IT environment, at the company’s headquarters. The company’s manager also revealed that the business had considered moving to the cloud, however decided against doing so as initial costings seemed too expensive. Without digging deeper, it seems that the retailer perhaps didn’t undertake a full business case of all the costs actually involved. So, what are the key considerations for establishing a solid rationale for moving an organisation’s infrastructure to the cloud?
Decisions about which strategy to adopt should be made on the basis of a strong business case. However, the mistake that organisations often make is that they focus too heavily on factors such as hardware CAPEX, the cost of rack space, and the cost of internet traffic, when they should be considering a wider range of financial issues. These include:
- Cost of purchasing equipment
- Allowing for interest payments on money spent at the beginning of a three-year investment cycle (unless cash purchases are involved)
- The cost of support from hardware vendors
- Contingency cost for space and/or replacement equipment. This includes at least one for all network equipment, as well as HDD, RAM and CPUs, or an expensive support contract with key vendors for rapid replacement
- Cost of licences (some software vendors offer different licence schedules on owned hardware compared to cloud offerings)
- The cost of IT staff who monitor and maintain the infrastructure
- If the organisation maintains the platform with in-house staff, allowing for 24/7 support costs for those IT staff, who can resolve emergency situations. Typically, this will require a minimum of three to five employees which, if they have to be recruited, is a risk that should be monetised in the business case
- Cost of training IT staff on key hosting technologies, such as VMware, Windows, or Cisco
To make these costs comparable to a monthly ‘pay as you go’ fee, they should be calculated back into a monthly cost. It’s sensible to make these calculations using a three-year write off period, as looking at anything longer means being brutally honest about whether hosting current infrastructure on hardware that is older than three years will really offer the reliability, security, and performance required. If the answer is yes, then the write-off period can be extended to four or maybe even five years.
These costs can then be compared with a monthly service provider fee that should also include some additional one-off items, including:
- Customer specific costs, such as modifying an ERP system to work with a private cloud service, or costs for new monitoring tools if the old monitoring system could only work within the existing network.
- Migration costs from the current solution to the private cloud infrastructure.
In particular, migration costs can be challenging to precisely calculate in advance, but is important to keep budget implications in mind as plans to move from colocation to cloud move forward. For instance, if relocation requires certain services to move to another data centre, there are two main options to consider.
The first is to virtualise existing servers and subsequently move them one by one to the private cloud (in most cases an IP change is still needed, something which could create issues). The complicating factor here is that the write off period for current equipment may not end at the same time. In this case, support contracts on older hardware have to be extended or some of the equipment has to be written off faster. The unpleasant alternative is to end up managing two separate clouds, and few organisations have the desire to cover the costs of duplication.
The second option, which can help get around this complication, is to move all current infrastructure from the office premises or existing data centre to the new data centre location. Choosing this option should be viewed as a ‘big bang’ exercise, where everything can be accomplished in one or two heavy nights of relocating servers. In some data centres, it is possible to connect current colocation rack(s) to the new private cloud, allowing a less intense move from colocation to cloud and, with a bit of luck, also retaining keep the current IP.
The bottom line is that any successful infrastructure change takes thought and preparation, because every organisation has unique requirements and varying levels of experience. Working with a partner who does it for a living can help to mitigate most problems and ensure a smooth transition from one infrastructure strategy to another with little – and ideally, no service interruptions.
Data Centre Tue 21 Apr 2020How much of a data centre can be controlled remotely?