Who supplies and installs servers and storage in Adelaide?
Subnet does. We are based in Forestville, Adelaide, and have worked with HPE since 2008. We size servers and storage from measured usage, build and test the platform, migrate workloads in stages, and look after backup, power protection and warranty afterwards.
How do we decide whether a workload belongs on premises, in a hosted facility or in the cloud?
Start with the workload rather than the platform. Useful questions include how quickly users and systems need to reach it, where the data must be held, how steady or variable the demand is, what the application vendor supports, and which skills are available to run it. Different workloads in one organisation often land in different places. A scope should record the reasons for each placement, so the decision can be revisited when requirements change.
What is the difference between on-premises, colocation and hosted infrastructure?
On premises means the equipment sits in a facility your organisation controls, such as a server room or its own data centre. Colocation means you own or lease the equipment but house it in a third-party facility that supplies space, power, cooling and physical security. Hosted infrastructure means a provider owns and runs the platform and you consume it, which moves hardware duties to the provider. The practical difference is who is responsible for the building, the hardware, the platform and the data, so confirm that split for each option.
Can on-premises and cloud services work together?
Yes. Many organisations keep some systems on site and use cloud services for others. The design then depends on the connection between the two, how people sign in, where data is copied or moved, and how each side is monitored and supported. Data transfer and network costs can add to running costs, so include them when comparing options. Agree which team or provider looks after each part, including the links between them.
How long should servers and storage stay in service?
There is no single right age. Planning is usually driven by the manufacturer’s end-of-support and end-of-sale dates, the warranty or support term, how much capacity and performance are left, and how often the equipment needs repair. Build a register of equipment with these dates, so refreshes can be budgeted ahead of time rather than triggered by a failure. A refresh is also the point to ask whether the workload should stay in the same place.
What should we consider when sizing capacity for a hardware refresh?
Work from measured use rather than the specification of the old equipment. Look at current and peak demand for processing, memory, storage and network, expected growth over the planned service life, and how much headroom is needed for failures and maintenance. Software licensing, rack space, power and cooling can also change with new hardware. Ask for the assumptions behind any sizing so they can be checked against your own figures.
How much power and cooling should we plan for?
Add up the power draw of the equipment you expect to run, then allow for growth, the redundancy you need and the efficiency of the power and cooling equipment. Cooling must remove the heat that the equipment produces. ASHRAE’s guidance, as summarised by a data-centre industry source, recommends an inlet temperature of 18 to 27 degrees Celsius, with wider allowable ranges set by equipment class and manufacturer conditions. Check the manufacturer’s specifications for the equipment you plan to buy.
What do data centre tier levels mean when comparing facilities?
Tier levels come from the Uptime Institute’s Tier Classification, which describes four design levels with increasing redundancy. Tier III is described as concurrently maintainable, meaning components can be serviced without shutting down IT equipment, while Tier IV adds fault tolerance. A facility claiming a tier is not the same as one certified against it, so ask whether the facility holds Uptime Institute certification and for which phase. A tier describes the facility rather than your applications, so your own design still matters.
What support and maintenance options exist for data centre equipment?
Common options include the manufacturer’s warranty and extended support, hardware maintenance from an independent provider, a managed service that looks after the environment, and an internal team handling day-to-day care. Each differs in what is covered, who attends, and how replacement parts and updates are handled. Coverage hours and response targets vary by agreement, so compare the written terms rather than assuming them. Subnet can scope support alongside a project, with the exact arrangement set out in an agreed scope.
How do data location and sovereignty requirements affect where infrastructure sits?
Some organisations must keep certain information within Australia, or within particular facilities, because of legislation, contracts, funding conditions or customer expectations. Identify which data sets are affected and what the obligation actually says, then match each to a location option. This also applies to backups, support access and any provider that handles the data. Your legal or compliance adviser should confirm the requirements, and the scope should record how they are met.