Planning IT infrastructure
By Redaktion techport.ai, IT-Beratung · Last updated on
Infrastructure only gets noticed when it is missing. It is the network production reports over, the server the ERP runs on, the link to the second site and the power supply in the technical room. As long as everything runs there is no reason to talk about it, and that is precisely why it usually gets renewed only when it fails or when a project hits its limits.
Infrastructure planning in the mid-market does not mean buying the newest equipment. It means knowing what has to be replaced when, where the limits are, and what happens when a part fails.
How you notice it
- Devices run considerably longer than intended because no replacement was budgeted.
- Systems get slow under peak load and nobody can say where the constraint sits.
- The technical room has grown over the years, cabling and labelling are out of date.
- There is no overview of which application runs on which machine and what depends on it.
Why this happens
Infrastructure gets bought project by project. One initiative needs a server, another more storage, a third a faster link. Each purchase solves its problem, but nobody assembles the parts into an overall picture and nobody plans the replacement, because the purchase sat in a project budget and the replacement sits in none. After five to seven years several replacements then fall due at once, usually unplanned.
How we go about it
- Record inventory and life cycle. We capture which components exist, how old they are, how long they will be supported and which applications run on them. From that comes a multi-year replacement plan that feeds into budget planning.
- Make dependencies visible. We document which application depends on which component, including network, directory service, name resolution and power supply. That chain is the most important information during an incident and the least known in daily work.
- Check capacity in the right place. We measure where bottlenecks arise rather than everywhere: storage, throughput, response times of the central systems and the links between sites. On that basis growth can be planned rather than reacted to.
- Secure operations. We set maintenance windows, arrange spare parts or contracts with response times for critical components, and check how long an outage would last per component. That figure belongs in the emergency plan.
What you gain
- Replacements that appear in the budget as planned rather than as a surprise.
- Incidents narrowed down faster because dependencies are documented.
- A solid basis for deciding what stays in house and what is operated externally.
From our projects
The most frequently underestimated dependency is the directory service. When it is unavailable, practically nothing works any more, including systems that appear to have nothing to do with it. In many companies it runs on a machine whose failure scenario has never been thought through. The second most common surprise is the power supply: uninterruptible supplies exist, but their batteries are older than intended and have not been tested for years, which in an emergency means the protection does not take effect. Both are small checks with large effect.
Häufige Fragen
Should we keep our own servers or outsource everything?
That is decided per system, not across the board. For standard applications external operation is usually more economical. For systems with tight response time requirements, for production control or where the external connection is itself a risk, running your own remains sensible. What matters is that the remaining infrastructure is then small enough to operate properly.
How long should servers and network components run?
As orientation, five years for servers and up to seven for network components, but vendor support is what decides. Once there are no more security updates, age is irrelevant and the component has to be replaced. That is why the end of support belongs in the inventory list, not the purchase date alone.
Let us talk about Planning IT infrastructure
In a thirty minute first call we work out where your biggest lever sits and whether we are the right people for it.
Further reading
Back to the field Operations and Support