It is the wrong water.

Particulate and biology block fine channels.

Pressure swings the loop cannot tolerate.

Cleanliness standards differ by an order of magnitude.

The heat exchanger keeps the two apart.

That separation is what a CDU exists for.

Only at matched water temperatures.

Every kilowatt rating assumes a supply condition.

Vendors publish at different ones.

Headline comparison across them is meaningless.

Ask for output at the site's actual supply.

Warmer facility water reduces available capacity.

Everything downstream has small passages.

Cold plates and manifolds clog rather than wear.

A blocked channel overheats one processor.

Filters need to be serviceable while running.

Ask about element cost and interval.

Commissioning debris is the first test.

Pumps, first and foremost.

A stopped secondary loop stops the hardware.

Dual pumps with automatic changeover are standard.

Replacement while running is the next step up.

Controls and power feeds follow.

Match it to the hall's availability target.

It follows pipework distance.

In-row units sit close to what they serve.

Floor-standing units feed many racks from plant.

The second keeps water out of the white space.

The first reduces distribution complexity.

Site policy on water in the hall often decides.

Supply temperature and flow, held stably.

Condensation avoidance where dew point is a risk.

Alarms that reach the site's management system.

Integration protocol matters more than the display.

Ask what it reports, not what it shows.

Trend data is what diagnoses a slow fault.

Cleanliness of the secondary loop.

Flush and fill procedure agreed in advance.

Pump changeover proven, not assumed.

Alarm paths tested end to end.

Leak detection interlocks exercised.

Record the baseline for later comparison.