Commercial / Monitoring and DCIM
Data Centre InfrastructureManagement and Monitoring
Direct answer
DCIM, or data centre infrastructure management, is software that watches power, cooling, space and assets across a whole facility. A computer room environmental monitoring system, commonly shortened to CREMS or EMS, is the sensor layer that watches the physical room: temperature, humidity, water, smoke, doors and access. They answer different questions, and most sites that need one need both.
Indigi Power and Cooling designs, supplies, installs and supports both across Australia on platforms from Vertiv, Schneider Electric, APC, Eaton and CyberPower, and maintains the UPS, cooling and electrical plant underneath them. Call 1800 046 344 or email sales@indigipc.com.au.
Start here
Three terms that get usedas if they mean the same thing
They do not. Buying the wrong layer is the most common and most expensive mistake in this category, so it is worth two minutes to get the vocabulary straight.
| Layer | What it watches | The question it answers |
|---|---|---|
| DCIM Facility |
Power, cooling, space, assets and how they depend on each other, across the whole site | What can this room still take, and what happens to the rest if I change this? |
| CREMS or EMS Room |
Temperature, humidity, water under floor, smoke, airflow, door access and video | Are the conditions around my equipment where they should be, right now? |
| Device monitoring Equipment |
One UPS, PDU or CRAC unit, through its own network card | Is this particular machine healthy, and what is its alarm log telling me? |
A note on CREMS, because the Australian market renders it three different ways. The hardware vendors who actually build these systems call it a Computer Room Environmental Monitoring System, and that is the correct expansion. You will also see it written as Computer Room Environmental Monitoring Solutions, and as Critical Environment Risk Management Services, sometimes both on the same page. They all describe the same thing: the sensor layer in the room. If a supplier has quoted you CREMS, this is the layer they are quoting.
Most sites arrive at this from one direction or the other. A facilities team wants capacity planning and ends up buying DCIM. An IT manager wants to know when the comms room gets hot and ends up buying sensors. Both are right for their problem, and the sensible order is to decide which question is actually keeping you up at night before anyone quotes a platform.
Pick the layer your problem lives in
Three layers of visibility,and what each one is for
Buying the wrong layer is the expensive mistake in this category. Select a layer to see what it watches, what it answers, and the hardware involved.
Select a layer
Data centre infrastructure management
Power, cooling, space and assets across the whole facility, and how they depend on each other. This is the layer that answers capacity questions, so it is what you need when the problem is planning rather than reacting.
Platforms: Vertiv Trellis / Schneider EcoStruxure IT
Computer room environmental monitoring
The physical room around the equipment. Heat, humidity, water, smoke, doors and video, with an alert the moment a threshold is crossed. This is the layer you need when the risk is not knowing the room has gone wrong.
Hardware: APC NetBotz / Vertiv Geist / Eaton EMP Gen 2
Device level monitoring
One machine at a time, through its own network card. Battery state, load, input supply and the alarm log on a single UPS, PDU or CRAC unit. Cheapest to deploy, and often already installed and unused on equipment you own.
Cards: Eaton Network-M2 and M3 / CyberPower RMCARD205 and 400 / Socomec Net Vision
The useful part
What gets monitored,and what each reading actually means
A monitoring quote usually lists sensors. What matters is what each one tells you and at what point it should wake somebody. These are the thresholds we configure by default and the reasoning behind them.
| Point | Typical threshold | What it means when it trips |
|---|---|---|
| Equipment inlet temperature | 18 to 27 degrees Celsius, per ASHRAE TC 9.9 | Measured at the cold aisle intake, not the return. A rising trend across a week is a cooling capacity problem; a step change in minutes is a unit that has tripped |
| Humidity | 5.5 to 15 degrees Celsius dew point and at or below 60% RH | Too high risks condensation on cold surfaces. Too low raises electrostatic discharge risk, which is the bound most monitoring setups forget to configure |
| Water under floor | Any detection | A CRAC condensate line, a chilled water leak or ingress from above. There is no sensible delay on this one; it alarms immediately |
| Smoke and VESDA interface | Any detection | Usually a dry contact from the fire indicator panel rather than a sensor of ours. It tells the monitoring platform what the fire system already knows |
| Door and access | Open beyond a set dwell, or any out of hours event | A propped door defeats the cooling design and is the single most common cause of a slow temperature rise nobody can explain |
| UPS battery and load | Load above design, battery on discharge, any alarm | Polled from the UPS network card. Repeated short transfers mean an upstream supply problem, not a UPS problem |
| PDU current, per outlet where supported | Circuit above 80% of rating | The number that tells you whether the next rack fits. Per outlet metering also identifies which device is actually drawing what |
| CRAC status and alarms | Any unit alarm, or fan or compressor state change | A failed unit in an N+1 room is not an outage but it is a countdown, and it should be a different alert severity from a failure in an N room |
| Differential pressure | Site specific, under floor to room | Only relevant with a raised floor. Falling pressure usually means tiles have been moved or a blanking panel has come out |
Room temperature targets and the reasoning behind them are covered further on our server room air conditioning page.
The bit that derails projects
How the piecesactually connect
Every platform brochure shows one tidy dashboard with everything feeding it. Getting there depends entirely on what your existing equipment is willing to tell you, and that is worth establishing before anybody quotes.
SNMP is the easy path. Anything with a network management card, which means most UPS made in the last fifteen years, most managed PDUs and many CRAC units, can be polled directly. If the card is missing it can usually be retrofitted, and that cost belongs in the quote rather than arriving as a surprise halfway through commissioning.
Modbus, RTU or TCP, covers most power meters and a good share of mechanical plant. It generally needs a protocol converter to join an IP network, which is a small piece of hardware and a non-trivial piece of configuration.
Dry contact inputs are how you capture anything that will only offer a volt free contact. Fire indicator panels, mechanical exhaust, generator run status and older switchboards all tend to fall here. It is a crude signal, just open or closed, but it is reliable and it is often the only thing a fire panel will give you. Expect to be told by the fire contractor what you may and may not connect to, because that panel is certified and your monitoring system is not part of its certification.
BACnet matters where there is a building management system already in place. The question is usually whether monitoring should feed the BMS, read from it, or stay independent. There is no universally right answer; it depends on who is on call and what they already watch.
The honest constraint: plenty of older UPS and CRAC units cannot be polled at all without adding hardware, and some legacy equipment simply has no path. We establish what can and cannot be read during the site audit, so the design reflects the building you have rather than the one in the brochure.
Integration, in one picture
How everything in the roomreports to one place
The brochure shows a tidy dashboard. This is what sits behind it. Hover or tab through a row to follow its path, and note that three of the six need something in the middle to translate.
Hover or tab a row to trace its path
Three of these six need hardware in the middle before they can join the network. That is the part most quotes leave out, and the reason we audit the site before designing.
Budgeting
What a monitoring build costs,and the four things that move it
Nobody in this market publishes a price, us included, because the range is genuinely enormous. What we can do is tell you what the money is actually spent on, so you can sanity check any quote you receive including ours.
Number of monitored points
The single biggest driver. A comms room with four sensors and a UPS is a different project from forty racks with per outlet metering. Points are also where quotes diverge most, because one supplier counts a four sensor probe as one point and another counts it as four. When comparing quotes, normalise this first.
Whether the cabling exists
Sensors need a path back to the appliance. In a room with free tray and spare cores this is trivial. In an occupied data hall with no spare containment, running cable is often the largest line on the quote and has nothing to do with the monitoring platform at all.
Retrofitting network cards
If the UPS or PDUs you want polled have no network card, each one needs hardware added and configured. On a fleet of older units this can quietly match the cost of the monitoring platform. It is also the easiest thing for a quote to leave out.
Licensing model
DCIM platforms license per node, per rack or per site, as a perpetual licence with annual maintenance or as a subscription. Over five years these land in very different places. Ask for the five year number, not the first year number, and check what happens to your data if you stop paying.
We quote the site audit separately and up front, so the design is based on what is actually in the room rather than on assumptions. See UPS maintenance plans for how ongoing support is priced.
Vendor agnostic, because we sell and service all of it
The platformswe supply and support
Monitoring platforms are specified and quoted per site rather than sold from a catalogue, so most of these are an enquiry rather than an add to cart. Eaton monitoring hardware is listed under Eaton UPS accessories. Where the right answer is to use the monitoring you already own rather than buy more, that is what we will tell you.
The question that actually matters
What happens whenthe alarm fires at 3am
Every supplier in this market will show you a dashboard. Very few will tell you what happens at three in the morning when the dashboard turns red and nobody is looking at it.
There are two separate things being sold here and they are worth keeping apart in your head. The first is configuration: setting thresholds, severities, escalation chains and delivery channels so the right alert reaches the right person. The second is response: somebody contractually obliged to act on it. A monitoring platform on its own gives you the first. It does not give you the second, and a dashboard nobody is watching is an expensive way to produce a log of how your outage developed.
On the configuration side, we set up severity tiers rather than one undifferentiated alarm, because an alert that fires for everything gets muted within a fortnight. A CRAC fault in an N+1 room is a next business day ticket. The same fault in a room with no redundancy is a phone call. Water under the floor is always a phone call. Delivery is by email, SNMP trap, SMS or webhook into whatever your team already watches, because an alert that arrives somewhere nobody looks has not actually been delivered.
On the response side, our maintenance plans already carry priority fault response, and the Critical Care level includes a 24/7 emergency response commitment. The sensible arrangement for most sites is monitoring configured to escalate into whatever response cover you hold, so the alert has somewhere to go. If you have an internal team on call, we configure into them instead; the point is that the escalation chain ends with a person, not a screen.
The question to ask any monitoring supplier, including us, is simple: when this alarm fires at 3am, who is obliged to do something, and what is the commitment? If the answer is vague, you are buying a dashboard.
See UPS maintenance plans for response levels, and UPS maintenance and service for what a visit covers.
Why us rather than a software reseller
Monitoring is only usefulif somebody turns up
A monitoring platform sold by a software reseller stops at the dashboard. When the temperature alarm fires, somebody still has to attend the CRAC unit, and somebody has to be licensed to open the switchboard. We are the same company for all three, which means the alert, the diagnosis and the repair do not pass through three sets of hands and two blame conversations.
It also means the thresholds get set by people who maintain the plant. The difference between an alert that is useful and one that gets ignored is usually whoever configured it having seen the failure mode before.
Coverage
Monitoring designacross Australia
Brisbane and QLD
Brisbane CBD, Tingalpa, Eight Mile Plains, Gold Coast, Sunshine Coast, Ipswich, Townsville, Cairns
Our HQ. Humidity and storm season make water detection and door monitoring worth more here than most people assume.
Sydney and NSW
Sydney CBD, Parramatta, North Ryde, Macquarie Park, Alexandria, Norwest, Newcastle, Wollongong
The largest colocation footprint in the country, so multi-tenant alert routing comes up most often here.
Melbourne and VIC
Melbourne CBD, Port Melbourne, Clayton, Derrimut, Tullamarine, Dandenong, Geelong, Ballarat
Heavy data hall presence, and the region where BMS integration is requested most.
Perth and WA
Perth CBD, Osborne Park, Welshpool, Canning Vale, Kwinana, Karratha, Port Hedland, Kalgoorlie
Remote and unmanned sites, where monitoring is often the only reason anyone knows a room is in trouble.
Canberra and ACT
Canberra CBD, Civic, Fyshwick, Mitchell, Barton, Tuggeranong, Belconnen, Queanbeyan
Departmental sites with audit and reporting obligations, where the log matters as much as the alert.
SA, NT and Tasmania
Adelaide, Port Adelaide, Edinburgh, Whyalla, Darwin, Palmerston, Katherine, Alice Springs, Hobart
Mobilisation charged per trip, with commissioning visits combined across nearby sites.
Questions we get asked
DCIM and monitoringFAQs
What is DCIM?
DCIM, or data centre infrastructure management, is software that brings power, cooling, space and asset data into one live view of a facility. It tracks how those systems depend on each other, supports capacity planning, and flags when a change in one place will cause a problem somewhere else. It is the facility layer, as distinct from environmental sensors that watch a single room or a network card that watches a single UPS.
What is CREMS?
CREMS stands for computer room environmental monitoring system. It is the sensor layer that watches the physical room around your equipment: temperature, humidity, water under floor, smoke, airflow, door access and in some cases video. It raises an alert when a threshold you have set is crossed. You will see the acronym expanded differently across the Australian market, including as Computer Room Environmental Monitoring Solutions and as Critical Environment Risk Management Services, but the hardware vendors who build these systems use System, and all three describe the same sensor layer.
What is the difference between DCIM and CREMS?
DCIM looks at the infrastructure and answers what your power, cooling and space are doing and what they can still take. CREMS looks at the room and answers whether conditions around the equipment are where they should be right now. Most critical sites benefit from both. If you only have budget for one, the question to ask is whether your risk is running out of capacity, which points to DCIM, or not knowing the room has gone wrong, which points to environmental monitoring.
What temperature should a server room be?
ASHRAE TC 9.9 recommends 18 to 27 degrees Celsius measured at the equipment inlet, not at the return air or on the wall. Humidity should sit between 5.5 and 15 degrees Celsius dew point and at or below 60 per cent relative humidity. The lower humidity bound matters as much as the upper one, because air that is too dry raises electrostatic discharge risk, and it is the threshold most monitoring deployments forget to configure.
Can monitoring be added to equipment we already own?
Usually, and it is where we would start. Most UPS made in the last fifteen years, most managed PDUs and many CRAC units already carry a network card that can be polled over SNMP, so a good deal of monitoring capability is often installed and simply unused. Where a card is missing it can generally be retrofitted. Older equipment sometimes has no path at all, which we establish during the site audit rather than discovering at commissioning.
What protocols do you support?
SNMP for anything with a network management card, which covers most modern UPS, managed PDUs and many CRAC units. Modbus RTU and TCP for power meters and mechanical plant, usually through a protocol converter. Dry contact inputs for fire indicator panels, mechanical exhaust, generator run status and older switchboards that will only offer a volt free contact. BACnet where there is a building management system to integrate with.
Who responds when an alarm triggers?
That depends on what you have arranged, and it is worth being precise about. Configuring the alerts and being contracted to respond to them are two different things. We set thresholds, severity tiers, escalation chains and delivery channels as part of any deployment. Response is covered under a maintenance plan: our plans carry priority fault response, and the Critical Care level includes a 24/7 emergency response commitment. If you have an internal team on call we configure the escalation into them instead. The important thing is that the chain ends with a person rather than a screen.
What does a monitoring deployment cost?
It varies too widely to publish a figure honestly, but four things drive it: the number of monitored points, whether cabling already exists back to the appliance, whether network cards need retrofitting to equipment you want polled, and the licensing model of the platform. When comparing quotes, check how each supplier counts a point, since a four sensor probe is sometimes counted as one and sometimes as four, and ask for the five year licensing cost rather than the first year.
Related
Critical infrastructureservices
Next step
Tell us how many rooms, roughly how many racks, and what you already have a network card on. That is enough for us to tell you what can be monitored today and what needs hardware.
We quote the site audit separately and up front, so the design reflects what is actually in the room. Call 1800 046 344, email sales@indigipc.com.au, or use the contact page. Indigi Power and Cooling is Indigenous and Veteran-owned, Supply Nation and ICN Gateway registered.
Scope a monitoring build →Partner with Us
Reliable Power & Cooling Solutions from Indigi Power & Cooling Expert supply, installation, and contracting services for critical infrastructure across Australia.