Ensuring uptime is a critical objective for data centre managers. Technology is helping to ease the load by providing intelligent insights into power usage. Calvin Nicholson, Senior Director of Product Management at Legrand DPC (Data, Power and Control Division), which includes the Server Technology brand, talks us through the key benefits of the company’s PRO3X PDU and how it is helping infrastructure teams to future-proof their devices.
How important is continued uptime for data centres and what are some of the factors that influence uptime?
Uptime is critical and there are a number of things that influence it.
Most of our customers use two PDUs, an A and a B feed that can be from two separate grids or one grid and a UPS, for example.
You ideally want to load each side to 40% so that if you get a failure and one PDU is picking up the whole cabinet’s load, you’re running roughly 80% of the load.
Organisations aren’t permitted to run continuously at above 80, so it’s really as simple as knowing the overall kW, distributing that load evenly, ensuring the cabinet is powered and plugged in properly during installation, and then making sure that power is distributed over both of the PDUs within the cabinet.
How far does real time monitoring reduce the risk of downtime and what tools can data centre managers use to assist with this?
Real time monitoring can refer to power, but it can also be environmental – we know there’s a focus today on saving money, and saving money is saving power. That can be within the cabinet and it also can be within your cooling infrastructure.
For example, we have a number of customers that are running at higher temperatures. The advantage of that is you don’t have to provide as much cooling, so you save power, but if you have a problem with your cooling system it’s critical that you know right away.
With switched intelligent PDUs, like the ones we provide, we have sensors for temperature, humidity, differential air pressure, water and contact closure. And various other things.
It allows confidence to know when there is a problem and that you can react before you have downtime or before you have a critical issue, ensuring that it doesn’t daisy chain within the data centre.
We have everything from devices that are measured at the in-feed for capacity planning to devices that measure at the outlet which is at the device and server level. We can really provide the information detailing what’s going on within the cabinet to ensure uptime and address problems quickly.
Can you introduce us to the PRO3X PDU and the challenges it aims to address?
The PRO3X PDU is really a hybrid design, taking some of the best Server Technology features and some of the best Raritan features.
One of the things that we’ve done is integrate mechanical locking for every outlet, which means no more sourcing locking cables in different lengths and different colours – it’s inherently built right into the PDU itself.
That really helps as we know that the more people that are in the data centre, frankly, the more problems you’re going to have. With locking outlets, cables can’t be bumped or pulled by accident, as an example.
We’ve incorporated our HDOT and our HDOT Cx technology, which is a custom outlet that we’ve patented and allows for a lot of flexibility as the hybrid Cx outlet can act as a C13, or a C19 outlet. This helps to future proof your cabinet configuration and make it much easier when there are equipment upgrades or cabinet refresh.
Could you share some use cases of how this technology is being used?
Looking at a cloud application, for example, behind that environment, there is a PDU. Power and environmental monitoring is going on behind the scenes in a cloud environment.
When you get into a data centre application, whether it’s a typical data centre that you own or a colocation, the same technologies are being used, but in a different way. In a colocation facility, we find a lot of focus on monitoring, making sure the colo meets its SLAs.
In a lot of regular data centres, power monitoring is key, right down to the device level.
It really comes down to what the customer wants to focus on – what is important to them, what their SLAs are and what they’re trying to accomplish. We really do everything from very basic PDUs, all the way up to the most intelligent monitoring.
How does this technology ultimately enable the likes of Smart Cities and emerging technologies such as Edge Computing?
If you look at Edge Computing and Artificial Intelligence, you’re really bringing the compute closer to where it’s needed.
An example here would be a self-driving truck with a tonne of data. You need to track all that data and keep it somewhere – whether it’s for storage, monitoring or to understand how the truck can be more fuel efficient. That’s all done at an Edge data centre.
That bandwidth and amount of data can’t be distributed huge distances. For Edge and AI it can be a regular data centre or a smaller data centre, but it’s really closer to where it’s needed.
You can also argue that it ties somewhat into Internet of Things, where you’ve got a large number of devices which the compute needs to be closer to. I think it’s really an enabling technology and we’re really just starting to see the uses.
How does flexibility extend the life of a data centre and how does the PRO3X enable this?
Being part of the critical physical infrastructure, the PDU is going to stay and last roughly the life of the cabinet.
We know that many customers refresh IT roughly every three to five years, depending on what they’re doing.
Let’s say we’ve got a product where half of the outlets are regular HDOT C13s and half of the outlets are HDOT Cx, that could be a C13, or a C19. So that one PDU can act as all 36 C13s. Or it can act as at least half of the outlets being a C19.
When you’re an industrial installation, you may need just a few C19s, because you don’t have a lot of high-power density, but if you do a refresh you might need a little more kW. You’ve got the C19s that are available to support that.
If you put technologies that are flexible like that into your data centre, when you do refreshes it enables you to grow and makes it easier when you’ve got these upgrades.
What is your best practice advice for data centre managers to ensure they’re creating a power strategy that will future-proof their infrastructure?
First off, look at what the current and future density is for the cabinet, kW wise. Many of our customers are going with 400-volt solutions in North America, which is what the rest of the world is already doing.
When you bring 480 power into the data centre you don’t convert it all the way down to 120, essentially, for a 208 application. So, there’s efficiency gains of bringing the power in at 480, and just converting it down to 400 or 415.
If you look at the devices themselves, believe it or not, they run more efficiently at higher voltages. So instead of running a server at 120 or 208, if you run it at 230 or 240 volts, then you have an efficiency there.
When you stack up all those efficiencies it really adds up – especially when you’re using 1000s of kW like data centres do.
Then, as I mentioned on the PDU side, I would look at technologies like integral locking. We mentioned the HDOT and HDOT Cx for the flexibility of outlets and the devices that you can plug in.
And then also, think about what you’re trying to do. Do you want to remotely turn things on and off? If you’re thinking about understanding the power usage, look at a product that can measure power, whether it’s at the end feed which is pretty much typical for an overall capacity, but also at the outlet level.
Also, a server sitting idly in your data centre not doing any useful work still uses roughly about half the power that it would be if it was doing something. You don’t want a bunch of servers just sitting around in your data centre.
There are decisions you can make that will help drive efficiency, both on the power chain but also in the application of the devices themselves.
