Sergey Nivens - Fotolia
Get the right SLA for enterprise liquid cooling services
If you're interested in data center liquid cooling, investigate the technology types, available service providers and outline an effective agreement for successful system use.
Once reserved for mainframes and supercomputers, liquid cooling is now available to enterprise data centers. New demanding workloads and hardware options increase power densities and place traditional cooling systems under strain. Data center managers are starting to look at enterprise liquid cooling services and systems as alternatives to air-based cooling setups.
The high cost of cooling systems is one of the primary reasons data center managers are exploring different cooling options. Hardware and processing power are becoming denser and higher powered as GPUs replace traditional processors, which puts more demands on the air-cooling systems. Costs for cooling can increase and may not be cost-effective.
Liquid cooling systems, on the other hand, are more efficient, take up less space, cost less and are quieter. They may be more suited to cool the processing-heavy hardware that runs in newer data centers.
The benefits of liquid cooling
The various types of liquid cooling make cooling hardware more targeted and scalable. It's more efficient because liquid cooling systems can be self-contained in sealed units and provide cooling directly to the device. These sealed units also are protected from dust and other harsh particulates.
Managers can see lower costs overall because the units concentrate their cooling efforts on the hardware itself, instead of at the building level. The systems use less power and water because they are smaller than traditional air-cooling units. Direct-to-chip liquid cooling systems can be installed on specific server units with high-use processing systems and applications, providing targeted cooling where needed.
Why use an enterprise liquid cooling service provider
Because liquid cooling is still relatively new for organizations, only a few providers offer enterprise liquid cooling services and systems. They're solely focused on liquid cooling for enterprises, and many of them have established benchmarks and proprietary technology for the industry.
As enterprises evolve their IT infrastructure and push the boundaries of what technology can do, they'll need to look at cooling alternatives, too. Their IT systems will not work properly without it.
Green Revolution Cooling (GRC) offers single-phase liquid immersion setups for data centers. They've also partnered with Hewlett Packard Enterprise to create a product that combines HPE servers with GRC immersion cooling technology.
Asetek offers a variety of direct-to-chip cooling products and has partnered with several leading high-performance computing and artificial intelligence providers, such as Fujitsu, Intel and Penguin Computing.
CoolIT Systems offers rack-based liquid cooling that admins can deploy with any server in any rack in a data center and partners with Dell EMC, HPE and Super Micro Computer.
Asperitas and Submer offer several immersion cooling options. These offerings are easy to install and maintain and enable managers to enjoy the full benefits of liquid cooling right away.
Schneider Electric has partnered with Iceotope and Avnet to offer chassis-level data center liquid cooling options.
What to include in a liquid cooling SLA
Liquid cooling system service and maintenance is a significant change for data center managers. Instead of calling a single service technician for support, they may have to contact several different vendors for enterprise liquid cooling services. The support workflow depends on the particular liquid cooling option they implement and the vendor they purchase from.
Some vendors have outsourced their warranty, service and support program to a third-party vendor, while others manage the systems themselves.
The first section of the service-level agreement (SLA) should be the summary. It summarizes the services being delivered, who's providing them and how the IT team can measure the success of the SLA.
Next, outline the goals for each party (the IT team and the vendor) that help admins map the support offerings with expected performance levels. For example, the vendor agrees to help maintain hardware and application performance levels of at least 80% and that they intend to make a certain number of technicians available 24/7.
The third section should list what each party needs to reach the previously defined goals. The vendor could offer a weekly status update or report, a quarterly meeting and a biannual physical inspection. As the customer, IT teams may need to provide usage information, status updates and any maintenance concerns to the vendor so they can regularly monitor and assign the appropriate resources to meet the SLA's terms.
Most importantly, the SLA must list all contact information that's needed for the vendor to provide service to the IT department. Be sure to include information for everyone that must be involved, such as maintenance staff, management stakeholders and project sponsors. If the liquid-cooled hardware is in a colocated data center, make sure to include colocation provider contact information and any unique access information requirements.
SLAs should outline what happens when the goals aren't met. This includes any financial penalties that are incurred, and by whom, and the form of compensation. Will the vendor offer service credits whenever they miss goals or is it a refund check that's issued once a quarter?
Finally, the SLA should include a cancellation section. This is especially critical if the cooling vendor outsources service and maintenance; there may be an opportunity to switch third-party providers with no penalty and use the same liquid cooling offering.