10-year server lifespan? That’s what one cloud service provider plans

A trend to extend the lifespan of servers beyond the typical three- to five-year range has companies such as Microsoft looking to add a few years of use to hardware that would otherwise be retired.

The latest is Scaleway, a European cloud services provider that’s sharing details about how it plans to get a decade of use out of its servers through a mix of reuse and repair.

Scaleway, based in Paris, decided the carbon footprint of new servers is just too large – server manufacturing alone accounts for 15% to 30% of each machine’s carbon impact. Reusing existing machines, rather than buying new ones, could significantly reduce e-waste.

So Scaleway decided to retrofit its 14,000 servers rather than dispose of them. Marc Raynaud, hardware support manager at Scaleway, documented the project in a blog post.

Scaleway found its old servers had a high RAID failure rate but were otherwise performing well. Batteries in the RAID controllers were the main source of failures. Raynaud explained that even if Scaleway replaced the batteries, the company didn’t expect it would lead to long-term, reliable performance, since batteries are chemical devices that degrade over time.

Scaleway decided to remove the RAID controllers, which led to a large-scale retrofitting project.

“This realization led us to explore more modern server options that do not rely on hardware cards like the ones originally equipped in our older servers,” Raynaud said via email. “As one of our key objectives was to eliminate the need for these hardware RAID cards altogether, we have shifted our focus towards procuring servers that align with this goal.”

So the company took the step of removing the RAID altogether and instead is now using onboard SATA controllers to directly attach disks to its servers. “With this setup, any necessary RAID functionality can be achieved through software RAID features,” Raynaud said.

The company’s goal is to achieve a high level of reliability and performance on these servers through a three-step process of qualification, testing, and validation.

Scaleway started by setting performance objectives for the finished product and taking a more detailed inventory of underperforming servers, looking at variables such as where they were located, the CPU, the use catalog each server was being sold in, and what catalog they could be sold in after the retrofit was done.

With that done, Scaleway needed to test its retrofitted servers to see if they could actually meet the proposed use cases. It put together a checklist for its hardware engineering team to determine the constraints and requirements for each lot of servers.

Physical checks were designed to see how the servers performed in a production environment. One of the tests is to make sure the RAID card could be physically removed, or that the team could bypass the RAID and allow access to the disks. Read/write tests were done in all SATA modes to ensure the performance was as good as, if not better, than before the retrofit.

Scaleway then performs a RAM upgrade and validates the performance, followed by a CPU performance check and a check of the firmware version for the different components (BIOS, BMC, etc.) to see if updates are required.

The company’s previous bare-metal offerings were equipped with memory configurations tailored to meet the needs of their clients at the time. However, as their needs evolved over time, there was an increased demand for more memory, Raynaud said. So Scaleway added more memory to the servers.

“It is worth noting that the reliability of memory DIMMs [dual in-line memory module] has significantly improved, and therefore, the failure rate was not a decisive factor in our decision-making process,” Raynaud said. Scaleway also reused a lot of their existing stock of DDR3 and DDR4 DIMMs as part of this upgrade initiative.

Once the hardware checks were completed, Scaleway began to physically move all 14,000 servers to a new data center. For this process, Scaleway hired a team to unrack and move them at a pace of hundreds of servers per week. It was only after the servers arrived at the new data center that the upgrades – removing the RAID card, new memory, firmware updates, etc. – were performed.

The servers have been relocated to multiple locations and data centers so as to optimize the network infrastructure and improve the overall performance and reliability of our systems, Raynaud said.

Not all servers qualified for the retrofit process, but those that didn’t qualify were not discarded. Instead, Scaleway used their parts for reuse and repair. “We harvested as many parts as possible to build up a stock of spares. After all, these servers are relatively old and will need maintenance in the future,” Raynaud wrote.

Scaleway’s servers have an age range of 7 to 10 years old. Raynaud notes that they have been properly maintained and upgraded over time to ensure that they remain reliable and fully operational.

“In the long run, these are investments that are good both for business and the environment. Resources needed to build new servers are not unlimited, and we’re proud to have developed a system to retrofit our old ones with minimal waste,” he added. “Today, our servers are used for up to 10 years — versus the industry average of three to four years — and nearly 80% of components are recycled.”