Do these 3 things before closing this tab:
1Fix the driver behind crashes, sound loss and screen glitches2Repair Windows errors before they cause bigger problems3Scan for outdated or missing drivers - takes under a minuteSome links on this page are affiliate links: if you buy through them we may earn a commission, at no extra cost to you.
In 2016, LinkedIn began reshaping its infrastructure to support growth from tens of thousands of servers toward hundreds of thousands. Its approach borrowed hyperscaler techniques—custom facility design, dense racks, a scale-out 100 GbE network and more direct hardware sourcing—while adapting them to a company that leased data-center space rather than owning an entire campus. The result was not a copy of Google or Facebook, but a distinctive effort to make large-scale infrastructure work within the constraints of colocation.
What “hyperscale” meant for LinkedIn
Here, “hyperscale” describes how infrastructure is designed and operated, not simply how many servers a company owns. The model uses repeatable building blocks and plans the facility, power, cooling, networking, servers and software as parts of a system that can expand predictably. It can also involve buying hardware more directly from original design manufacturers (ODMs) and specifying equipment around a company’s workloads rather than relying entirely on conventional branded products.
LinkedIn’s 2016 initiative was a hyperscale-style transformation for its own services—not evidence that it had become a general-purpose cloud provider or matched the physical scale and ownership model of the largest campus-owning hyperscalers. The company’s first major implementation was LOR1, a leased facility in Hillsboro, Oregon. Data Center Knowledge’s 2016 report described the facility and the infrastructure choices behind it; LinkedIn’s engineers outlined the network effort as Project Altair.
Why LinkedIn needed a different infrastructure model
LinkedIn said its existing infrastructure needed to grow from a fleet in the tens of thousands of servers to hundreds of thousands over the following years. Expanding a dispersed set of smaller deployments incrementally could make capacity planning, network design and operations harder as demand rose. LinkedIn wanted a more repeatable platform for adding capacity at that scale.
#1 Best Overall
- Save valuable floor space: 6U wall mount server cabinet Dimensions: 13.78" H x21.65" W x17.72" D.Maximum mounting depth is 14.2"
- Keep critical network equipment secure: glass door and side panels are lockable to prevent unauthorized access. Front door can be installed on either side of the front of the cabinet to satisfy your door swing orientation preference
- Easy equipment configuration: Fully adjustable mounting rails and numbered U positions, with square holes for easy equipment mounting with top and bottom punch-out panels for easy cable access
- Durability: Made of high quality cold rolled steel holds up to 110lb (50kg) (Easy Assembly Required)
- PCI & HIPPA and EIA/ECA-310-E compliant
The network was central to the problem. More servers mean more communication within and between groups of machines, so simply adding compute capacity is not enough: the fabric connecting it must scale as well. Project Altair paired a redesigned data-center network with a larger facility architecture, rather than treating faster switches as an isolated upgrade.
LOR1: high density in leased space
LOR1 was in Hillsboro, Oregon, and leased from Infomart Data Centers; it was not a LinkedIn-owned campus. Related coverage described an 8 MW custom data hall. That tenant status helps explain the design: unlike an operator building an entire campus around its own equipment, LinkedIn had to work within a leased facility’s available space and power.
According to Yuval Bachar’s statements reported by Data Center Knowledge, the facility used custom electrical and mechanical designs and was planned to accommodate more than 100,000 servers. Initial cabinets were reported at 96 servers and slightly below 18 kW each. The cooling design was described as capable of supporting about 32 kW per rack. These are reported design figures, not independently audited measurements of continuous operating load.
Free tools Windows power users keep installed
One-click scans. No signup required.
LinkedIn used heat-conducting cabinet doors to contain hot exhaust within the rack. The room side was consequently described as effectively cold-air space, rather than a conventional arrangement built around hot-aisle containment. This was a rack-level air-management approach; it should not be confused with liquid cooling.
Higher density could put more computing capacity into constrained leased space, but it is not automatically more efficient overall. It makes power distribution, cooling, maintenance access and failure planning more demanding. A facility’s ability to support a stated rack-density ceiling does not establish that every rack operated at that load, or that higher density reduced total energy consumption.
Project Altair and LinkedIn’s 100G network
Project Altair introduced a flatter scale-out fabric intended to connect pods containing thousands of servers. LinkedIn described the architecture in terms of predictable end-to-end latency and oversubscription characteristics. In other words, it aimed to make the network’s behavior more consistent as the deployment grew, rather than letting each expansion add an ad hoc layer of complexity.
LinkedIn specified a 100 GbE switch design for use across its facilities and planned a progression through 10G, 25G, 50G and 100G connectivity. The optical approach used PSM4 and split a 100G connection into two 50G ports. Bachar presented the design as a cost-effective way to reach high bandwidth while retaining intermediate steps; that was LinkedIn’s engineering assessment, not a universal cost comparison for every data center.
Rank #2
- Save valuable floor space: 12U wall mount server cabinet Dimensions: 24.25" H x21.65" W x17.72" D. MAXIMUM MOUNTING DEPTH is 14.2".
- Keep critical network equipment secure: glass door and side panels are lockable to prevent unauthorized access; Front door can be installed on either side of the front of the cabinet to satisfy your door swing orientation preference
- Easy equipment configuration: Fully adjustable mounting rails and numbered U positions, with square holes for easy equipment mounting with top and bottom punchout panels for easy cable access
- Durability: Made of high quality cold rolled steel holds up to 110lb (50kg) (Easy Assembly Required)
- PCI & HIPPA and EIA/ECA-310-E compliant
“Designed its own switches” needs qualification. The evidence describes LinkedIn specifying the switches and using white-box or ODM-style hardware, not manufacturing all the equipment itself. The strategic change was greater control over the network design and procurement, with the corresponding responsibility for integration, support and lifecycle management.
Servers: modified ODM hardware, not a fully custom fleet
LinkedIn was buying servers from ODMs and modifying standard configurations. In 2016, more fully custom servers were under consideration for a later generation; the available reporting does not establish that LinkedIn had already deployed an entirely proprietary server platform. The distinction matters: custom facility and network design were part of the documented implementation, while deeper server customization remained a possible next step.
Direct sourcing and tailored specifications can provide control over configuration and purchasing, but they shift more work to the buyer. Validation, firmware and management integration, spare-parts planning, supplier diversity and support arrangements all become important parts of the economics—not incidental details.
Why LinkedIn pursued high density
The reported rationale was a detailed comparison of server, power and space costs. In a leased data hall, concentrating more compute into each rack can make better use of limited floor area and available power capacity. That is particularly valuable when moving from retail colocation—often organized around individual cabinets—to larger, more standardized deployments.
Windows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstallCrashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minuteDensity is a trade-off, not a shortcut. It can improve capacity per square foot while increasing the consequences of a cooling or power-distribution problem. It may also complicate service access and constrain equipment choices. Whether it lowers total cost depends on utilization, electricity and lease terms, facility design, hardware, redundancy and the engineering capability available to operate it.
Open19: modular hardware for conventional racks
In July 2016, LinkedIn announced Open19, an effort to bring modular and open principles to server and rack design. Its initial concept targeted standard 19-inch, four-post racks and used modular server “bricks,” with a power shelf and optional battery-backup and top-of-rack networking components. Snap-on power and data connections were intended to simplify deployment; the initial design described up to 100G per brick and bandwidth and power that scaled with brick size. LinkedIn set out the concept in its Open19 announcement.
The goals included lower rack and server costs, improved power utilization, simpler sourcing and interoperability among suppliers. Those goals connect Open19 to the broader infrastructure strategy: modularity could make it easier to add capacity without committing every part of the system to one supplier’s proprietary design.
Rank #3
- Sturdy:4u server rack is construct from cold rolled steel, with a weight capacity of 110lbs(50kg); Electrostatic powder coat prevents rust and corrosion,quality finish
- Direct use:Open and use, not having to assemble it.Network rack can be placed flat or mounted on the wall,also can be installed vertically under the table
- Design Features:maximum mounting depth of 14 in,cables can be fixed on the side panel;Open frame server rack achieves effortless inspection, replacement and assemble
- Installation:wall mount network rack is easy to install,with instructions or videos for reference;Equipped with multiple accessories, suitable for different needs
- Application:EIA/ECA-310-E Compliant;wall mounted 4u rack fits all 19" racks and cabinets to hold various IT, network, and AV equipment;wall mount rack available in 4U, 6U, and 8U to choose
Open19 began as an initiative, not a mature, universally adopted standard. LinkedIn announced the Open19 Foundation with founding members in 2017, then said in 2018 it would contribute the platform’s mechanical, electrical and networking designs to the community. See the company’s accounts of the foundation launch and design contribution.
Why Open Compute hardware was not an immediate fit
LinkedIn’s reported reason for not using Open Compute Project (OCP) hardware in the described facilities was compatibility with standard data-center racks and infrastructure. Its leased environment shaped what could be installed: rack dimensions, power distribution, cooling, clearances, cabling and switch topology all matter. An open design is not automatically interchangeable with another facility’s equipment.
LinkedIn’s choice was therefore practical rather than a rejection of openness as a principle. Open19 pursued modularity within a conventional rack form factor. The case illustrates why buyers must assess the facility and deployment model alongside a hardware standard—and why the initial incompatibility does not establish a permanent corporate policy against OCP.
The colocation business impact
The infrastructure shift had consequences beyond LinkedIn’s own data halls. Equinix reported that LinkedIn moved equipment out of about 1,300 cabinets in the Americas in the fourth quarter of 2016, while retaining interconnection services. Equinix estimated the departure would reduce first-quarter 2017 revenue by $6.8 million, according to contemporary reporting.
This was not a complete exit from Equinix. It shows how a customer can reduce retail cabinet usage while keeping valuable network connections, and how a move toward larger leased deployments can shift revenue between retail colocation and wholesale capacity. Large customers’ infrastructure decisions can materially affect providers.
Quick wins for a faster PC:
Repair Windows errors before they cause bigger problemsFix Now →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →What changed after Microsoft acquired LinkedIn?
Microsoft announced a $26.2 billion acquisition of LinkedIn in 2016 and completed it that December. At the time, Microsoft’s CFO Amy Hood was reported as indicating that the company did not intend immediate disruptive changes to LinkedIn’s capital expenditure or infrastructure. Contemporary coverage described that initial hands-off posture.
Microsoft’s infrastructure practices differed in important respects, including support for OCP-based hardware, but the acquisition alone does not reveal what happened to LinkedIn’s architecture over the long term. Public evidence cited here does not establish whether LinkedIn fully retained, abandoned or merged the 2016 approach. The accurate conclusion is limited: the initial strategy was real, and there was no immediate public confirmation of a wholesale replacement.
What infrastructure planners can learn
- Scale changes the economics. At large server counts, repeatable designs and standardized capacity additions can be more manageable than continually extending a patchwork of deployments.
- Lease terms shape architecture. A tenant may prioritize density and conventional rack compatibility in ways that differ from a campus owner able to design buildings around its hardware.
- Network design belongs in capacity planning. Compute growth without a scalable fabric can move the bottleneck rather than remove it.
- Customization creates obligations. White-box equipment and open designs can increase choice, but require engineering for validation, integration, spares and support.
- Open does not mean plug-and-play. Confirm mechanical, electrical, cooling and network fit before adopting a platform.
- Measure the whole system. Rack density can save space without guaranteeing lower energy use or lower total cost. Model power, cooling, redundancy, utilization and operations together.
LinkedIn’s significance was not that it became identical to the largest hyperscalers. It adapted hyperscale methods—especially a custom network, denser deployment and modular hardware ambitions—to a leased, multi-site environment. That makes the effort a useful example of how infrastructure strategy depends as much on facility constraints and procurement as on server count.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
The Tool Desk
Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →
