9 design goals for cloud data center networks
New white paper from Arista Networks details the company's two-tier spine/leaf and single-tier spline universal cloud network designs.
A new white paper from Arista Networks details the company's two-tier spine/leaf and single-tier spline universal cloud network designs, designed to provide data centers with "unprecedented scale, performance and density without proprietary protocols, lock-ins or forklift upgrades." The company contends that all of its universal cloud network configurations revolve around 9 central design goals. Directly quoted, these goals are as follows:
1. No proprietary protocols or vendor lock-ins.
2. Fewer Tiers is better than more Tiers.
3. No protocol religion.
4. Modern infrastructure should be run active/active.
5. Designs should be agile and allow for flexibility in port speeds.
6. Scale-out designs enable infrastructure to start small and evolve over time.
7. Large Buffers can be important.
8. Consistent features and OS.
The paper also touches on specific design considerations for leaf/spine network designs. "A design with more tiers offers higher scalability compared to a design with less tiers," states the analysis. "However, it trades this off against both higher capital expense (capex) and operational expense (opex). More tiers means more devices, which is more devices to manage as well as more ports used between switches for the fan-out interconnects between switches."
Related: Arista claims industry’s fastest data center switch
Educating on the concept of "spline network designs," Arista specifies, "Spline designs collapse what have historically been the spine and leaf tiers into a single spline. Single tier spline designs will always offer the lowest capex and opex (as there are no ports used for interconnecting tiers of switches), the lowest latency, are inherently non-oversubscribed with at most two management touch points."
The company adds that "flexible airflow options (front-to-rear or rear-to-front) on a modular spline switch enable its deployment in server/compute racks in the data center, with ports on the same side as the servers with airflow that matches the thermal containment of the servers."
The analysis goes on to conclude that "capex costs go up with increased scale. However, capex costs can be dramatically reduced if a network can be built using fewer tiers, as less cost is sunk into the interconnects between tiers. Opex costs also decrease dramatically with fewer devices to manage, power and cool, etc. All network designs should be looked at from the perspective of the cost per usable port (i.e. those ports used for servers/storage) over the lifetime of the network."
View/Download the white paper.