English

Optimising Data Centre Performance with FS IP Fabric Solution

Posted on Aug 29, 2024 by
201

To meet the demands of high-performance computing and storage, data centres require robust, flexible, and scalable network solutions. FS IP Fabric solution, built on a Spine-Leaf architecture, perfectly addresses these challenges. Read on to discover how this solution ensures your data centre operates at optimal performance.

The Drawbacks of Traditional Data Centre Network Architectures

Traditional data centre networks are typically based on Layer 2 connectivity, utilising mainly Layer 2 switches and at least one pair of Layer 3 devices, often routers or firewalls. Designed around a three-tier architecture (access, aggregation, and core layers), these networks have revealed several significant issues when faced with modern data centre demands.

  • Limited Scalability: Traditional three-tier architecture struggles with horizontal scaling, leading to bandwidth bottlenecks and an inability to meet the needs of contemporary data centres.

  • High Latency and Inefficiency: The multi-layer design increases network latency, particularly as east-west traffic grows, with traffic bottlenecking at the core layer, resulting in inefficiencies.

  • Redundancy and Reliability Challenges: Complex redundancy designs increase network complexity and costs, and a failure at the core layer can cause widespread network outages.

  • Complex Management and Operations: Traditional architectures rely on manual configurations, and compatibility issues between devices, combined with low levels of automation, make management and operations complex and time-consuming.

Understanding FS IP Fabric Solution Architecture

Efficient Spine-Leaf Architecture

Next-generation data centres based on Spine-Leaf architecture have been around for years and are a mature and stable technology. FS IP Fabric solution is based on the Fat-Tree Spine-Leaf topology, which facilitates seamless scaling by adding more Spine nodes. This architecture avoids the bottlenecks that slow down network transmission speeds by utilising multiple high-bandwidth direct connections, ensuring high forwarding efficiency and low latency. This design is particularly beneficial for data centres that need to maintain optimal performance under high traffic conditions.

Spine Layer:

Deploying the N8560-32C switches at the Spine layer ensures the network remains operational even under heavy traffic loads. Supporting VXLAN, EVPN, and other virtual network extensions, these switches facilitate unlimited data transmission. Interconnections between switches use QSFP-SR4-100G optical modules, which are easy to install and hot-swappable, making them suitable for various applications such as data centres, high-performance computing networks, and enterprise core and distribution layers.

Leaf Layer:

The N5860-48SC switch is an ideal choice for data centres and cloud computing services of all sizes. The switches are interconnected using MLAG technology, enhancing redundancy and network reliability. Additionally, SFP-10GSR-85 optical modules are used to connect with the server layer, ensuring efficient 10G bandwidth transmission.

Server Layer:

The server layer deploys two RS5230 servers, each equipped with X710BM2-2SP Ethernet network interface cards to provide sufficient bandwidth for line-rate traffic on dual 10Gb ports. They are connected to Leaf switches via LACP, achieving dynamic link management and redundancy.

FS IP Fabric Solution Architecture

Simplified Management: Ensuring Business Continuity and Flexibility

While ensuring high network performance, FS IP Fabric solution also optimises network operations through advanced management tools and technologies, ensuring business continuity and network flexibility.

Unified Management with PicOS® and AmpCon™

A highlight of FS IP Fabric solution is the unified management capability provided by the PicOS® software and AmpCon™ management platform. This combination simplifies network configuration, monitoring, and maintenance, effectively eliminating costly downtime and time-consuming manual tasks. The PicOS® switches, powered by Broadcom chips, support EVPN-VXLAN and MLAG technology, ensuring high network reliability and scalability. Through zero-touch provisioning, centralised updates, and intelligent backups, AmpCon™ ensures secure network expansion while reducing labour costs and preventing downtime.

PicOS® and AmpCon™

Enhanced Network Reliability with EBGP and ECMP

To guarantee business continuity, FS IP Fabric solution combines EBGP and ECMP for load balancing and high availability across autonomous systems. By configuring multiple EBGP peers and enabling ECMP, traffic can be forwarded across multiple equal paths, enhancing redundancy, performance, and reliability. This ensures that even in the event of network failures, routes are quickly recalculated and updated, minimising downtime.

Improved Network Virtualisation Efficiency with EVPN-VXLAN

FS IP Fabric solution also supports large-scale multi-tenant environments through EVPN-VXLAN, simplifying the network architecture while providing efficient network virtualisation, dynamic migration, and unified management. With 24-bit VNID, the solution supports up to 16 million virtual networks, making it ideal for data centres that need to scale and accommodate a large number of tenants. Moreover, EVPN-VXLAN supports seamless migration of virtual machines and containers, ensuring optimal network resource utilisation under dynamic workload changes.

Increased Bandwidth Utilisation with MLAG and LACP

The MLAG and LACP implementations by FS significantly enhance network bandwidth utilisation. MLAG combines multiple physical links into a single logical link, increasing total bandwidth. At the same time, LACP dynamically manages, monitors, and maintains the health of aggregated links, providing fast failure recovery and stable data transmission. This configuration ensures higher network reliability and efficient traffic distribution.

MLAG and LACP

Final Thoughts

If your new infrastructure deployment requires more than one switch and at least a two-layer design, Spine-Leaf is the optimal choice. FS IP Fabric solution is designed to meet the stringent demands of modern data centres, offering a perfect combination of performance, flexibility, and scalability. By leveraging advanced technologies such as Spine-Leaf architecture, PicOS® software, the AmpCon™ platform, and EVPN-VXLAN, FS ensures your data centre can operate efficiently while being prepared to face future challenges. Choose FS, and build a data centre network that is not only powerful but also ready for the future.

You might be interested in

Knowledge
Knowledge
Knowledge
See profile for Sheldon.
Sheldon
Decoding OLT, ONU, ONT, and ODN in PON Network
Mar 14, 2023
440.9k
Knowledge
See profile for Irving.
Irving
What's the Difference? Hub vs Switch vs Router
Dec 17, 2021
385.3k
Knowledge
See profile for Sheldon.
Sheldon
What Is SFP Port of Gigabit Switch?
Jan 6, 2023
379.0k
Knowledge
See profile for Migelle.
Migelle
PoE vs PoE+ vs PoE++ Switch: How to Choose?
May 30, 2024
465.0k
Knowledge
Knowledge
Knowledge
Knowledge
See profile for Moris.
Moris
How Much Do You Know About Power Cord Types?
Sep 29, 2021
335.7k