Data Center Fabric

Data Center Fabric

Geo-redundant data center with "Leaf & Spine" architecture for a financial company.

Performance infrastructure with a focus on maximum high availability and business continuity

A southern German financial service provider outsourced its data center, previously hosted by itself, to a housing service of two regional data centers.

Due to historical "low-level" applications, the customer required a data center design that offers connectivity and flexibility at the data link layer (OSI Layer 2) without bringing with it the disadvantages of a switched design. However, the primary focus of the infrastructure should be maximum high availability and business continuity (BCM).

In cooperation with Cisco Solution Engineers, BEUL developed a concept for a geo-redundant data center network that spans a "Leaf & Spine" architecture across both data center locations. "Cisco Fabric Path" is used as an overlay, a TRILL implementation that combines the advantages of a switched and routed design. Here, Cisco Nexus switches of the 7000, 5000 and 2000 series were used as ToR and EoR systems, which, in addition to channelized 10G Ethernet, also provide 8G Fiber Channel and 10G FCoE for connecting servers and storage. The data center locations themselves were multi-redundantly coupled using DWDM technology on Dark Fiber.

BEUL prepared the turnkey DC infrastructure for the customer in the new data centers and accompanied the entire relocation process, in particular the multi-day "hot" migration phase with increased work shifts at night and at weekends.

Motivation

  • Cost and migration pressure from moving to a new data center environment
  • Implementation of a symmetrical data center and DMZ zone for secure business continuity in the event of a disaster (through clustering and geo-redundancy)
  • Consolidation of the previous data center volume through switch aggregation and SAN integration

Challenges

  • Elaborate and lengthy server migration due to the hardware volume with 24/7 on-site service, installation shifts and on-call service
  • Maintaining standard operations through staged migration
  • Replacement of the existing SAN switches with Multiprotocol Nexus systems with Ethernet, FCoE and FC

Technologies

images/technologien/cisco.png
images/technologien/datwyler.png
images/technologien/hp-enterprise.png
images/technologien/radware.png

The project in figures

2
Geo-redundant data centers
72
Hours of shift work for system relocations
500
servers and storage systems
1.200
Virtual Machines
10.000
meters of fiber optic and copper cable
20.000
kilograms of network and server hardware