We’ve just completed a new Network Topology POC this week, for an existing customer, that will now add hardware and connectivity redundancy from the customer’s ISP (Cogent) to the customers’ existing HPE C7000 Chassis, that are full of HPE BL460c Gen8 Blade Servers, located across 19x Data Centres around the world.
Current Setup
The customer’s current setup consisted of 10Gbit Cogent ISP Links directly connected into the back of the HPE Virtual Connect Switches with a static route, routing a /24 public address range down each 10Gbit link.
The customer’s current setup offers no hardware or routing redundancy if the customer has a 10Gbit ISP Link, HPE Virtual Connect Switch, Cable or SFP failure.
For example, when a HPE Virtual Connect Switch failure occurs, which has happened recently, the Data Centre Remote hands have to be called and then have to physically move the 10Gbit link from the failed HPE Virtual Connect Switch, into a spare port on the working HPE Virtual Connect Switch. The customer then has to temporarily reconfigure the HPE Virtual Connect Domain Server Profiles for each HPE BL460c Gen8 Blade Server that are connected to the VLAN Network on the failed switch.
Once the failed HPE Virtual Connect has been replaced, Data Centre Remote hands have to swap the 10Gbit link back into the replacement HPE Virtual Connect Switch. The HPE Virtual Connect Domain Server Profiles have to be reconfigured again, back to the original VLAN Network.
All of the above causes long periods of downtime to each HPE BL460c Gen8 Blade Server connected to the failed HPE Virtual Connect Switch.
New Setup
The customer came to Intelligent Servers and asked if we could redesign the network configuration to allow for hardware, Layer 2 and Layer 3 routing failure, ensuring that each HPE BL460c Gen8 Blade Server has a redundant path, to and from the Internet, in the event of hardware or ISP Uplink failure.
With the customers’ requirements in mind, we created and tested a new network topology using Cisco Nexus 3064PQ-10GX acting as the Core switches for each Data Centre location, with the following features enabled:
- Inter-VLAN Routing
- LACP (Link Aggregated Control Protocol)
- vPC (Virtual Port Channel)
- VRRP (Virtual Router Redundancy Protocol)
With these features enabled and configured correctly, the customer can now afford to have a single hardware failure, or multiple uplink failure at both the ISP (Cogent), Core (Cisco Nexus Switches), and/or the Access Layer (HPE Virtual Connect Switches).
Switch Hardware, ISP 10Gbit Uplink, HPE Virtual Connect Uplink, Layer 2/Layer 3 redundancy and ECMP (Equal Cost Multi-Pathing) is now in place, because of the configurations listed below, that have been tested using a Virtual Lab (EVE-NG) as well as been tested using the same Physical Hardware in our Intelligent Servers test facility:
- Cogent (ISP) have created LAG (Link Aggregation Group) bundles of the 10Gbit ISP Links which provide both Uplink Failure Redundancy and ECMP (Equal Cost Multi-Pathing) from Cogent to the customer’s core Cisco Nexus Switches.
- Cogent (ISP) now routes the customers /24 Public IP Subnets to the VRRP Virtual Gateway IP Address of the 10Gbit WAN Peering Links /29 Subnet configured on the Cisco Nexus Switches.
- Each customer /24 Public Subnet now has an SVI VLAN Interface on each Cisco Nexus Switch, with the first IP address of the /24 subnet configured as the VRRP Virtual Gateway IP Address for the customers HPE BL460c Gen8 Blade Servers to use as their Default Gateway.
- Cisco Nexus vPC (Virtual Port Channel) groups have been created at the Cisco Nexus Switch end between the Cisco Nexus Switches and the HPE Virtual Connect Switches, to protect against Core Switch and Uplink Failure. VRRP (Virtual Router Redundancy Protocol) complements vPC to enable ECMP (Equal Cost Multi-Pathing) load balancing across the Uplinks between the ISP, Core (Cisco Nexus) and Access Layer (HPE Virtual Connect) Switches.
This configuration was created and tested by our in-house Cisco CCNA engineer with input from our in-house Cisco CCIE engineer. If you have something similar that you would like to achieve for your own network infrastructure, please get in touch with us via our main phone number +44(0)1423 223430 or speak to your existing Sales Account Manager who will arrange a call with one of our Cisco Certified engineers.