Design network backhaul for DC-2-DC connectivity
Budget: $250 – $750 AUD
We need to link two data centers together, however, this is not possible by dark fiber due to the distance between the two sites. We have resulted to either a managed backhaul link or wavelength service between the two sites. The wavelength or backhaul will be 100 Gbps, and will need to offer the ability for managed service providers and co-location customers to get a cross-connect from us, in order to get to the other side and vice versa. Theoretically, a point-to-point link, but with 144 ports at either end that are segregated.
I personally have a somewhat decent understanding of networking including BGP, DHCP, DNS, etc.. but i'm not familiar with how to go about this as there needs to be the ability to have up to 144 cross-connects between the two sites, meaning that everything would need to be VLAN'ed or something similar (I think). There might be a better way.
Here is what I'm after:
- A solution that is scalable and cost-effective
- The ability to expedite some customer's traffic over the network if they require mission-critical latency and no packet discards, and even provide customers with different classes of service.
- The ability to simply configure a port at either end of the wavelength or backhaul link (at either site), to allow the flow of data for the customer.
- A detailed network diagram and explanation on what the best approach to this would be, what's involved, the configuration, alternative options (if any), and your opinion on what's best.
*Our upstream provider who will be providing the managed backhaul or wavelength service will simply hand off their service with a network termination unit, and provide us with an SFP out of one of the fibre ports. From there, we would bring a patch between that NTU port and whatever equipment we have to deliver these cross-connects.
**Our upstream provider will be providing fully geo-diverse routing to the site, meaning there will be two NTUs at either end (2 at DC1, 2 at DC2). When one link fails, it moves to the other. This will need to be accounted for when you design the network. Also, I don't believe we can run active DWDM over the NTUs because they are also using DWDM in the upstream's core transport network, so we would just need data center grade switches (correct me if I'm wrong).
I personally have a somewhat decent understanding of networking including BGP, DHCP, DNS, etc.. but i'm not familiar with how to go about this as there needs to be the ability to have up to 144 cross-connects between the two sites, meaning that everything would need to be VLAN'ed or something similar (I think). There might be a better way.
Here is what I'm after:
- A solution that is scalable and cost-effective
- The ability to expedite some customer's traffic over the network if they require mission-critical latency and no packet discards, and even provide customers with different classes of service.
- The ability to simply configure a port at either end of the wavelength or backhaul link (at either site), to allow the flow of data for the customer.
- A detailed network diagram and explanation on what the best approach to this would be, what's involved, the configuration, alternative options (if any), and your opinion on what's best.
*Our upstream provider who will be providing the managed backhaul or wavelength service will simply hand off their service with a network termination unit, and provide us with an SFP out of one of the fibre ports. From there, we would bring a patch between that NTU port and whatever equipment we have to deliver these cross-connects.
**Our upstream provider will be providing fully geo-diverse routing to the site, meaning there will be two NTUs at either end (2 at DC1, 2 at DC2). When one link fails, it moves to the other. This will need to be accounted for when you design the network. Also, I don't believe we can run active DWDM over the NTUs because they are also using DWDM in the upstream's core transport network, so we would just need data center grade switches (correct me if I'm wrong).