Controlled Congestion Can Increase Margins
Economy datacenter networks providing multiple unmetered services are growing rapidly. Adding transit capacity to meet customer demands can be challenging and costly. Growth that exceeds forecasts, suprising loops costs, and carrier delays can all result in a degraded service for high and low-traffic customers alike. By applying modern quality-of-service features to such a datacenter network, economy operators can regain control of the quality of their service offerings and retain low-use customers whlie maximizing margins.
As an example, let us compare three customers with different utilization patterns, a transit cost of $10Mbps, and varying revenue levels and service plans. Also assume similar provisioning and administrative overhead costs per-account. This motivates the datacenter operator to provide the best service available to high-margin customers. When necessary, service quality for low-margin customers may be sacrificed.
| A | Port Speed | Peak Use | EF Queue | Transit CoGS | Revenue | Margin |
|---|---|---|---|---|---|---|
| A | 10Mbps | 5Mbps | 100.00% | $50 | $99 | $49 |
| B | 20Mbps | 20Mbps | 50.00% | $200 | $239 | $39 |
| C | 100Mbps | 100Mbps | 10.00% | $1,000 | $999 | -$1 |
| G | 1Gbps | 400Mbps | 40.00% | $4,000 | $4,800 | $800 |
Of our four example customers, G generates the most margin on a per-account basis. Customers A generates the highest per-Mbps margin. Customer B consumes a good deal of network resources, but still produces more margin than Customer C, whose peak utilization actually works out to a loss.
In an uncongested network with plenty of headroom on infrastructure and transit circuits, all customers should receive high throughput, have minimal packet loss, and in general, be happy with network performance.
In a congested network without quality-of-service features, however; all customers will experience some degree of packet loss and reduced throughput. This may be acceptable to customers B and C, who may select the datacenter operator's product based largely on cost; however customers A and G are more likely to balance cost and quality concerns, and may choose another vendor if they experience frequent or heavy congestion.
Quality of Service ("QOS") can eliminate this problem and allow the datacenter operator to exert control over customer's traffic by classifying none, some, or all of each customer's traffic into and expedited forwarding ("EF") queue, treating EF traffic with greater priority than best effort ("BE") traffic.
The datacenter operator gains the ability to serve both cost-conscious, low-margin customers as well as quality-minded, higher-margin customers over the same network infrastructure.
INC StaffProprietary
