Data center interconnect – build or buy?
Many enterprises, industries, government and public sector organizations use managed data center interconnect (DCI) services to connect their data centers.
But as data volumes continue to increase, managed DCI services are becoming costly to scale. They are also inflexible with poor response times to changing needs. DCI solutions need to change to take full advantage of new agile, flexible and dynamic cloud IT models.
Some organizations are implementing private DCI based on dark fiber as a strategic investment to support private and hybrid clouds. But what is the business case for building private DCI? And how is the cost justified when managed DCI services are widely available?
Data center challenges
Large organizations face a number of data center challenges including:
- Data volumes are constantly increasing due to trends such as Big Data, mobility, IoT and bring your own device or application
- It is becoming too expensive to expand or build more data centers and keep hardware and software up to date
- Organizations are under pressure to reduce complexity and cost, such as reduce the space and power their data centers consume to meet green obligations
- Organizations want to avoid technology obsolescence and increase flexibility and agility by using cloud-based IT.
To address these challenges, large organizations are consolidating and transforming their data centers so they benefit from economies of scale. They are also implementing data center virtualization and SDN to improve server and storage efficiency and increase flexibility. Finally, they are moving to a cloud IT model to enable agility, flexibility and lower costs.
By moving to a cloud IT model, organizations can use different types of cloud to suit different business needs. For example, they can use private and virtual private clouds where they need security and control for business-critical applications and data, and hybrid and public clouds where they need agility and flexibility for non-business-critical applications and data.
Cloud IT helps organisations to scale quickly and cost efficiently to match demand for compute and storage. It supports an IT model that matches expense to need, preserves cash and increases business agility.
Cloud forces a re-evaluation of DCI needs
Data center consolidation, growth in inter-data center traffic and moving to cloud increases DCI bandwidth needs. As a result, organizations need to reconsider their DCI requirements.
Managed DCI network services are becoming expensive to scale, are inflexible and take too long to provision. They are simply not agile and flexible enough to meet the dynamic nature of the cloud. In addition, many organizations want more control over their DCI, particularly to ensure data compliance, integrity, security and sovereignty.
For these reasons, and to cope with the increase in bandwidth needs that cloud brings, many organizations are building their own private DCI using dark fiber. This approach is best for organizations that need very high bandwidth, physical security and complete control of their data centers, network and the data that flows over it.
Private DCI build using dark fiber
With a private DCI build, organizations lease or buy dark fiber between their data center locations, and install and manage their own optical DWDM equipment. These provide scalable bandwidth and offer the best cost-performance ratio and lowest latency required for many DCI applications. Organizations can also reduce the risk and time to market by using a partner such as a network integrator.
In many cases, the cost of building a private DCI solution is offset significantly by the cost savings achieved through consolidating multiple data centers for critical applications and data and moving to a cloud IT model for additional capacity, agility and flexibility when needed.
Figure 1 summarizes the benefits of a private DCI compared to a managed DCI service. It compares a prior mode of operation (PMO) on the left in which multiple data centers are connected using a managed DCI service provided by a service provider with a future mode of operation (FMO) on the right in which consolidated and virtualized data centers are connected using a private DCI build based on dark fiber.
Figure 1. Summary of the benefits of private DCI build versus managed DCI services
The benefits of the private DCI build include:
- OPEX savings and improved server utilization as a result of consolidating and reducing the number of data centers and introducing virtualization.
- Greater scalability and performance as well as greater resiliency at less cost and lower complexity by using leased dark fiber with diverse paths
- Better control and security with increased agility and much higher bandwidth at incremental cost by installing and managing optical DWDM equipment.
When does private build DCI make sense?
Overall, the costs of a private cloud DCI solution can be significantly less than a managed DCI solution, especially when consolidating and virtualizing data centers. The cost of dark fiber linking fewer data centers is often less than the cost of managed DCI services between multiple data centers, particularly when future bandwidth needs are included.
While the monthly recurring cost of a 10G managed Ethernet or managed wavelength service may be less than a dark fiber solution, for bandwidth in the range of 10G to 40G private becomes cost effective as managed DCI services become cost prohibitive at bandwidths of greater than 10G.
For bandwidth greater than or equal to 40G, private DCI build using leased fiber can be a good economical choice, particularly as bandwidth can be increased at marginal cost by lighting additional wavelengths on existing optical equipment.
A private DCI build business case
To illustrate the business benefits of a private DCI build, Nokia has developed a business case for a 100G private DCI build versus a 100G managed DCI service. It uses an example of a typical large enterprise with a total of 15 sites and a PMO that uses managed Ethernet Private Line (EPL) both for DCI and connectivity to a managed IP VPN service.
The business case compares the PMO, in which DCI is provided using two point-to-point 100G managed Ethernet services for resilience, with a FMO using 100G private build DCI based on diverse dark fiber paths for resilience and purchased and self-managed DWDM optical equipment with redundant hardware. In both PMO and FMO, managed Ethernet services provide access connections at each site to the service provider’s local central offices where they connect to the service provider’s IP VPN service.
Figure 2 shows the incremental cash-flows generated by the FMO. The red line shows a cumulative view of the incremental cash flows generated. The red line below the zero axis shows the initial period where investment is required. As the project progresses, the red line crosses the zero axis, showing that the project has generated as much cash flow as the PMO would have. From this point forward, the project generates higher cash flows than not doing the project.
Figure 2. Incremental cash flows generated by the FMO
The bars in Figure 2 show the incremental cash inflows and outflows. The blue bars show CAPEX for the DWDM optical hardware and software required for the project, with peak funding occurring in the second quarter of the project.
The grey bars above the zero axis show the project generates net OPEX savings from the first quarter. The OPEX assumptions include a one-time initial set up cost to cover connection to the local dark fiber access infrastructure, plus monthly fiber maintenance and support costs. Hardware and software maintenance costs and network support costs are also included.
The FMO has a discounted payback period, or breakeven point, occurring in the 12th month or the fourth quarter after the start of the project. Reducing the OPEX, for example by using a network integrator to provide hardware and software maintenance and network support externally at a lower cost, would result in higher cash flows and a reduced payback period.
Finally, in the FMO increasing bandwidth from 100G to 200G can easily be achieved by lighting additional wavelengths over the dark fiber at a marginal incremental cost, as the CAPEX assumptions include the cost of additional hardware for future bandwidth increases. Increasing bandwidth in the PMO using additional managed Ethernet or wavelength services would incur significant additional cost – assuming that additional capacity is readily available.