Network Planning

Last updated: 2020-11-11 11:58:33

    Read this document to learn about the connection planning before building a network architecture for Direct Connect.

    Background

    A connection planning improves the stability and high availability of the network architecture for Direct Connect, and minimizes the impact from failures including device, port/fiber optic component, connection and data center at the access point. See the table below for the description and cause of failures.

    Failure Description Cause
    Connection The connection communication fails or many packets are discarded. The connection is damaged. For example, the cable is cut off.
    Port/fiber optic component The port/fiber optic component fails to be read or has an transfer exception due to hardware or software failures.
    • Hardware failure: incompatible version, contaminated or damaged port.
    • Software failure: incorrect port status, port in **UP** status without receiving or sending packets, frequent port on/off, or cyclic redundancy check (CRC) error.
    Network device The exchange or router on the IDC side is unavailable.
    • Hardware failure: power supply, port, component or cable failures.
    • Software failure: exchange/router error or incorrectly configured.
    Data center at the access point The data center encounters network disconnection, or the access point is unavailable. The data center cannot function normally due to earthquakes, fires or other disasters.

    Planning Ideas

    Capacity panning

    The capacity planning is designed to meet business bandwidth requirements at a reasonable cost and guarantee the business operation in the event of a connection failure. To achieve this, you can:

    • Apply for connections twice your actual needs.
    • Maintain the connection utilization (current peak bandwidth/connection bandwidth * 100%) 50% or less.

    Assume your business needs 3 Mpbs of bandwidth, you can apply for two connections with 5 Mpbs of bandwidth to each Tencent Cloud access point, each connection will be used about 30%. If one connection faults, the business traffic will quickly switch to the standby connection to ensure the business continuity. After the switchover, the standby connection will be used about 60%. In this way, only the connection load temporarily increases, and the business data is unaffected.

    Expansion planning

    The expansion planning is designed to meet the surging business demand at a reasonable cost. Depending on the expansion cycle, you need to plan differently as follows:

    • If you need an urgent expansion, you can apply for a connection based on the estimated business bandwidth, so that you can adjust the bandwidth limit to maintain your business when the traffic surges.
    • If you don’t need an expansion in the near future, you can apply for a new connection based on your actual expansion needs. The expansion will take 2-3 months.

      Note:

      A single expansion exceeding 100 Gbps of bandwidth requires a longer period. Please develop an expansion plan in advance for high-bandwidth businesses.

    Disaster recovery planning

    The disaster recovery is designed to improve the high availability of a network architecture and minimize the failure (including port/fiber optic component, network device, and data center at the access point) impact on business operations.
    To avoid single points of failures, we recommend that you develop a disaster recovery plan for access points, physical lines, and hardware devices.

    • Access point: the local IDC uses connections of different physical lines to two intra-region Tencent Cloud access points.
    • Physical line: different physical lines provided by carriers are used to connect to a Tencent Cloud access point.
    • Hardware devices: both network devices and fiber optic components have redundant backups.

    Assume you connect your local IDC to two intra-region Tencent Cloud access points according to [capacity planning (#capacity) requirements, as shown in the figure below. If the connection to access point A is disconnected due to port/fiber optic component, network devices or data center failures, the business traffic will quickly switch to the connection to access point B, without losing business data or interrupting businesses.

    Sample Architectures

    Tencent Cloud provides the following four network architectures for Direct Connect to facilitate your network planning.

    Architecture Use Case Business Elasticity Availability Cost
    Four connections to two access points (recommended) Suitable for use cases that require excellent business availability and elasticity, including the key production and real-time data transaction. Highest Highest Highest
    Two connections to two access points (recommended) Suitable for key business that requires high availability and elasticity. Medium Higher Medium
    Two connections to one access point Suitable for non-critical business, including the cloud development and testing environments. Medium Medium Medium
    Connection to one access point Suitable for non-critical business that does not require high elasticity and availability. Low Low Low

    Four Connections to Two Access Points (Recommended)

    Overview

    • Description: this architecture connects your IDC to two intra-region Tencent Cloud access points using primary and secondary connections respectively, and then accesses Tencent Cloud VPCs.
    • Use cases: suitable for use cases that require excellent business availability and elasticity, including the key production and real-time data transaction.
    • Cost: high.

    Disaster recovery

    The following table describes the impact of this network architecture on business under various failures when the planning ideas are met.

    Failure Business Impact
    Connection The connections loads increase, and the business continues.
    Port/fiber optic component The connections loads increase, and the business continues.
    Network device The connections loads increase, and the business continues.
    Data center at the access point The connections loads increase, and the business continues.

    Two Connections to Two Access Points (Recommended)

    Overview

    • Description: this architecture connects your IDC to two intra-region Tencent Cloud access points using one connection respectively, and then accesses Tencent Cloud VPCs.
    • Use cases: suitable for key business that requires high availability and elasticity.
    • Cost: medium.

    Disaster recovery

    The following table describes the impact of this network architecture on businesses under various failures.

    Failure Business Impact Recovery Measures
    Connection
    • If each connection is used 50% or less, the connection loads burst and the business continues.
    • If each connection is used greater than 50%, the connection will be full-loaded, some data will be lost, and the business continues.
    If the business data is lost, you need to apply for a new connection and spend 2-3 months to restore the business.
    Network device
    • If each connection is used 50% or less, the connection loads burst and the business continues.
    • If each connection is used greater than 50%, the connection will be full-loaded, some data will be lost, and the business continues.
    If business data is lost, you need to check and repair the network devices. The recovery time depends on the specific failure.
    Port/fiber optic component
    • If each connection is used 50% or less, the connection loads burst and the business continues.
    • If each connection is used greater than 50%, the connection will be full-loaded, some data will be lost, and the business continues.
    If business data is lost, you need to check and repair the ports or fiber optic components. The recovery time depends on the specific failure.
    Data center at the access point
    • If each connection is used 50% or less, the connection loads burst and the business continues.
    • If each connection is used greater than 50%, the connection will be full-loaded, some data will be lost, and the business continues.
    If business data is lost, you can:
    • Contact the data center carrier to repair the failure. The recovery time depends on the specific failure.
    • Reapply connections to other access points. The recovery takes about 2-3 months.

    Two Connections to One Access Point

    Overview

    • Description: this architecture connects your IDC to one Tencent Cloud access point using two connections, and then accesses Tencent Cloud VPCs.
    • Use cases: suitable for non-critical business, such as cloud development and testing environments.
    • Cost: medium.

    Disaster recovery

    The following table describes the impact of this network architecture on businesses under various failures.

    • If each connection is used 50% or less, the connection loads burst and the business continues.
    • If each connection is used greater than 50%, the connection will be full-loaded, some data will be lost, and the business continues.
    Failure Business Impact Recovery Measures
    Connection If the business data is lost, you need to apply for a new connection and spend 2-3 months to restore the business.
    Network device The business will be interrupted. You need to troubleshoot and repair the faulty network devices. The recovery time depends on the specific failure.
    Port/fiber optic module The business will be interrupted. You need to troubleshoot and repair the faulty local ports or fiber optic components. The recovery time depends on the specific failure.
    Data center at the access point The business will be interrupted.
    • Contact the data center carrier to fix the failure. The recovery time depends on the specific failure.
    • Reapply connections to other access points. The recovery takes about 2-3 months.

    A Connection to One Access Point

    Overview

    • Description: this architecture connects your IDC to one Tencent Cloud access point using a connection, and then accesses Tencent Cloud VPCs.
    • Use cases: suitable for non-critical business that does not require high elasticity and availability.
    • Cost: medium.

    Disaster recovery

    The following table describes the impact of this network architecture on businesses under various failures.

    Failure Business Impact Recovery Measures
    Connection The business will be interrupted. Reapply for a connection. The recovery takes about 2-3 months.
    Network device The business will be interrupted. You need to troubleshoot and repair the faulty network devices. The recovery time depends on the specific failure.
    Port/fiber optic module The business will be interrupted. You need to troubleshoot and repair the faulty local ports or fiber optic components. The recovery time depends on the specific failure.
    Data center at the access point The business will be interrupted.
    • Contact the data center ISP to fix the failure. The recovery time depends on the specific failure.
    • Reapply connections to other access points. The recovery takes about 2-3 months.

    Was this page helpful?

    Was this page helpful?

    • Not at all
    • Not very helpful
    • Somewhat helpful
    • Very helpful
    • Extremely helpful
    Send Feedback
    Help