ナビ付きでトピックを開く

 

External Web Servers (Customer Origin Group)

Applies To:
HTTP LargeThis platform has been optimized to cache and deliver static content (e.g., HTML, CSS, JavaScript, ISO, multimedia, and software downloads, etc.) over HTTP or HTTPS.
HTTP SmallLegacy. If you are currently serving traffic over this platform, then you may continue to do so. However, we recommend that you serve your traffic over our more robust HTTP Large platform.
ADNThe Application Delivery Network platform has been optimized to deliver dynamic content (e.g., login credentials, account information, etc.) over HTTP or HTTPS. Typically, user-specific and database-driven content are served over this platform.

This article explains how to set up customer origin groups. Customer origin groups is a new capability. If your account has not been updated to use this capability, please refer to the External Servers (Customer Origin - Legacy) article for information on how to set up a customer origin.

ClosedHow can I tell whether my account supports customer origin groups?

From the main menu, navigate to [HTTP Large, HTTP Small, or ADN] | Customer Origin.

If your account supports customer origin groups, it should look similar to the following illustration:

If it still uses the legacy method for customer origin configuration and therefore doesn't support customer origin groups, it will look like the following illustration:

Customer origin groups allow you to leverage our built-in support for connecting to new or preexisting Azure block blob containers. Contact your CDN account manager to upgrade your account to support customer origin groups.

Serve content stored or generated by third-party web servers (e.g., web hosting) via the CDN by performing the following steps:

  1. Optional. Prepare for HTTPS delivery by requesting a TLS certificate.

    HTTPS delivery requires the installation of a TLS certificate on our network. Please contact your CDN account manager to request HTTPS support.
    Learn more.

  2. Create a customer origin group.

    A customer origin group maps your origin (e.g., your web servers) to a CDN URL through which your content may be served. A CDN URL consists of a system-defined base URL followed by the relative path to your content.

  3. Optional. Create an edge CNAME configuration.

    Use an edge CNAME configuration to serve traffic via the CDN without having to update your links. This type of configuration maps a customer origin group to a CNAME recordA Canonical Name (CNAME) record is used to indicate that a hostname is an alias of another hostname. A CNAME record must be registered on a Domain Name System (DNS). This term should not be confused with edge CNAME..

  4. Verify that your firewall accepts HTTP and/or HTTPS traffic from our network. Additionally, requests that originate from our network should not be rate limited.

    Learn more.

Key information:

Creating a Customer Origin Group

This section provides step-by-step instructions on how to create a customer origin group.

Key information:

The maximum number of customer origin groups per platform is 100.

To create a customer origin group

  1. Navigate to the Origins page corresponding to the desired platform. ClosedHow?From the main menu, navigate to [HTTP Large, HTTP Small, or ADN] | Customer Origin.
  2. Click Create.
  3. Click Customer Origin.

  4. Click within the Group option and then perform the following steps:

    1. Type the name that will be assigned to the new customer origin group.

      Use the following characters when defining this name: alphanumeric, hyphens, periods, and underscores.

      This name will be incorporated into the CDN URL (e.g., http://wpc.0001.omegacdn.net/800001/Customer Origin Group).

    2. Click + Add "Customer Origin Group Name".

  5. In the Name option, type the name by which you will identify the hostname or IP address being defined.

    Use the following characters when defining this name: alphanumeric, hyphens, periods, and underscores.

  6. Identify your web server(s) and how our network will communicate with them via one of the following methods:

    • End-to-End Encryption: This type of setup only allows the HTTPS protocol for the communication between clients, our network, and your web servers.

      ClosedView steps.

      HTTPS delivery requires the installation of a TLS certificate on our network. Please contact your CDN account manager to request HTTPS support.
      Learn more.

      1. From the Protocol Type option, select HTTPS Only.

      2. Set the Hostname or IP Address option to either an IP address or a hostname that points to your web server(s).

        Example:

        origin.example.com
      3. Set the Port option to the port over which our network will serve traffic to the above hostname or IP address.

    • Client-to-Edge Encryption: This type of setup only allows the HTTPS protocol for communication between clients and our network. Communication between our network and your web servers will be performed using the HTTP protocol.

      ClosedView steps.

      HTTPS delivery requires the installation of a TLS certificate on our network. Please contact your CDN account manager to request HTTPS support.
      Learn more.

      1. From the Protocol Type option, select HTTPS Only.

      2. Set the Hostname or IP Address option to http:// followed by either an IP address or a hostname that points to your web server(s).

        Example:

        http://origin.example.com

      3. Set the Port option to the port over which our network will serve traffic to the above hostname or IP address.

    • HTTP Only: Only allow the HTTP protocol for the communication between clients, our network, and your web servers.

      ClosedView steps.

      HTTPS requests cannot be served by a hostname or IP address set to this mode.

      1. From the Protocol Type option, select HTTP Only.

      2. Set the Hostname or IP Address option to either an IP address or a hostname that points to your web server(s).

        Example:

        origin.example.com

      3. Set the Port option to the port over which our network will serve traffic to the above hostname or IP address.

    • Match Client: Match the protocol defined in the client's request.

      ClosedView steps.

      HTTPS delivery requires the installation of a TLS certificate on our network. Please contact your CDN account manager to request HTTPS support.
      Learn more.

      Our network will serve HTTP traffic to your web servers over port 80, while HTTPS traffic will be served over port 443.

      1. From the Protocol Type option, select Match Client.

      2. Set the Hostname or IP Address option to either an IP address or a hostname that points to your web server(s).

        Example:

        origin.example.com

  7. From the Primary option, determine how requests will be load balanced when multiple hostnames or IP addresses have been associated with a customer origin group.

    • Round-Robin: Make sure that the Primary option is disabled.
    • Primary/Failover: If the current hostname or IP address is the primary web server, then enable the Primary option. Otherwise, disable it.

  8. From the Group Settings section, click HTTP Host Header. Verify that the Host Header option is set to one of the following:

    • A hostname defined in step 5.
    • An edge CNAME whose configuration points to this customer origin.
    • Blank. The request will determine the value assigned to the Host request header.

  9. ADN Only

    Perform the following steps when creating a customer origin group for the ADN platform:

    1. Upload a sample asset to each server that has been specified for this customer origin group.

      Our servers use this asset to choose the optimal ADN Gateway servers and data routes for your customer origin group. The optimal file size for this sample asset is 5 KB.

    2. From the Group Settings section, click ADN Gateway.

    3. In the Validation Path option, specify a URL that points to a sample asset on your customer origin server.

      The hostname specified in this URL should match the one defined in the Host Header option unless that option has been set to blank.

    4. Mark or clear the Follow redirects option according to whether our service will respect a URL redirect for the URL defined in the previous step.
    5. Click Validate. If the result returns "200 OK" for all hostnames/IP addresses, then proceed to the next step.

  10. Origin Shield Only

    Origin Shield, which must be purchased separately, is available on the HTTP Large and HTTP Small platforms.

    • Set up origin shield on a customer origin by performing the following steps:

      1. From the Group Settings section, click Origin Shield.
      2. Enable the Origin Shield option.

      3. Perform one of the following:

        • Set up a recommended origin shield configuration by enabling the Single POP option. You should then select the POP closest to your customer origin server(s) from the ALL POPs list.

        • Create a custom origin shield configuration by enabling the Multiple POPs option and then selecting the origin shield action that will take place for each region.

          Learn more.

    • Set up a customer origin to handle all requests that are not served from cache by disabling the Origin Shield option.

  11. Optional. From the Group Settings section, click Resolve to IPv6 or IPv4. Review whether requests will be resolved to IPv4 or IPv6.

    Learn more.

  12. HTTPS Delivery Only

    Optional. If your customer origin group supports HTTPS delivery, click Origin TLS from the Group Settings section. Review your SNI setup and whether self-signed certificates will be allowed.

    Learn more.

  13. Click Save to save your customer origin group.

  14. Optional. Add another hostname or IP address to your customer origin group.

    1. Click Create.
    2. From the Group option, select the customer origin group created above.
    3. Repeat steps 4 - 6.
    4. Click Save.
    5. Repeat as needed.
  15. Verify that the firewall for your web servers does not restrict access to the IP address blocks listed within the Whitelist IP Blocks page.

Modifying a Customer Origin Group

A customer origin group contains individual origin entries and settings that apply across the entire group. Modify a customer origin group by clicking within one of its origin entries.

If an edge CNAME points to a customer origin group, then you will not be allowed to switch one of its origin entries to a different customer origin group. Consider creating a new customer origin group. Alternatively, you can delete the edge CNAME configuration and then switch the desired origin entries to a different customer origin group.

It may take up to an hour for changes to your customer origin group to take effect.

To modify a customer origin group

  1. Navigate to the Origins page corresponding to the desired platform. ClosedHow?From the main menu, navigate to [HTTP Large, HTTP Small, or ADN] | Customer Origin.
  2. Click on the desired customer origin group to expand it.
  3. Click on the desired origin entry.
  4. Make the desired changes to the origin entry. See steps 4 - 6 from the To create a customer origin group procedure above.
  5. Make the desired changes to group settings. See steps 7 - 11 from the To create a customer origin group procedure above.
  6. Click Save to save your customer origin group.

Deleting a Customer Origin Group

Permanently delete a customer origin group by deleting all of its origin entries.

If an edge CNAME points to a customer origin group, then you will not be allowed to delete its last origin entry. Delete the corresponding edge CNAME configuration and then delete the last origin entry.

It may take up to an hour for customer origin group deletions to take effect.

To delete a customer origin group

  1. Navigate to the Origins page corresponding to the desired platform. ClosedHow?From the main menu, navigate to [HTTP Large, HTTP Small, or ADN] | Customer Origin.
  2. Click on the desired customer origin group to expand it.
  3. Hover over the desired origin entry and then click . When prompted, click Delete to confirm the deletion of the origin entry.
  4. Repeat the previous step until you have deleted all of the origin entries associated with the desired customer origin group.

Customer Origin Group Name

The Group option uniquely identifies your customer origin group. This name is included as a URL segment within a CDN URLA system-defined URL that points to a CDN hostname. A CDN URL allows content delivery via our network. Simplify your CDN setup by also creating an edge CNAME configuration which potentially allows you to deliver traffic via the CDN using the same links as your current setup. as indicated below.

Syntax:

Protocol://CDN Hostname/80Indicates that a standard HTTP/HTTPS request should be forwarded to the customer origin identified in the CDN URL.ANThis term represents your customer account number (e.g., 0001) which can be found in the upper right-hand corner of the MCC./Customer Origin Group/Path/File Name

Key information:

Example

If the primary purpose of your web servers is to serve images, then you might create a customer origin group called images. An example of what a CDN URL for this type of customer origin might look like is provided below.

Sample CDN URL:

http://wpc.ANThis term represents your customer account number (e.g., 0001) which can be found in the upper right-hand corner of the MCC..omegacdn.net/80ANThis term represents your customer account number (e.g., 0001) which can be found in the upper right-hand corner of the MCC./imagesIdentifies a customer origin group by its directory name.

The above sample CDN URL points to the webroot on the server(s) associated with the images customer origin group. Append the desired relative path to the content that you would like to request. This relative path is highlighted in the following sample CDN URL:

http://wpc.0001.omegacdn.net/800001/images/photography/clientX/

Origin Entries

A customer origin group may consist of up to 10 origin entries per protocol. An origin entry identifies a set of web server(s) by hostname or IP address.

Create an origin entry by clicking Create from the Origins page. Set the Group option to the customer origin group to which the origin entry will be added.

Define a customer origin's web servers through the following options:

Option Description

Name

Assigns a friendly name to an origin entry.

Protocol Type

Determines whether an origin entry will fulfill:

  • HTTP Requests: Set this option to HTTP Only.
  • HTTPS Requests: Set this option to HTTPS Only.

  • HTTP and HTTPS Requests: You may either:

Please contact your CDN account manager to set up HTTPS traffic via a customer origin.
Learn more.

Hostname or IP Address

Identifies the web server(s) that will be associated with the current origin entry via a hostname or IP address. If you selected either the HTTPS Only or HTTP Only mode in the Protocol Type option, then you may also define a protocol for edge to origin communication.

You may not specify a hostname or IP address that points to our network. This type of configuration is disallowed since it may cause your traffic to infinitely loop within our network.

Use any combination of hostnames and IP addresses when defining a customer origin group's origin entries.

A hostname must be resolved to an IP address before a request can be forwarded to it. The Origin Configuration option defines how hostnames will be resolved to an IP address.
Learn more.

Syntax:

  • HTTPS Only and HTTP Only Modes:

    ProtocolOptional. Replace with either "http://" or "https://".Hostname or IP Address

    If you choose to assign a protocol, make sure that all origin entries within a customer origin group for a given mode use the same protocol. For example, you may not configure some HTTPS Only origin entries to use the HTTP protocol and others to use the HTTPS protocol.

    Hostname Example:

    https://www.mydomain.com

    IPv6 Example:

    [1:2:3:4:5:6:7:8]

  • Match Client Mode:

    Hostname or IP Address

    IPv4 Example:

    10.10.10.255

Brackets are required when identifying an origin server through the use of IPv6 notation. This is the standard URI convention for IPv6 addresses.

Port

Determines the port for edge to origin communication.

This option cannot be set when an origin entry has been configured to use Match Client mode.

Default value:

80 | 443

Key information:

HTTP Host Header

HTTP 1.1 requires a Host header to be sent with each request. A Host header identifies the hostname/IP address and port associated with a request. This information is especially useful when there are multiple virtual domains hosted on a single physical server or load-balanced set of servers.

Each customer origin group allows a Host header value to be configured. Typically, the Host header value should be set to either of the following:

If the Host Header option is set to blank, then the request URI determines the value for the Host request header.

Load Balancing

A load balancing configuration defines how traffic will be managed between the edge of our network and a customer origin group for a particular protocol (i.e., HTTP or HTTPS).

Traffic is load balanced when a customer origin group contains multiple origin entries for either the HTTP or HTTPS protocol. The customer origin group's load balancing configuration determines which IP address will handle the next request.

The available load balancing options are:

The above load-balancing options are completely independent from any load balancing configuration that may already distribute traffic to your web servers. For instance, traffic for a single IP address might be load balanced across several physical servers.

Unavailable Servers

A server is considered unavailable when either of the following conditions are true:

The manner in which an unavailable server affects load balancing is described below.

  1. Once a server is designated as unavailable, CDN traffic will not be load balanced to the corresponding hostname or IP address for a brief time period.
  2. Upon the expiration of this time period, CDN traffic may once again flow through the corresponding hostname or IP address according to the customer origin's load balancing configuration (i.e., round robin or primary & failover).
  3. If the server is still unavailable, then CDN traffic will not be load balanced to the corresponding hostname or IP address for a brief, but slightly longer time period.
  4. Steps 2 and 3 repeat until the server becomes available.

IP Version

Hostnames associated with a customer origin group must be resolved to an IP address before a request can be served to it. The Origin Configuration option determines whether our servers will prefer to resolve a hostname to an IPv4 or IPv6 address.

Setting Description

Default

Indicates that hostnames should be resolved to IPv4 addresses only.

We reserve the right to change the behavior of this default setting at any time.

V6 Preferred Over V4

Indicates that our edge servers will resolve hostnames to IPv6 addresses whenever possible. If an IPv6 address for that hostname does not exist, then the hostname will be resolved to an IPv4 address.

V4 Preferred Over V6

Indicates that our edge servers will resolve hostnames to IPv4 addresses whenever possible. If an IPv4 address for that hostname does not exist, then the hostname will be resolved to an IPv6 address.

V4 Only

Indicates that hostnames should be resolved to IPv4 addresses only.

V6 Only

Indicates that hostnames should be resolved to IPv6 addresses only.

TLS Verification

Configure how our edge servers perform TLS verification with your origin servers.

Please contact your CDN account manager to set up HTTPS delivery.
Learn more.

SNI for HTTPS Origin Requests

Our network can leverage Server Name Indication (SNI) during the TLS handshake. If the SNI hint is not found, then your origin server's implementation determines the TLS certificate that will be returned.

Additionally, an edge server will compare the hostname used for the SNI hint to the certificate's Subject Alternative Name (SAN) or Common Name (CN) during the TLS handshake. If the hostname does not match, then the edge server will respond with a 502 Bad Gateway response.

To enable SNI

  1. Modify the desired customer origin group.
  2. From the Group Settings section, click Origin TLS.
  3. Enable the Use SNI for HTTPS Origin Requests option.

  4. From the SNI Hostname option, define the hostname that will be sent as a SNI hint during the TLS handshake.

    Our service will also perform a strict check using this hostname against the certificate's Subject Alternative Name (SAN) or Common Name (CN) during the TLS handshake.

  5. Click Save to save your changes to your customer origin group.

Self-Signed Certificates

Our network can disable delivery when an edge server detects a self-signed certificate from the origin server during the TLS handshake. Enable the Disallow Self-Signed option to require our edge servers to respond with a 502 Bad Gateway response upon detecting a self-signed certificate from the origin server during the TLS handshake.

Public Key Pinning

Register the SHA-1 digest for the public key of your end-entity (i.e., leaf) certificate. After which, our edge servers will respond with a 502 Bad Gateway response when the SHA-1 digest for the public key detected from the origin server does not match one of the values defined within this option.

To pin your public key(s)

  1. Generate a SHA-1 digest of your public key by running the following command on the server hosting your end-entity certificate(s):

    openssl x509 -in <public_cert>.crt -pubkey -noout | openssl rsa -pubin -outform DER | openssl dgst -sha1
  2. Modify the desired customer origin group.
  3. From the Group Settings section, click Origin TLS.
  4. From the Subject Public Key Info Pin Verification option, paste the SHA-1 digest generated in step 1.
  5. Register an additional public key by clicking + Pin and then repeating the above steps.
  6. Click Save to save your changes to your customer origin group.

Origin Shield

Origin Shield establishes an additional buffer between a customer origin server and your clients. This buffer is useful for protecting a customer origin server from:

  • Denial of service attacks
  • Spikes in traffic

The Origin Shield feature is only available after it has been activated on the HTTP Large and/or the HTTP Small platform. This feature is unavailable for all other platforms. For more information, please contact your CDN account manager.

How Does It Work?

The Origin Shield feature reduces the number of requests that are sent to the customer origin server. This results in reduced server and network load on the customer origin server. It is able to accomplish this by establishing an intermediate caching layer between the customer origin server and an edge server. This intermediate caching layer is illustrated below.

This intermediate caching layer augments the default CDN caching behavior in the following ways:

Origin Shield Configuration

Protecting a customer origin through the use of the Origin Shield feature requires the selection of a single (recommended) or multiple origin shield locations. Each configuration method is described below.

Each origin shield server is identified by the city and the three-letter abbreviation for the POP where it is located. The use of a POP abbreviation allows you to distinguish between multiple origin shield servers in the same city.

Single Origin Shield Location (Recommended)

The recommended configuration for reducing requests to your customer origin server is to define a single origin shield location. This can be achieved by enabling the Single POP option and then selecting the location closest to the server(s) associated with the customer origin.

The workflow through which a request for content protected by Origin Shield is handled by our CDN service is illustrated below.

The above workflow is described below.

  1. If an edge server does not have a cached version of the requested content, then it will forward the request to the specified origin shield location.

    If a cached version is found, the requested content will be served immediately to the client.

  2. If an origin shield server does not have the requested asset, it will forward the request to your web servers.

    If a cached version is found, then the origin shield server will serve it immediately to the client via the edge server from the previous step. Proceed to step 5.

  3. Once the origin shield server receives the requested content from your web servers, it will forward it to the edge server that requested it. After which, it may cache the requested content.
  4. The edge server will deliver the content to the client that requested it.
  5. The edge server may then cache it.

Multiple Origin Shield Locations

An alternative configuration is to define several origin shield locations for a single customer origin group. This can be accomplished by enabling the Multiple POPs option and then choosing how requests are handled for each of the following regions: North America: West Coast United States, North America: East Coast United States, Europe, and Asia/South America. Choose one of the following options for each region:

  • Blank: Leaving a region blank indicates that requests for this region will skip the origin shield server in the selected POP and attempt to retrieve it from the next closest origin shield server.
  • POP: Selecting a POP activates origin shield for that region. Requests for this region will go through the selected origin shield. If the origin shield does not have the requested asset, it will request it from the customer origin server.
  • Bypass: Selecting to bypass a region indicates that requests for this region will bypass the origin shield and go directly to the customer origin server. This type of configuration is the equivalent of turning origin shield off for a particular region.

Origin shield locations in Asia and South America require the activation of the Premium Asia and Latin America geographic delivery regions, respectively.

ADN Gateway

ADN Gateway servers are responsible for ensuring efficient data delivery between the CDN and external web servers (i.e., customer origin servers). Before anADN Gateway server can assume this responsibility, our network will need to calculate the top three ADN Gateway servers that can provide the best performance for a customer origin group. This is calculated by requesting an asset from each web server associated with a customer origin group.

Configuration:

Key information:

Sample Validation Path

The following sample scenario provides an example as to how the Validation Path option should be configured.

Item Scenario Description

HTTP Host Header

A customer origin server's HTTP Host header is set to:

adn.mydomain.com

Sample Asset Configuration

A sample asset has been uploaded to all of the web servers associated with your customer origin group. The relative path to this sample asset is:

/Permanent/Background.png

Validation Path

The Validation Path option should be set to:

http://adn.mydomain.com/permanent/background.png

Key Response Headers

Configure your web server(s) or Rules Engine to include key headers in the response sent to the client to ensure optimal CDN performance.

Cache Revalidation

The web server(s) associated with a customer origin should include one of the following headers with each response that should be cached:

The above response headers allow our edge servers to perform cache revalidationRefers to the process that occurs when a request for stale content requires that our edge servers check for a new version of the requested content on the origin server. when staleIdentifies cached content whose TTL has expired. Our edge servers revalidate stale content with the origin server. This step ensures that the latest version of the requested content is served to the requester. content is requested. If both response headers are missing, then our edge servers will perform an unconditional GET request to the customer origin server.

QUIC

Allow QUIC-compatible clients to leverage QUIC by including the alt-svc header in the response.

The easiest way to enable QUIC for all CDN traffic is to add the QUIC feature under the Always match condition.

Learn more.

Setting Response Headers

Response headers may be defined via either of the following methods:

QUIC

QUIC is an emerging technology that is developing at a rapid pace. For this reason, we are occasionally required to temporarily suspend this capability when upgrading our QUIC implementation. Although QUIC will be disabled during these maintenance windows, your traffic will continue to be accelerated via our network.

QUIC, which stands for Quick UDP Internet Connections, is a new transport protocol developed by Google that reduces latency and provides better HTTP/2 stream-multiplexing support when compared to TCP. QUIC provides functionality equivalent to TCP + TLS + HTTP/2 except that it is implemented over UDP.

QUIC is only supported for client-to-edge traffic to customer origins that have been configured to support TLS.
Learn more (Certificate Provisioning System).
Learn more (Legacy HTTPS).

According to The Chromium Projects, it provides the following advantages over TCP + TLS + HTTP/2:

QUIC Setup

Enable QUIC by setting up a rule within Rules Engine that either:

ClosedWhy should I only enable QUIC for all traffic or by edge CNAME?

Once a QUIC-compatible user agent discovers that a server supports QUIC, it will attempt to leverage QUIC for all subsequent requests to the same domain until the connection ends. This means that enabling QUIC using more granular control will not accomplish the desired result.

Sample Scenario

In this sample scenario, you have created a rule that only enables QUIC for HTML files.

Let's assume that a QUIC-compatible user agent requests the following file:

https://cdn.example.com/marketing/index.html

This user agent will leverage QUIC for all subsequent requests to cdn.example.com until the connection ends. This will occur regardless of whether HTML files are being requested.

To enable QUIC via Rules Engine

Before enabling QUIC, please verify that the desired customer origin(s) support TLS.

  1. Create a draft.

  2. Update a rule within that draft to define the condition(s) under which QUIC should be enabled.

  3. Perform one of the following steps:

    • Automatic Setup (Recommended): Add the QUIC feature directly below the match condition(s) created in step 2. Set it to Yes.

    • Manual Setup: Add the Modify Client Response Header feature directly below the match condition(s) created in step 2.

      Set it to the following value:

      quic=":443"; ma=2592000; v="%{quic_versions}",%{quic_altsvc_versions}
  4. Click Save to save the rule.
  5. Click Lock Draft as Policy to convert the draft into a policy.
  6. Deploy the policy to the Production or Staging environment.

How Does QUIC Work?

Before a user agentRefers to software that acts on behalf of a user. For example, a web browser (e.g., FireFox, Chrome, and Internet Explorer) is a user agent. A web browser will make HTTP/HTTPS requests based on user actions (e.g., requesting a web site or clicking a link). may leverage QUIC, it must be informed via the alt-svc (Alternative Service) response header that it may communicate with the CDN via QUIC. This response header also informs the client the set of supported QUIC versions and the length of time that this data should be cached by the client.

By default, QUIC is supported on the latest versions of Google Chrome, Chromium, and Opera. However, it may require enablement. If a user agent doesn't support QUIC, then it will communicate with the CDN using HTTP/2 over TCP.

Our QUIC implementation supports the Bottleneck Bandwidth and Round-trip propagation time (BBR) congestion control algorithm without requiring additional CDN setup. However, BBR will only be used when a QUIC-enabled client (e.g., Google Chrome) requests it.

Sample alt-svc header name/value:

alt-svc: quic=":443"; ma=2592000; v="49,48,46,43",h3-Q049=":443"; ma=2592000,h3-Q048=":443"; ma=2592000,h3-Q046=":443"; ma=2592000,h3-Q043=":443"; ma=2592000

The above sample response header indicates to the client that:

WebSocket Protocol

The WebSocket protocol is a TCP-based protocol that enables an open, continuous, and full-duplex connection between a client (e.g., web browser) and a server. This allows a server to send data to a client without requiring the client to request it. In contrast, the HTTP protocol only allows a server to respond when a client requests content.

Leverage the WebSocket protocol for low latency applications that require real-time, interactive communication. For example, WebSocket communication improves the user experience for multiplayer games and chat applications by providing higher responsiveness via lower latency.

Key benefits:

WebSocket Setup

Set up involves:

  1. WebSocket Protocol Activation

    You must activate support for the WebSocket protocol before you can leverage it.

    Contact your CDN account manager to request support for the WebSocket protocol. We will determine whether to activate it based on an analysis of your use case.

  2. Web Server Setup

    Verify that your web server supports the use of the WebSocket protocol. After which, you should implement a WebSocket server-side application and verify that it allows bidirectional communication with your client.

    Our network acts as a middleman between your client and server for bidirectional communication via the WebSocket protocol. If your web server does not properly support WebSocket communication, then a client's request to upgrade to the WebSocket protocol will fail and communication will proceed via HTTP.

  3. Client Setup

    1. Verify that the client natively supports the WebSocket protocol. Most modern browsers already support the WebSocket protocol. If your custom application does not already support it, then you will need to add support for it before proceeding to the next step.

    2. Update your client to include the following request headers when initially requesting a connection that requires low latency:

      Connection: Upgrade

      Upgrade: websocket

      These headers indicate that the client wants to change the protocol from HTTP to WebSocket.

    3. Update your client to re-initiate the connection with the server upon disconnection.

      A client, server, or a network disruption may disconnect a WebSocket connection.

How Does the WebSocket Protocol Work?

A WebSocket connection starts when a client submits a request to the CDN over HTTP. This request must inform an edge serverThis type of server is located near the edge of our network where its close proximity to your end-users allows it to deliver data more quickly than normal Internet communication. It is responsible for handling requests and caching assets. that the client would like to upgrade the connection to use the WebSocket protocol via the following request headers:

Connection: Upgrade

Upgrade: websocket

If the WebSocket protocol has been activated on your account, an edge server will forward the request to your web server(s). Your web server must then either:

Firewall Access

For the purpose of fulfilling requests, our edge servers require access to all servers associated with a customer origin group. Please ensure that your firewall allows access to all of the IP blocks listed in the Whitelist IP Blocks page.

Export a list of the IPv4 and IPv6 blocks that should be whitelisted by clicking CSV from the Whitelist IP Blocks page.

The Whitelist IP Blocks page contains a superset of IP addresses that includes:

Once the IP blocks defined within the Whitelist IP Blocks page have been whitelisted on your firewall, it is unnecessary to add the IP blocks defined within the CDN IP's page.

Best Practices (Dynamic Application)

If your customer origin hosts a dynamic application, then it is highly recommended that you do not use a user's IP address to maintain a session instance. This type of configuration is unsupported, since the client does not connect directly to your web servers. Instead, the client connects to a server on our network, and then that server connects to your web server. If you would like to maintain a session for your dynamic application, we recommend that you use a cookie to identify the session. For example, a cookie could keep track of a unique ID for each client's session.

More Information