|
| 1 | +# NPEP-133: FQDN Selector for Egress Traffic |
| 2 | + |
| 3 | +* Issue: |
| 4 | + [#133](https://github.com/kubernetes-sigs/network-policy-api/issues/133) |
| 5 | +* Status: Provisional |
| 6 | + |
| 7 | +## TLDR |
| 8 | + |
| 9 | +This enhancement proposes adding a new optional selector to specify egress peers |
| 10 | +using [Fully Qualified Domain Names](https://www.wikipedia.org/wiki/Fully_qualified_domain_name) |
| 11 | +(FQDNs). |
| 12 | + |
| 13 | +## Goals |
| 14 | + |
| 15 | +* Provide a selector to specify egress peers using a Fully Qualified Domain Name |
| 16 | + (for example `kubernetes.io`). |
| 17 | +* Support a restricted set of regex matching capabilities when specifying FQDNs. |
| 18 | +* Currently only `ALLOW` type rules are proposed. |
| 19 | + * Correctly enforcing DENY rules based on FQDN selectors is difficult as there |
| 20 | + is no guarantee a Network Policy plugin is aware of all IPs backing a FQDN |
| 21 | + policy. |
| 22 | +* Currently only AdminNetworkPolicy is the intended scope for this proposal. |
| 23 | + * Since Kubernetes NetworkPolicy does not have a FQDN selector, adding this |
| 24 | + capability to BaselineAdminNetworkPolicy can result in unintended behavior. |
| 25 | + For example, if BANP allows traffic to `example.io`, but the namespace admin |
| 26 | + installs a Kubernetes Network Policy, the namespace admin has no way to |
| 27 | + replicate the `example.io` selector using just Kubernetes Network Policies. |
| 28 | + |
| 29 | +## Non-Goals |
| 30 | + |
| 31 | +* This enhancement does not include a FQDN selector for allowing ingress |
| 32 | + traffic. |
| 33 | +* This enhancement does not include any L7 matching or filtering capabilities, |
| 34 | + like matching HTTP traffic or URL paths. |
| 35 | + * This selector should not control what DNS records are resolvable from a |
| 36 | + particular workload. |
| 37 | +* This enhancement does not provide a mechanism for selecting in-cluster |
| 38 | + endpoints using FQDNs. This is explicitly disallowed by the spec. |
| 39 | + * To select Pods, Nodes, or the API Server, AdminNetworkPolicy has other more |
| 40 | + specific selectors. |
| 41 | +* This enhancement does not specify the details of how traffic is routed to the |
| 42 | + specified destination. For example, it does not prescribe details around NAT |
| 43 | + or egress gateways. |
| 44 | +* This enhancement does not require any mechanism for securing DNS resolution |
| 45 | + (e.g. DNSSEC or DNS-over-TLS). Unsecured DNS requests are expected to be |
| 46 | + sufficient for looking up FQDNs. |
| 47 | + |
| 48 | +## Introduction |
| 49 | + |
| 50 | +FQDN-based egress controls are a common enterprise security practice. |
| 51 | +Administrators often prefer to write security policies using DNS names such as |
| 52 | +“www.kubernetes.io” instead of capturing all the IP addresses the DNS name might |
| 53 | +resolve to. Keeping up with changing IP addresses is a maintenance burden, and |
| 54 | +hampers the readability of the network policies. |
| 55 | + |
| 56 | +## User Stories |
| 57 | + |
| 58 | +* As a cluster admin, I want to allow all Pods in the cluster to send traffic to |
| 59 | + an external service specified by a well-known domain name. For example, all |
| 60 | + Pods must be able to talk to `my-service.com`. |
| 61 | + |
| 62 | +* As a cluster admin, I want to allow Pods in the "monitoring" namespace to be |
| 63 | + able to send traffic to a logs-sink, hosted at `logs-storage.com` |
| 64 | + |
| 65 | +* As a cluster admin, I want to allow all Pods in the cluster to send traffic to |
| 66 | + any of the managed services provided by my Cloud Provider. Since the cloud |
| 67 | + provider has a well known parent domain, I want to allow Pods to send traffic |
| 68 | + to all sub-domains using a wild-card selector -- `*.my-cloud-provider.com` |
| 69 | + |
| 70 | +### Future User Stories |
| 71 | + |
| 72 | +These are some user stories we want to keep in mind, but due to limitations of |
| 73 | +the existing Network Policy API, cannot be implemented currently. The design |
| 74 | +goal in this case is to ensure we do not make these unimplementable down the line. |
| 75 | + |
| 76 | +* As a cluster admin, I want to block all cluster egress traffic by default, and |
| 77 | + require namespace admins to create NetworkPolicies explicitly allowing egress |
| 78 | + to the domains they need to talk to. |
| 79 | + |
| 80 | + The Cluster admin would use a `BaselineAdminNetworkPolicy` object to switch |
| 81 | + the default disposition of the cluster. Namespace admins would then use |
| 82 | + a FQDN selector in the Kubernetes `NetworkPolicy` objects to allow |
| 83 | + `my-service.com`. |
| 84 | + |
| 85 | +## API |
| 86 | + |
| 87 | +TODO |
| 88 | + |
| 89 | +## Alternatives |
| 90 | + |
| 91 | +### IP Block Selector |
| 92 | + |
| 93 | +IP blocks are an important tool for specifying Network Policies. However, they |
| 94 | +do not address all user needs and have a few short-comings when compared to FQDN |
| 95 | +selectors: |
| 96 | + |
| 97 | +* IP-based selectors can become verbose if a single logical service has numerous |
| 98 | + IPs backing it. |
| 99 | +* IP-based selectors pose an ongoing maintanance burden for administrators, who |
| 100 | + need to be aware of changing IPs. |
| 101 | +* IP-based selectors can result in policies that are difficult to read and |
| 102 | + audit. |
| 103 | + |
| 104 | +### L7 Policy |
| 105 | + |
| 106 | +Another alternative is to provide a true L7 selector, similar to the policies |
| 107 | +provided by Service Mesh providers. While L7 selectors can offer more |
| 108 | +expressibility, they often come trade-offs that are not suitable for all users: |
| 109 | + |
| 110 | +* L7 selectors necessarily support a select set of protocols. Customers may be |
| 111 | + using a custom protocol for application-level communication, but still want |
| 112 | + the ability to specify endpoints using DNS. |
| 113 | +* L7 selectors often require proxies to perform deep packet inspection and |
| 114 | + enforce the policies. These proxies can introduce un-desireable latencies in |
| 115 | + the datapath of applications. |
| 116 | + |
| 117 | +## References |
| 118 | + |
| 119 | +* [NPEP #126](https://github.com/kubernetes-sigs/network-policy-api/issues/126): |
| 120 | + Egress Control in ANP |
| 121 | + |
| 122 | +### Implementations |
| 123 | + |
| 124 | +* [Antrea](https://antrea.io/docs/main/docs/antrea-network-policy/#fqdn-based-filtering) |
| 125 | +* [Calico](https://docs.tigera.io/calico-enterprise/latest/network-policy/domain-based-policy) |
| 126 | +* [Cilium](https://docs.cilium.io/en/latest/security/policy/language/#dns-based) |
| 127 | +* [Open Shift](https://docs.openshift.com/container-platform/latest/networking/openshift_sdn/configuring-egress-firewall.html) |
| 128 | + |
| 129 | +The following is a best-effort breakdown of capabilities of different |
| 130 | +NetworkPolicy providers, as of 2023-09-25. This information may be out-of-date, |
| 131 | +or inaccurate. |
| 132 | + |
| 133 | +| | Antrea | Calico | Cilium | OpenShift <br/> (current) | OpenShift <br/> (future) | |
| 134 | +| -------------- | ------------------------------ | ------------ | ------------ | ------------------------- | ------------------------ | |
| 135 | +| Implementation | DNS Snooping <br/> + Async DNS | DNS Snooping | DNS Snooping | Async DNS | DNS Snooping | |
| 136 | +| Wildcards | ✔ | ️✔ | ✔ | ❌ | ✔ | |
| 137 | +| Egress Rules | ✔ | ️✔ | ✔ | ✔ | ✔ | |
| 138 | +| Ingress Rules | ❌ | ️❌ | ❌ | ❌ | ❌ | |
| 139 | +| Allow Rules | ✔ | ️✔ | ✔ | ✔ | ✔ | |
| 140 | +| Deny Rules | ✔ | ️❌(?) | ❌ | ✔ | ❌(?) | |
0 commit comments