|
| 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 selector to specify egress peers using |
| 10 | +Fully Qualified Domain Names (FQDNs). |
| 11 | + |
| 12 | +## Goals |
| 13 | + |
| 14 | +* Provide a selector to specify egress peers using a Fully Qualified Domain Name |
| 15 | + (for example `kubernetes.io`). |
| 16 | +* Support a restricted set of regex matching capabilities when specifying FQDNs. |
| 17 | + |
| 18 | +## Non-Goals |
| 19 | + |
| 20 | +* This enhancement does not include a FQDN selector for allowing ingress |
| 21 | + traffic. |
| 22 | +* This enhancement does not include any L7 matching or filtering capabilities, |
| 23 | + like matching HTTP traffic or URL paths. |
| 24 | + * This selector should not control what DNS records are resolvable from a |
| 25 | + particular workload. |
| 26 | +* This enhancement does not provide a mechanism for selecting in-cluster |
| 27 | + endpoints using FQDNs. This is explicitly disallowed by the spec. |
| 28 | + * To select Pods, Nodes, API Server, AdminNetworkPolicy has more first party |
| 29 | + selector with better UX. |
| 30 | +* This enhancement does not specify the details of how traffic is routed to the |
| 31 | + specified destination. For example, it does not prescribe details around NAT |
| 32 | + or egress gateways. |
| 33 | +* This enhancement does not require a particular mechanism for securing DNS |
| 34 | + resolution (e.g. DNSSEC or DNS-over-TLS). |
| 35 | + * TODO: Is this up to implementations? Maybe an optional enum describing |
| 36 | + what security mechanims to use and providers can chose whther to support |
| 37 | + or not. |
| 38 | + |
| 39 | +## Introduction |
| 40 | + |
| 41 | +FQDN-based egress controls are a common enterprise security practice. |
| 42 | +Administrators often prefer to write security policies using DNS names such as |
| 43 | +“www.kubernetes.io” instead of capturing all the IP addresses the DNS name might |
| 44 | +resolve to. Keeping up with changing IP addresses is a maintenance burden, and |
| 45 | +hampers the readability of the network policies. |
| 46 | + |
| 47 | +## User Stories |
| 48 | + |
| 49 | +* As a cluster admin, I want to allow certain workloads to send traffic to a |
| 50 | + service specified by a well-known domain name (e.g. on-prem logging service) |
| 51 | +* As a cluster admin, I want to allow the cluster to communitcate with services |
| 52 | + provided by a Cloud provider by allowing their whole domain. |
| 53 | + |
| 54 | +## API |
| 55 | + |
| 56 | +TODO |
| 57 | + |
| 58 | +## Alternatives |
| 59 | + |
| 60 | +### IP Block Selector |
| 61 | + |
| 62 | +IP blocks are an important tool for specifying Network Policies. However, they |
| 63 | +do not address all user needs and have a few short-comings when compared to FQDN |
| 64 | +selectors: |
| 65 | + |
| 66 | +* IP-based selectors can become verbose if a single logical service has numerous |
| 67 | + IPs backing it. |
| 68 | +* IP-based selectors pose an ongoing maintanance burden for administrators, who |
| 69 | + need to be aware of changing IPs. |
| 70 | +* IP-based selectors can result in policies that are difficult to read and |
| 71 | + audit. |
| 72 | + |
| 73 | +### L7 Policy |
| 74 | + |
| 75 | +Another alternative is to provide a true L7 selector, similar to the policies |
| 76 | +provided by Service Mesh providers. While L7 selectors can offer more |
| 77 | +expressibility, they often come trade-offs that are not suitable for all users: |
| 78 | + |
| 79 | +* L7 selectors necessarily support a select set of protocols. Customers may be |
| 80 | + using a custom protocol for application-level communication, but still want |
| 81 | + the ability to specify endpoints using DNS. |
| 82 | +* L7 selectors often require proxies to perform deep packet inspection and |
| 83 | + enforce the policies. These proxies can introduce un-desireable latencies in |
| 84 | + the datapath of applications. |
| 85 | + |
| 86 | +## References |
| 87 | + |
| 88 | +* [NPEP #126](https://github.com/kubernetes-sigs/network-policy-api/issues/126): |
| 89 | + Egress Control in ANP |
0 commit comments