]> Team Digitale, Italian Government

Italy robipolli@gmail.com

Red Hat

amr@redhat.com

Applications and Real-Time HTTPAPI Internet-Draft This document defines the RateLimit-Limit, RateLimit-Remaining, RateLimit-Reset fields for HTTP, thus allowing servers to publish current service limits and clients to shape their request policy and avoid being throttled out. Note to Readers RFC EDITOR: please remove this section before publication Discussion of this draft takes place on the HTTP working group mailing list (httpapi@ietf.org), which is archived at https://mailarchive.ietf.org/arch/browse/httpapi/. The source code and issues list for this draft can be found at https://github.com/ietf-wg-httpapi/ratelimit-headers. References to ThisRFC in the IANA Considerations section would be replaced with the RFC number when assigned.

Introduction The widespreading of HTTP as a distributed computation protocol requires an explicit way of communicating service status and usage quotas. This was partially addressed by the Retry-After header field defined in to be returned in 429 Too Many Requests (see ) or 503 Service Unavailable responses. Widely deployed quota mechanisms limit the number of acceptable requests in a given time window, e.g. 10 requests per second; currently, there is no standard way to communicate service quotas so that the client can throttle its requests and prevent 4xx or 5xx responses. See for further information on the current usage of rate limiting in HTTP. This document defines syntax and semantics for the following fields:

• RateLimit-Limit: containing the requests quota in the time window;
• RateLimit-Remaining: containing the remaining requests quota in the current window;
• RateLimit-Reset: containing the time remaining in the current window, specified in seconds.

The behavior of RateLimit-Reset is compatible with the delay-seconds notation of Retry-After. The fields definition allows to describe complex policies, including the ones using multiple and variable time windows and dynamic quotas, or implementing concurrency limits.

Goals The goals of the RateLimit fields are:

Interoperability:

Standardization of the names and semantics of rate-limit headers to ease their enforcement and adoption;

Resiliency:

Improve resiliency of HTTP infrastructure by providing clients with information useful to throttle their requests and prevent 4xx or 5xx responses;

Documentation:

Simplify API documentation by eliminating the need to include detailed quota limits and related fields in API documentation.

The following features are out of the scope of this document:

Authorization:

RateLimit fields are not meant to support authorization or other kinds of access controls.

Throttling scope:

This specification does not cover the throttling scope, that may be the given resource-target, its parent path or the whole Origin (see Section 7 of ). This can be addressed using extensibility mechanisms such as the parameter registry .

Response status code:

RateLimit fields may be returned in both successful (see Section 15.3 of ) and non-successful responses. This specification does not cover whether non Successful responses count on quota usage, nor it mandates any correlation between the RateLimit values and the returned status code.

Throttling policy:

This specification does not mandate a specific throttling policy. The values published in the fields, including the window size, can be statically or dynamically evaluated.

Service Level Agreement:

Conveyed quota hints do not imply any service guarantee. Server is free to throttle respectful clients under certain circumstances.

Notational Conventions The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in BCP 14 when, and only when, they appear in all capitals, as shown here. This document uses the Augmented BNF defined in and updated by along with the "#rule" extension defined in Section 5.6.1 of . The term Origin is to be interpreted as described in Section 7 of . This specification uses Structured Fields to specify syntax. The terms sf-list, sf-item, sf-string, sf-token, sf-integer, bare-item and key refer to the structured types defined therein.

Expressing rate-limit policies

Time window Rate limit policies limit the number of acceptable requests in a given time window. A time window is expressed in seconds, using the following syntax: Where delay-seconds is a non-negative sf-integer compatible with the "delay-seconds" rule defined in Section 10.2.3 of . Subsecond precision is not supported.

Service limit The service-limit is a value associated to the maximum number of requests that the server is willing to accept from one or more clients on a given basis (originating IP, authenticated user, geographical, ..) during a time-window as defined in . The service-limit is expressed in quota-units and has the following syntax: where quota-units is a non-negative sf-integer. The service-limit SHOULD match the maximum number of acceptable requests. The service-limit MAY differ from the total number of acceptable requests when weight mechanisms, bursts, or other server policies are implemented. If the service-limit does not match the maximum number of acceptable requests the relation with that SHOULD be communicated out-of-band. Example: A server could

• count once requests like /books/{id}
• count twice search requests like /books?author=WuMing

so that we have the following counters

Quota policy This specification allows describing a quota policy with the following syntax: where the associated bare-item is a service-limit and parameters are supported. The following parameters are defined:

w:

The REQUIRED "w" parameter specifies a time window. Its syntax is a "time-window" defined in .

Other parameters are allowed and can be regarded as comments. They ought to be registered within the "Hypertext Transfer Protocol (HTTP) RateLimit Parameters Registry", as described in . An example policy of 100 quota-units per minute. The definition of a quota-policy does not imply any specific distribution of quota-units over time. Such service specific details can be conveyed as parameters. Two policy examples containing further details via custom parameters To avoid clashes, implementers SHOULD prefix unregistered parameters with an x-<vendor> identifier, e.g. x-acme-policy, x-acme-burst. While it is useful to define a clear syntax and semantics even for custom parameters, it is important to note that user agents are not required to process quota policy information.

Providing RateLimit fields A server uses the RateLimit response fields defined in this document to communicate its quota policies according to the following rules:

• RateLimit-Limit and RateLimit-Reset are REQUIRED;
• RateLimit-Remaining is RECOMMENDED.

The returned values refers to the metrics used to evaluate if the current request respects the quota policy and MAY not apply to subsequent requests. Example: a successful response with the following fields does not guarantee that the next request will be successful. Server metrics may be subject to other conditions like the one shown in the example from . A server MAY return RateLimit response fields independently of the response status code. This includes throttled responses. This document does not mandate any correlation between the RateLimit values and the returned status code. Servers should be careful in returning RateLimit fields in redirection responses (e.g. 3xx status codes) because a low RateLimit-Remaining value could prevent the client from issuing requests. For example, given the rate limiting fields below, a client could decide to wait 10 seconds before following the Location header, because RateLimit-Remaining is 0. If a response contains both the Retry-After and the RateLimit-Reset fields, the value of RateLimit-Reset SHOULD reference the same point in time as Retry-After. When using a policy involving more than one time-window, the server MUST reply with the RateLimit fields related to the window with the lower RateLimit-Remaining values. A service returning RateLimit fields MUST NOT convey values exposing an unwanted volume of requests and SHOULD implement mechanisms to cap the ratio between RateLimit-Remaining and RateLimit-Reset (see ); this is especially important when quota-policies use a large time-window. Under certain conditions, a server MAY artificially lower RateLimit field values between subsequent requests, e.g. to respond to Denial of Service attacks or in case of resource saturation. Servers usually establish whether the request is in-quota before creating a response, so the RateLimit field values should be already available in that moment. Nonetheless servers MAY decide to send the RateLimit fields in a trailer section. To ease the migration from existing rate limit headers, a server SHOULD be able to provide the RateLimit-Limit field even without the optional quota-policy section.

Performance considerations Servers are not required to return RateLimit fields in every response, and clients need to take this into account. For example, an implementer concerned with performance might provide RateLimit fields only when a given quota is going to expire. Implementers concerned with response fields' size, might take into account their ratio with respect to the payload data, or use header-compression http features such as .

Receiving RateLimit fields A client MUST process the received RateLimit fields. A client MUST validate the values received in the RateLimit fields before using them and check if there are significant discrepancies with the expected ones. This includes a RateLimit-Reset moment too far in the future or a service-limit too high. A client receiving RateLimit fields MUST NOT assume that subsequent responses contain the same RateLimit fields, or any RateLimit fields at all. Malformed RateLimit fields MAY be ignored. A client SHOULD NOT exceed the quota-units expressed in RateLimit-Remaining before the time-window expressed in RateLimit-Reset. A client MAY still probe the server if the RateLimit-Reset is considered too high. The value of RateLimit-Reset is generated at response time: a client aware of a significant network latency MAY behave accordingly and use other information (e.g. the Date response header field, or otherwise gathered metrics) to better estimate the RateLimit-Reset moment intended by the server. The quota-policy values and comments provided in RateLimit-Limit are informative and MAY be ignored. If a response contains both the RateLimit-Reset and Retry-After fields, Retry-After MUST take precedence and RateLimit-Reset MAY be ignored. This specification does not mandate a specific throttling behavior and implementers can adopt their preferred policies, including:

• slowing down or preemptively back-off their request rate when approaching quota limits;
• consuming all the quota according to the exposed limits and then wait.

Intermediaries This section documents the considerations advised in Section 16.3.2 of . An intermediary that is not part of the originating service infrastructure and is not aware of the quota-policy semantic used by the Origin Server SHOULD NOT alter the RateLimit fields' values in such a way as to communicate a more permissive quota-policy; this includes removing the RateLimit fields. An intermediary MAY alter the RateLimit fields in such a way as to communicate a more restrictive quota-policy when:

• it is aware of the quota-unit semantic used by the Origin Server;
• it implements this specification and enforces a quota-policy which is more restrictive than the one conveyed in the fields.

An intermediary SHOULD forward a request even when presuming that it might not be serviced; the service returning the RateLimit fields is the sole responsible of enforcing the communicated quota-policy, and it is always free to service incoming requests. This specification does not mandate any behavior on intermediaries respect to retries, nor requires that intermediaries have any role in respecting quota-policies. For example, it is legitimate for a proxy to retransmit a request without notifying the client, and thus consuming quota-units.

Caching As is the ordinary case for HTTP caching (), a response with RateLimit fields might be cached and re-used for subsequent requests. A cached RateLimit response does not modify quota counters but could contain stale information. Clients interested in determining the freshness of the RateLimit fields could rely on fields such as Date and on the time-window of a quota-policy.

Fields definition The following RateLimit response fields are defined

RateLimit-Limit The RateLimit-Limit response field indicates the service-limit associated to the client in the current time-window. If the client exceeds that limit, it MAY not be served. The field is a List Structured Field of positive length. The first member is named expiring-limit and its syntax is service-limit, while the syntax of the other optional members is quota-policy The expiring-limit value MUST be set to the service-limit that is closer to reach its limit. The quota-policy is defined in , and its values are informative. A time-window associated to expiring-limit can be communicated via an optional quota-policy value, like shown in the following example If the expiring-limit is not associated to a time-window, the time-window MUST either be:

• inferred by the value of RateLimit-Reset at the moment of the reset, or
• communicated out-of-band (e.g. in the documentation).

Policies using multiple quota limits MAY be returned using multiple quota-policy items, like shown in the following two examples: This field MUST NOT occur multiple times and can be sent in a trailer section.

RateLimit-Remaining The RateLimit-Remaining response field indicates the remaining quota-units defined in associated to the client. The field is an Integer Structured Field and its value is This field MUST NOT occur multiple times and can be sent in a trailer section. Clients MUST NOT assume that a positive RateLimit-Remaining value is a guarantee that further requests will be served. A low RateLimit-Remaining value is like a yellow traffic-light for either the number of requests issued in the time-window or the request throughput: the red light may arrive suddenly (see ). One example of RateLimit-Remaining use is below.

RateLimit-Reset The RateLimit-Reset response field indicates either

• the number of seconds until the quota resets.

The field is an Integer Structured Field and its value is The delay-seconds format is used because:

• it does not rely on clock synchronization and is resilient to clock adjustment and clock skew between client and server (see Section 5.6.7 of );
• it mitigates the risk related to thundering herd when too many clients are serviced with the same timestamp.

This field MUST NOT occur multiple times and can be sent in a trailer section. An example of RateLimit-Reset use is below. The client MUST NOT assume that all its service-limit will be restored after the moment referenced by RateLimit-Reset. The server MAY arbitrarily alter the RateLimit-Reset value between subsequent requests e.g. in case of resource saturation or to implement sliding window policies.

Security Considerations

Throttling does not prevent clients from issuing requests This specification does not prevent clients to make over-quota requests. Servers should always implement mechanisms to prevent resource exhaustion.

Information disclosure Servers should not disclose to untrusted parties operational capacity information that can be used to saturate its infrastructural resources. While this specification does not mandate whether non 2xx responses consume quota, if 401 and 403 responses count on quota a malicious client could probe the endpoint to get traffic information of another user. As intermediaries might retransmit requests and consume quota-units without prior knowledge of the User Agent, RateLimit fields might reveal the existence of an intermediary to the User Agent.

Remaining quota-units are not granted requests RateLimit-* fields convey hints from the server to the clients in order to avoid being throttled out. Clients MUST NOT consider the quota-units returned in RateLimit-Remaining as a service level agreement. In case of resource saturation, the server MAY artificially lower the returned values or not serve the request regardless of the advertised quotas.

Reliability of RateLimit-Reset Consider that service-limit may not be restored after the moment referenced by RateLimit-Reset, and the RateLimit-Reset value should not be considered fixed nor constant. Subsequent requests may return a higher RateLimit-Reset value to limit concurrency or implement dynamic or adaptive throttling policies.

Resource exhaustion When returning RateLimit-Reset you must be aware that many throttled clients may come back at the very moment specified. This is true for Retry-After too. For example, if the quota resets every day at 18:00:00 and your server returns the RateLimit-Reset accordingly there's a high probability that all clients will show up at 18:00:00. This could be mitigated by adding some jitter to the field-value. Resource exhaustion issues can be associated with quota policies using a large time-window, because a user agent by chance or on purpose might consume most of its quota-units in a significantly shorter interval. This behavior can be even triggered by the provided RateLimit fields. The following example describes a service with an unconsumed quota-policy of 10000 quota-units per 1000 seconds. A client implementing a simple ratio between RateLimit-Remaining and RateLimit-Reset could infer an average throughput of 1000 quota-units per second, while RateLimit-Limit conveys a quota-policy with an average of 10 quota-units per second. If the service cannot handle such load, it should return either a lower RateLimit-Remaining value or an higher RateLimit-Reset value. Moreover, complementing large time-window quota-policies with a short time-window one mitigates those risks.

Denial of Service RateLimit fields may assume unexpected values by chance or purpose. For example, an excessively high RateLimit-Remaining value may be:

• used by a malicious intermediary to trigger a Denial of Service attack or consume client resources boosting its requests;
• passed by a misconfigured server;

or an high RateLimit-Reset value could inhibit clients to contact the server. Clients MUST validate the received values to mitigate those risks.

IANA Considerations IANA is requested to update one registry and create one new registry. Please add the following entries to the "Hypertext Transfer Protocol (HTTP) Field Name Registry" registry ():

Field Name	Status	Specification
RateLimit-Limit	permanent	of ThisRFC
RateLimit-Remaining	permanent	of ThisRFC
RateLimit-Reset	permanent	of ThisRFC

RateLimit Parameters Registration IANA is requested to create a new registry to be called "Hypertext Transfer Protocol (HTTP) RateLimit Parameters Registry", to be located at https://www.iana.org/assignments/http-ratelimit-parameters. Registration is done on the advice of a Designated Expert, appointed by the IESG or their delegate. All entries are Specification Required (). Registration requests consist of the following information:

• Parameter name: The parameter name, conforming to .
• Field name: The RateLimit field for which the parameter is registered. If a parameter is intended to be used with multiple fields, it has to be registered for each one.
• Description: A brief description of the parameter.
• Specification document: A reference to the document that specifies the parameter, preferably including a URI that can be used to retrieve a copy of the document.
• Comments (optional): Any additional information that can be useful.

The initial contents of this registry should be:

Field Name	Parameter name	Description	Specification	Comments (optional)
RateLimit-Limit	w	Time window	of ThisRFC

References Normative References Many protocols make use of points of extensibility that use constants to identify various protocol parameters. To ensure that the values in these fields do not have conflicting uses and to promote interoperability, their allocations are often coordinated by a central record keeper. For IETF protocols, that role is filled by the Internet Assigned Numbers Authority (IANA). To make assignments in a given registry prudently, guidance describing the conditions under which new values should be assigned, as well as when and how modifications to existing values can be made, is needed. This document defines a framework for the documentation of these guidelines by specification authors, in order to assure that the provided guidance for the IANA Considerations is clear and addresses the various issues that are likely in the operation of a registry. This is the third edition of this document; it obsoletes RFC 5226. Adobe Fastly greenbytes GmbH The Hypertext Transfer Protocol (HTTP) is a stateless application- level protocol for distributed, collaborative, hypertext information systems. This document describes the overall architecture of HTTP, establishes common terminology, and defines aspects of the protocol that are shared by all versions. In this definition are core protocol elements, extensibility mechanisms, and the "http" and "https" Uniform Resource Identifier (URI) schemes. This document updates RFC 3864 and obsoletes RFC 2818, RFC 7231, RFC 7232, RFC 7233, RFC 7235, RFC 7538, RFC 7615, RFC 7694, and portions of RFC 7230. This document defines the concept of an "origin", which is often used as the scope of authority or privilege by user agents. Typically, user agents isolate content retrieved from different origins to prevent malicious web site operators from interfering with the operation of benign web sites. In addition to outlining the principles that underlie the concept of origin, this document details how to determine the origin of a URI and how to serialize an origin into a string. It also defines an HTTP header field, named "Origin", that indicates which origins are associated with an HTTP request. [STANDARDS-TRACK] In many standards track documents several words are used to signify the requirements in the specification. These words are often capitalized. This document defines these words as they should be interpreted in IETF documents. This document specifies an Internet Best Current Practices for the Internet Community, and requests discussion and suggestions for improvements. RFC 2119 specifies common key words that may be used in protocol specifications. This document aims to reduce the ambiguity by clarifying that only UPPERCASE usage of the key words have the defined special meanings. Internet technical specifications often need to define a formal syntax. Over the years, a modified version of Backus-Naur Form (BNF), called Augmented BNF (ABNF), has been popular among many Internet specifications. The current specification documents ABNF. It balances compactness and simplicity with reasonable representational power. The differences between standard BNF and ABNF involve naming rules, repetition, alternatives, order-independence, and value ranges. This specification also supplies additional rule definitions and encoding for a core lexical analyzer of the type common to several Internet specifications. [STANDARDS-TRACK] This document extends the base definition of ABNF (Augmented Backus-Naur Form) to include a way to specify US-ASCII string literals that are matched in a case-sensitive manner. This document describes a set of data types and associated algorithms that are intended to make it easier and safer to define and handle HTTP header and trailer fields, known as "Structured Fields", "Structured Headers", or "Structured Trailers". It is intended for use by specifications of new HTTP fields that wish to use a common syntax that is more restrictive than traditional HTTP field values. Informative References Many applications that use the Stream Control Transmission Protocol (SCTP) want the ability to "reset" a stream. The intention of resetting a stream is to set the numbering sequence of the stream back to 'zero' with a corresponding notification to the application layer that the reset has been performed. Applications requiring this feature want it so that they can "reuse" streams for different purposes but still utilize the stream sequence number so that the application can track the message flows. Thus, without this feature, a new use of an old stream would result in message numbers greater than expected, unless there is a protocol mechanism to "reset the streams back to zero". This document also includes methods for resetting the transmission sequence numbers, adding additional streams, and resetting all stream sequence numbers. [STANDARDS-TRACK] This specification defines HPACK, a compression format for efficiently representing HTTP header fields, to be used in HTTP/2. The Hypertext Transfer Protocol (HTTP) is a stateless \%application- level protocol for distributed, collaborative, hypertext information systems. This document defines HTTP caches and the associated header fields that control cache behavior or indicate cacheable response messages. This document specifies additional HyperText Transfer Protocol (HTTP) status codes for a variety of common situations. [STANDARDS-TRACK] This document defines a date and time format for use in Internet protocols that is a profile of the ISO 8601 standard for representation of dates and times using the Gregorian calendar.

Rate-limiting and quotas Servers use quota mechanisms to avoid systems overload, to ensure an equitable distribution of computational resources or to enforce other policies - e.g. monetization. A basic quota mechanism limits the number of acceptable requests in a given time window, e.g. 10 requests per second. When quota is exceeded, servers usually do not serve the request replying instead with a 4xx HTTP status code (e.g. 429 or 403) or adopt more aggressive policies like dropping connections. Quotas may be enforced on different basis (e.g. per user, per IP, per geographic area, ..) and at different levels. For example, an user may be allowed to issue:

• 10 requests per second;
• limited to 60 requests per minute;
• limited to 1000 requests per hour.

Moreover system metrics, statistics and heuristics can be used to implement more complex policies, where the number of acceptable requests and the time window are computed dynamically. To help clients throttling their requests, servers may expose the counters used to evaluate quota policies via HTTP header fields. Those response headers may be added by HTTP intermediaries such as API gateways and reverse proxies. On the web we can find many different rate-limit headers, usually containing the number of allowed requests in a given time window, and when the window is reset. The common choice is to return three headers containing:

• the maximum number of allowed requests in the time window;
• the number of remaining requests in the current window;
• the time remaining in the current window expressed in seconds or as a timestamp;

Interoperability issues A major interoperability issue in throttling is the lack of standard headers, because:

• each implementation associates different semantics to the same header field names;
• header field names proliferates.

User Agents interfacing with different servers may thus need to process different headers, or the very same application interface that sits behind different reverse proxies may reply with different throttling headers.

Examples

Unparameterized responses

Throttling information in responses The client exhausted its service-limit for the next 50 seconds. The time-window is communicated out-of-band or inferred by the field values. Request: Response: Since the field values are not necessarily correlated with the response status code, a subsequent request is not required to fail. The example below shows that the server decided to serve the request even if RateLimit-Remaining is 0. Another server, or the same server under other load conditions, could have decided to throttle the request instead. Request: Response:

Use in conjunction with custom fields The server uses two custom fields, namely acme-RateLimit-DayLimit and acme-RateLimit-HourLimit to expose the following policy:

• 5000 daily quota-units;
• 1000 hourly quota-units.

The client consumed 4900 quota-units in the first 14 hours. Despite the next hourly limit of 1000 quota-units, the closest limit to reach is the daily one. The server then exposes the RateLimit-* fields to inform the client that:

• it has only 100 quota-units left;
• the window will reset in 10 hours.

Request: Response:

Use for limiting concurrency Throttling fields may be used to limit concurrency, advertising limits that are lower than the usual ones in case of saturation, thus increasing availability. The server adopted a basic policy of 100 quota-units per minute, and in case of resource exhaustion adapts the returned values reducing both RateLimit-Limit and RateLimit-Remaining. After 2 seconds the client consumed 40 quota-units Request: Response: At the subsequent request - due to resource exhaustion - the server advertises only RateLimit-Remaining: 20. Request: Response:

Use in throttled responses A client exhausted its quota and the server throttles it sending Retry-After. In this example, the values of Retry-After and RateLimit-Reset reference the same moment, but this is not a requirement. The 429 Too Many Requests HTTP status code is just used as an example. Request: Response:

Parameterized responses

Throttling window specified via parameter The client has 99 quota-units left for the next 50 seconds. The time-window is communicated by the w parameter, so we know the throughput is 100 quota-units per minute. Request: Response:

Dynamic limits with parameterized windows The policy conveyed by RateLimit-Limit states that the server accepts 100 quota-units per minute. To avoid resource exhaustion, the server artificially lowers the actual limits returned in the throttling headers. The RateLimit-Remaining then advertises only 9 quota-units for the next 50 seconds to slow down the client. Note that the server could have lowered even the other values in RateLimit-Limit: this specification does not mandate any relation between the field values contained in subsequent responses. Request: Response:

Dynamic limits for pushing back and slowing down Continuing the previous example, let's say the client waits 10 seconds and performs a new request which, due to resource exhaustion, the server rejects and pushes back, advertising RateLimit-Remaining: 0 for the next 20 seconds. The server advertises a smaller window with a lower limit to slow down the client for the rest of its original window after the 20 seconds elapse. Request: Response:

Dynamic limits for pushing back with Retry-After and slow down Alternatively, given the same context where the previous example starts, we can convey the same information to the client via Retry-After, with the advantage that the server can now specify the policy's nominal limit and window that will apply after the reset, e.g. assuming the resource exhaustion is likely to be gone by then, so the advertised policy does not need to be adjusted, yet we managed to stop requests for a while and slow down the rest of the current window. Request: Response: Note that in this last response the client is expected to honor Retry-After and perform no requests for the specified amount of time, whereas the previous example would not force the client to stop requests before the reset time is elapsed, as it would still be free to query again the server even if it is likely to have the request rejected.

Missing Remaining information The server does not expose RateLimit-Remaining values (for example, because the underlying counters are not available). Instead, it resets the limit counter every second. It communicates to the client the limit of 10 quota-units per second always returning the couple RateLimit-Limit and RateLimit-Reset. Request: Response: Request: Response:

Use with multiple windows This is a standardized way of describing the policy detailed in :

• 5000 daily quota-units;
• 1000 hourly quota-units.

The client consumed 4900 quota-units in the first 14 hours. Despite the next hourly limit of 1000 quota-units, the closest limit to reach is the daily one. The server then exposes the RateLimit fields to inform the client that:

• it has only 100 quota-units left;
• the window will reset in 10 hours;
• the expiring-limit is 5000.

Request: Response:

FAQ RFC Editor: Please remove this section before publication.

1. Why defining standard fields for throttling? To simplify enforcement of throttling policies.
2. Can I use RateLimit-* in throttled responses (eg with status code 429)? Yes, you can.
3. Are those specs tied to RFC 6585? No. defines the 429 status code and we use it just as an example of a throttled request, that could instead use even 403 or whatever status code. The goal of this specification is to standardize the name and semantic of three ratelimit fields widely used on the internet. Stricter relations with status codes or error response payloads would impose behaviors to all the existing implementations making the adoption more complex.
4. Why don't pass the throttling scope as a parameter? The word "scope" can have different meanings: for example it can be an URL, or an authorization scope. Since authorization is out of the scope of this document (see ), and that we rely only on , in we defined "scope" in terms of URL. Since clients are not required to process quota policies (see ), we could add a new "RateLimit-Scope" field to this spec. See this discussion on a similar thread Specific ecosystems can still bake their own prefixed parameters, such as acme-auth-scope or acme-url-scope and ensure that clients process them. This behavior cannot be relied upon when communicating between different ecosystems. We are open to suggestions: comment on this issue
5. Why using delay-seconds instead of a UNIX Timestamp? Why not using subsecond precision? Using delay-seconds aligns with Retry-After, which is returned in similar contexts, eg on 429 responses. Timestamps require a clock synchronization protocol (see Section 5.6.7 of ). This may be problematic (e.g. clock adjustment, clock skew, failure of hardcoded clock synchronization servers, IoT devices, ..). Moreover timestamps may not be monotonically increasing due to clock adjustment. See Another NTP client failure story We did not use subsecond precision because:
   • that is more subject to system clock correction like the one implemented via the adjtimex() Linux system call;
   • response-time latency may not make it worth. A brief discussion on the subject is on the httpwg ml
   • almost all rate-limit headers implementations do not use it.
6. Why not support multiple quota remaining? While this might be of some value, my experience suggests that overly-complex quota implementations results in lower effectiveness of this policy. This spec allows the client to easily focusing on RateLimit-Remaining and RateLimit-Reset.
7. Shouldn't I limit concurrency instead of request rate? You can use this specification to limit concurrency at the HTTP level (see {#use-for-limiting-concurrency}) and help clients to shape their requests avoiding being throttled out. A problematic way to limit concurrency is connection dropping, especially when connections are multiplexed (e.g. HTTP/2) because this results in unserviced client requests, which is something we want to avoid. A semantic way to limit concurrency is to return 503 + Retry-After in case of resource saturation (e.g. thrashing, connection queues too long, Service Level Objectives not meet, ..). Saturation conditions can be either dynamic or static: all this is out of the scope for the current document.
8. Do a positive value of RateLimit-Remaining imply any service guarantee for my future requests to be served? No. FAQ integrated in .
9. Is the quota-policy definition too complex? You can always return the simplest form of the 3 fields

The key runtime value is the first element of the list: expiring-limit, the others quota-policy are informative. So for the following field: the key value is the one referencing the lowest limit: 100

1. Can we use shorter names? Why don't put everything in one field?

The most common syntax we found on the web is X-RateLimit-* and when starting this I-D we opted for it The basic form of those fields is easily parseable, even by implementers processing responses using technologies like dynamic interpreter with limited syntax. Using a single field complicates parsing and takes a significantly different approach from the existing ones: this can limit adoption.

1. Why don't mention connections? Beware of the term "connection": &#65532; &#65532; - it is just one possible saturation cause. Once you go that path &#65532; you will expose other infrastructural details (bandwidth, CPU, .. see ) &#65532; and complicate client compliance; &#65532; - it is an infrastructural detail defined in terms of server and network &#65532; rather than the consumed service. This specification protects the services first, and then the infrastructures through client cooperation (see ). &#65532; &#65532; RateLimit fields enable sending on the same connection different limit values &#65532; on each response, depending on the policy scope (e.g. per-user, per-custom-key, ..) &#65532;
2. Can intermediaries alter RateLimit fields? Generally, they should not because it might result in unserviced requests. There are reasonable use cases for intermediaries mangling RateLimit fields though, e.g. when they enforce stricter quota-policies, or when they are an active component of the service. In those case we will consider them as part of the originating infrastructure.
3. Why the w parameter is just informative? Could it be used by a client to determine the request rate? A non-informative w parameter might be fine in an environment where clients and servers are tightly coupled. Conveying policies with this detail on a large scale would be very complex and implementations would be likely not interoperable. We thus decided to leave w as an informational parameter and only rely on RateLimit-Limit, RateLimit-Remaining and RateLimit-Reset for defining the throttling behavior.

RateLimit fields currently used on the web RFC Editor: Please remove this section before publication. Commonly used header field names are:

• X-RateLimit-Limit, X-RateLimit-Remaining, X-RateLimit-Reset;
• X-Rate-Limit-Limit, X-Rate-Limit-Remaining, X-Rate-Limit-Reset.

There are variants too, where the window is specified in the header field name, eg:

• x-ratelimit-limit-minute, x-ratelimit-limit-hour, x-ratelimit-limit-day
• x-ratelimit-remaining-minute, x-ratelimit-remaining-hour, x-ratelimit-remaining-day

Here are some interoperability issues:

• X-RateLimit-Remaining references different values, depending on the implementation:
  • seconds remaining to the window expiration
  • milliseconds remaining to the window expiration
  • seconds since UTC, in UNIX Timestamp
  • a datetime, either IMF-fixdate or
• different headers, with the same semantic, are used by different implementers:
  • X-RateLimit-Limit and X-Rate-Limit-Limit
  • X-RateLimit-Remaining and X-Rate-Limit-Remaining
  • X-RateLimit-Reset and X-Rate-Limit-Reset

The semantic of RateLimit-Remaining depends on the windowing algorithm. A sliding window policy for example may result in having a RateLimit-Remaining value related to the ratio between the current and the maximum throughput. e.g. If this is the case, the optimal solution is to achieve At this point you should stop increasing your request rate.

Acknowledgements Thanks to Willi Schoenborn, Alejandro Martinez Ruiz, Alessandro Ranellucci, Amos Jeffries, Martin Thomson, Erik Wilde and Mark Nottingham for being the initial contributors of these specifications. Kudos to the first community implementers: Aapo Talvensaari, Nathan Friedly and Sanyam Dogra. In addition to the people above, this document owes a lot to the extensive discussion in the HTTPAPI workgroup, including Rich Salz, Darrel Miller and Julian Reschke.

Changes RFC Editor: Please remove this section before publication.

Since draft-ietf-httpapi-ratelimit-headers-01

• Update IANA considerations #60
• Use Structured fields #58
• Reorganize document #67

Since draft-ietf-httpapi-ratelimit-headers-00

• Use I-D.httpbis-semantics, which includes referencing delay-seconds instead of delta-seconds. #5