-
Notifications
You must be signed in to change notification settings - Fork 22
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
New API porposal - IP high-throughput elastic network API #90
base: main
Are you sure you want to change the base?
Conversation
| ---- | ----- | | ||
| API family name | IP high-throughput elastic network API | | ||
| API family owner | China Mobile | | ||
| API summary | This API provides scheduling ability of huge data transmission in network. It can calculate load-balancing paths automatically with given time duration, total data amount and bandwidth limit, and satisfy huge data transmission requirements with dynamic and elastic bandwidth allocation algorithms, tidal-traffic-aware segment list adjustment and SRv6 multiple segment list technology.<br /><br />IP high-throughput elastic network API has been successfully commercialized across a variety of sectors. This API enable network operators to monetize their dynamic bandwidth requirements effectively. The capabilities of high throughput and extensive bandwidth present considerable potential for future market applications. It is particularly important in advanced fields such as technological research, astronomical and meteorological analysis, data backup, and video editing, where significant advances have led to the emergence of a large number of commercial use cases.<br/>This API is suitable for customers in industries with massive data transfer requirements, such as Cross-cloud backup, model training at intelligent compute centers, video editing, and scientific computing. For example, the Five-hundred-meter Aperture Spherical radio Telescope (FAST), located in GuiZhou province of China, could generate more than 10 PB of observational data per year. By calling this API, users can dynamically adjust bandwidth and elastically orchestrate network resources according to their specific requirements for bandwidth, latency, and file size. This capability facilitates the efficient and expedient transmission of large observational datasets within short timeframes. <br /><br />Input:<br/>1. Bandwidth Requirement: Time-segmented bandwidth requirements (can include multiple time points)<br/>2. Bandwidth Requirement Type: Upload bandwidth, download bandwidth, two-way bandwidth Service<br/>3. Latency Requirement: Latency guarantee requirement range<br/>4. File Requirements: Size of files to be transmitted <br/>5. Guarantee Requirements: Bandwidth guarantee and load sharing requirements<br/><br />Output:<br/>1. Estimation of transmission rate by time segment<br/>2. Estimated transmission duration,<br/>3. Statistical conclusions after successful transmission | |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
The API Summary states : "This API provides scheduling ability of huge data transmission in network.".
So I think the primary operation is a request to schedule a huge data transmission, based on the bandwidth/latency/loss etc. input parameters.
And I think the output response should include:
- an indication that the operator is able to meet the scheduling request (or not).
- a schedule of date(s) showing when the operator can achieve the huge data transmission (e.g. an array of network quiet times) . I think the 'estimation of transmission rate', 'estimated transmission duration' are specific to each individual item in that array (i.e. each possible schedule time that the operator can offer). The API Consumer may choose the schedule date with the best performance, or they could choose a schedule date with inferior performance which fulfils the transmission requirements but with less bandwidth than the best schedule date. For example, the API Consumer may require the earliest possible date, or the cheapest option.
- Any cost difference between the different schedule dates (if appropriate).
- (optional) any conditions required for the operator to meet the request (e.g. a particular entry point to the network must be used).
- Statistical conclusions after successful and unsuccessful transmission.
(note this captures my comment from the API Backlog meeting 10-10-24)
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thank you very much for Kevsy's excellent suggestion. We respond to your API output parameter suggestions as follows:
- Yes, cloud operators match and compare their service capabilities with the user-side input of bandwidth requirements, types of bandwidth demand, business latency requirements, file requirements, and assurance requirements to determine whether the operator is able to meet the scheduling request.
- The API can provide multiple data transfer plans: each includes the data transfer period, estimated transfer time, and transfer rate; different charges are calculated by domain B for different schemes, allowing users to choose from them. The API consumer may choose the schedule date with the best performance, or they could choose a schedule date with inferior performance which fulfills the transmission requirements but with less bandwidth than the best schedule time(date), for the earliest possible date, or the cheapest option. The cost calculation in domain B is beyond the scope of consideration for this API.
- The cost difference can be mapped and calculated based on the different data transfer periods, estimated transfer times, and transfer rates among various transmission plans.
- There is no specific value application scenario for the particular entry point to the network as an API output at present, so it can be considered as an optional item.
- Data transmission statistical conclusions after successful and unsuccessful transmissions have been supported in the API.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
More information on technology dependency:
The IP high-throughput elastic network API relies on an SRv6-based IP bearer network, and this IP bearer network:
- Is based on the Segment Routing Architecture (IETF RFC8402);
- Implements SR multi-Segment load-sharing technology through SR v6 network programming (IETF RFC8986);
- Adjusts QoS in real-time using the Network Configuration Protocol (NETCONF) (IETF RFC6241) to ensure the availability of the high-throughput large bandwidth capability API.
Differences from current proposals:
IP High-Throughput Elastic Network API: The IP high-throughput elastic network API relies on an IPSR v6 network and achieves the capability of huge data transmission in the network through srv6 technology, such as UCMP or ECMP. It is a technology within the domain of IP bearer networks and thus differs significantly in terms of network domain and business functionality from the Dedicated Networks API (5G NPN), The QoD API (Home Devices), and the Network Slicing API (RAN slicing).
Dedicated Networks API: This API relies on networking features such as the 5G NPN (specifically the PNI-NPN, TS 23.501 Clause 5.30) or the Network Communication Service (GSMA OPG.02 Annex H), which offers several realization options including Slices or APN/DNN.
The QoD API (Home Devices Quality on Demand (QoD)): This API necessitates the support of Differentiated Services (DiffServ) on home network routers as a mechanism for classifying and managing network traffic and providing quality of service (QoS).
Network Slicing API: This API requires the availability of RAN slicing, UDM, PCF, and network slicing management services as specified in 3GPP TS 28.531 (network slice provisioning operations), TS 28.541 (Information model definitions for network slice NRM) standards.
Modified the output parameters in the API summary column and added the Uniqueness column to explain the differences between this API and NetworkSlicing, QoD, DedicatedNetwork API
What type of PR is this?
New API proposal 'IP high-throughput elastic network API'
What this PR does / why we need it:
'IP high-throughput elastic network API' provides the scheduling ability of huge data transmission in the network.
Which issue(s) this PR fixes:
Fixes #89