Chapter 1. Input data
Prerequisites for creating the «Incoplax» platform
-
Sanctions and market constraints, the rising cost of Western products and their support, and the import substitution course.
-
An increased number of requests for communication solutions from large organizations that cannot be satisfied by a single-server solution (up to 1,000 subscribers).
-
Growth in requests for UC (unified communications) products, omnichannel service, and video conferencing.
-
Over a couple of decades, a number of companies consolidated and, for the most part, automated spontaneously. As a result, large amounts of heterogeneous equipment and software responsible for communications accumulated. They "somehow" work, having been configured by someone at some point, but are not maintained or are maintained with insufficient quality and cannot be adapted to new tasks. According to known data, several equipment models fail every day. In large companies, where simultaneous replacement is impossible but communications are a strategic resource, this is a kind of roulette. Such a material asset is managed by armies of administrators distributed across the country with fragmented knowledge. Among such assets are products from companies that have left the domestic market.
-
Mass abandonment of Western solutions in favor of freely distributed software.
-
Technology is accelerating; communications are increasingly becoming the basis of activity, penetrating deep business processes and integrating more with other tools.
A solution is required that is ready for functional development, whose architecture allows combining the results of different teams working with different technologies and at different speeds without losing manageability and coherence, and that scales horizontally to 1,000,000+ subscribers and users.
Main objectives of the «Incoplax» platform
-
A distributed server platform for telephony and related business tasks (call center, Omni, UC, etc.) built on it.
-
Serves a large number of geographically distributed corporate divisions with high fault-tolerance requirements.
-
Supports separation into several independent companies with independent data management, served on the same physical infrastructure.
-
Can integrate with and be used from arbitrary CRM systems as a communications environment.
-
Is configured, monitored, and managed through an open API; gradually gains its own client application working through the same API.
-
Supports telephony services familiar to the large enterprise sector.
-
Managed scale of 1000+. Methods for increasing scale.
-
Managed availability of 99.(9)%. Methods for increasing availability.
-
Distribution across geographically dispersed sites while keeping them operational during network unavailability.
-
Ability to separate and combine the management point (one admin for everything, or a separate admin in each domain).
-
Support for embedding into other products.
-
Ability to expand functionally toward enterprise, SaaS, and PaaS.
-
No-code and low-code tools for creating information systems seamlessly connected to communication processes.
-
No-code and low-code tools for creating service scripts, business processes, data exchange, and ETL.
-
RAD constructor for information and communication systems.
Requirements
Basic functional requirements
-
Full support for the SIP protocol.
-
Registration of phones under user accounts.
-
Calls between users, hold, transfer, pickup.
-
Second incoming call during an active conversation.
-
Redial to the previous subscriber.
-
Connection to external PBXs and providers with and without registration, with number substitution.
-
Conference calling on the phone.
-
Server conference room (>100 subscribers).
-
Selective meetings with voting (>100 subscribers).
-
IVR scripts for voice call handling with speech recognition and synthesis support.
-
Data management via API.
-
Activity and availability monitoring via API.
-
Two-step call routing based on rules with support for regular expressions, masks, table filters, and modifiers.
-
Logging of calls, scripts, and business logic.
-
LDAP integration.
-
Centralized updates.
-
Group numbers.
-
Queue support.
-
Multiple call support.
-
Storage of telephony events and data, sending to a broker.
-
Call recording storage.
-
Softphone application support.
-
Browser softphone support.
-
Virtualization: several independent companies on one infrastructure.
-
Mechanisms for notifying external systems about events.
-
Video call support.
-
Call control via API.
-
Data model management with change subscriptions.
-
Support for various storages and protocols.
-
Connection to mailboxes.
-
Connection to messengers.
-
Bots for automated service of text and voice channels.
-
Support and management of availability of third-party microservices and web applications.
-
Sniffers from external networks receive no responses from the system and show no interest in it.
Non-functional requirements
Quality attributes are listed in descending priority order — architecture drivers.
-
Scalability/extensibility
-
Availability/reliability
-
Changeability/modifiability
-
Development speed
-
Standards compliance
-
Performance
-
Maintainability
-
Protection against attacks
-
Documentability
-
Integrability
Selected quality attribute scenarios
-
An arbitrary SIP phone connects to the system in 2 minutes, registers, and can make and receive calls, hold, and transfer.
-
With an arbitrary server switched off, a subscriber dials a number and makes a successful call and connection to another subscriber.
-
With 1,000 active conversations, a subscriber dials a number and connects to another subscriber.
-
With 200 cps, a subscriber dials a number and connects to another subscriber.
-
After any server is switched off, the phone remains registered, and the account can be called without significant delays.
-
The system is deployed in Moscow and Kazan. Communication between the cities is absent. A subscriber in Kazan registers and calls another subscriber in Kazan. A subscriber in Moscow registers and calls another subscriber in Moscow.
-
After communication between the cities is restored, a subscriber in Moscow dials a Kazan subscriber’s number and connects to them.
-
An administrator deploys a site in Novosibirsk in less than 1 hour given servers and a deployment plan.
-
A support employee identifies the cause of a server failure within an hour.
-
A support employee identifies the cause of a call failure within an hour.
-
An attacker tries to register or call under a non-existent account or with an incorrect password for the 56th time within 30 seconds. The system does not respond and blocks the IP address at a lower level for 20 minutes. Subsequent attacker requests receive no responses and create ten times less load on the system’s server processors.
-
Administrators of two independent organizations agree and configure routes, each in their own office, within 1 minute, making it possible to call users from the first organization to the second.
-
During a telephone conversation, any server is switched off, and the connection is not interrupted; sound is lost for no more than 1 second.
-
One server serves 200 simultaneous calls without losses and delays at 10 cps.
-
The 201st user joins a conference in less than 3 seconds and hears the speaking subscriber; everyone also hears them.
-
A programmer adds a new script component in less than 1 day.
-
A programmer adds a new script microservice in less than 1 day.
-
An arbitrary web CRM system integrates the API of a system based on «Incoplax» in less than 1 week.
-
An administrator adds 1 server to the cluster on a site and increases performance by 100 cps / 200 simultaneous calls / 1,000 BLF per second / 1,000 registrations per second.
-
During a restart of any server, a subscriber makes a call and completes it with a delay of no more than 5 seconds.
-
During a telephone conversation, communication with external devices is lost. The conversation is terminated by the server forcibly within 2 hours, and the call recording is placed in storage within 2 hours.
-
An MVP of a request management system or help desk is created in less than 8 hours.
-
An MVP of an information system package based on an object model of 10 classes and 50 requirements is created in less than 20 hours.
-
A tabular report, a diagram based on aggregated data, or a dashboard with indicators based on data in the object model is created in 30 minutes.
Quality attribute scenarios are described in more detail in the wiki.
Technical requirements
-
Support for SIP/2.0 over UDP, TCP, TLS, WebSocket, and WebSocket secure.
-
Support for RTP and DTLS (including for WebRTC).
-
HTTP, HTTPS, and WebSocket APIs.
-
Support for requirements of various SIP providers:
-
requiring media traffic to be sent from the connected device’s address;
-
prohibiting REFER;
-
requiring requests to be sent from a single account;
-
limiting the number of simultaneous dialogs;
-
requiring OPTIONS support or reciprocal pings via OPTIONS.
-
-
Support for various SIP devices (that is, focus on the most basic set of SIP RFCs):
-
not supporting negotiation of multiple codecs;
-
unable to work with NAT and substitute the media address/port in SDP;
-
not understanding the interface language;
-
unable to re-register on request.
-
-
Support for wideband codecs (Speex, OPUS).
-
Support for voice codecs g.711a, g.711u, g.729ab, g722, g726, gsm610, iLBC.
-
Support for video codecs H.263, H.264, VP8/VP9.
-
Use of the freely distributed PostgreSQL DB, including one installed at the customer site.
-
Support for distribution across several sites.
-
Support for independent operation of sites when communication between them is absent.
-
Support for the x64 architecture.
-
Cross-platform: support for Linux, Windows, VMware virtual machines, and Hyper-V.
-
Support for multiprocessor systems (NUMA, etc.).
-
Support for delivery in a Docker container.
-
Next chapter: Basic categories and technologies
-
Previous chapter: Introduction