Frequently asked questions.

The BQN Platform is Bequant’s traffic optimization and network visibility platform for ISPs and mobile operators. It helps operators improve subscriber quality of experience by reducing congestion, latency, packet loss and retransmissions, while also giving network and support teams better visibility into subscriber, application and network-level performance.

For ISPs and mobile operators, the main problems it solves are overloaded access links, inconsistent customer experience during peak hours, complex subscriber policy enforcement, limited router scalability and slow troubleshooting. BQN addresses these issues with TCP optimization, active queue management, AI-based congestion mitigation, DPI-enabled application awareness and per-subscriber policy enforcement.

In practice, this helps operators get more value from existing infrastructure, delay unnecessary capacity upgrades, reduce support workload and deliver a more consistent service to subscribers.

Bequant detects network congestion and manages the network capacity while improving the subscriber quality of experience. Keeping QoE constant, the network can accommodate more subscribers. Anibal Enríquez, CEO of Communicate, has reported 10% more users per Access Point.

Keeping the number of subscribers constant, the subscriber QoE is improved by reducing the network latency and losses. Cajun Networks has reported 83% reduction in latency and retransmissions going down from a high of 6,6% to an average of 0.32%.

Bequant traffic AI congestion mitigation (ACM) mitigates the congestion, active queue management (AQM) enforces rate limits in an optimal way and application-based traffic policies (DPI enabled) moderate the demand of heavy applications during peak hours.

More information can be found on Cajun Case study and Enhanced Subscriber QoE.

It helps by providing key insights into how the subscriber is using the Internet connection. For example, using a DPI enabled product such as Bequant, a technical support team can see what applications are being used at the home network and realize that the reported problem is related to that usage, for example to a big download in combination with streaming sessions by several members of the family.

Improving network performance reduces churn and support costs by removing many of the problems that make subscribers dissatisfied and trigger support calls: high latency, packet loss, congestion, buffering, unstable video calls and inconsistent speeds during peak hours.

For churn, better performance means subscribers experience the service they expected more consistently. When gaming is responsive, video streams do not stall, calls remain stable and downloads complete at expected speeds, customers have fewer reasons to complain or switch providers. Bequant helps by reducing congestion, latency and packet loss while giving ISPs visibility into subscriber-level QoE before problems become churn risks.

For support costs, better performance lowers the number of tickets caused by network-quality issues. It also gives support and network teams clearer data when problems do occur, so they can identify the cause. That reduces repeat calls, escalations and unnecessary truck rolls.

The benefits are manifold:

• Reduction in latency and packet loss. This appears across many deployments. Cajun Broadband achieved latencies below 10 ms and losses under 1%. InfoWest now consistently sees latencies between 15 ms and 28 ms, with losses below 0.4% — an 80% reduction.
• Fewer support calls. Bequant’s QoE improvements reduce the volume of customer support interactions. Resound Networks reports a 35% drop in calls related to network issues. Arbuckle has also seen a measurable decrease after deploying the solution.
• Congestion mitigation during peak hours. Bequant’s traffic optimization and application-aware policies help operators mitigate the effects of network congestion — whether in the network core or at the subscriber’s access link. La Harpe Telephone used Bequant to successfully manage traffic surges during special events. DigitalC is another example, resolving capacity constraints in one of their service areas.
• Opportunity to increase subscriber counts or plan speeds without degrading QoE. In some cases, the QoE gains can be reinvested to grow the subscriber base or raise plan speeds while maintaining service quality, avoiding the cost of a capacity upgrade. Arbuckle is a clear case: they were able to double the speed of some of their plans.

The main differences between Bequant and Preseem are:

• TCP optimization accelerates TCP traffic and provides the rest of the Bequant features with accurate metrics for loss, latency and congestion. Preseem has no TCP optimization.
• Application awareness thanks to DPI. Preseem has no DPI feature.
• AQM with TCP-native speed control. Preseem AQM does not have TCP-native speed control and senders are slowed down by dropping packets.
• Up to 400 Gbps of throughput capacity per server. Preseem’s maximum capacity is 100 Gbps.
• Clustering several Bequant servers enables seamless management. Preseem has no clustering feature.

The Bequant BQN Platform is deployed as an inline element on standard Intel or AMD servers. It is placed where it can see and manage subscriber traffic, usually between the subscriber access network and the upstream/core network.

It does not require replacing the ISP’s existing routers or redesigning the network architecture. In most deployments, Bequant complements the existing router infrastructure by offloading traffic optimization, subscriber policy enforcement and analytics from routers to a dedicated platform. This allows routers to focus on routing while BQN handles TCP optimization, active queue management, congestion mitigation, DPI-based policies and subscriber-level visibility.

The main network changes are normally limited to inserting the BQN server or cluster into the traffic path and integrating it with existing operational systems such as RADIUS and billing platforms when subscriber-specific policies are required. For multi-site or higher-capacity networks, multiple BQN servers can be clustered and centrally managed.

Bequant integrates with existing billing and subscriber management systems by importing subscriber-plan information and using it to enforce per-subscriber traffic policies automatically.

The BQN Platform can receive subscriber data through RADIUS signaling, direct interaction with supported billing systems, or the Bequant REST API. This allows the platform to map each subscriber to the correct service plan without requiring engineers to configure queues manually on routers.

Bequant already supports integrations with multiple ISP billing platforms — currently 20 vendors and growing — including systems such as Powercode, Sonar, Splynx, UISP, Wisphub, Wispro and others.

TCP optimization is important for ISPs because most subscriber traffic still runs over TCP, and not all TCP implementations perform well in today's access networks, especially in wireless links with random losses. In practice this means subscribers often get less speed than the network can deliver (slower downloads, unstable video) while paid-for capacity sits unused. TCP optimization fixes the transport itself, recovering that lost throughput without any change to subscriber devices or applications.

The Bequant BQN Platform does this with a transparent TCP optimization (TCPO) proxy that handles packet loss more efficiently, reducing retransmissions and latency and delivering more consistent application performance. The gains are measurable in production: InfoWest, for example, reported an 80% reduction in packet retransmissions after deploying Bequant.

Bequant TCPO also gives the rest of the platform better tools. Because TCP transport headers are not encrypted, each flow exposes accurate per-subscriber metrics (round-trip time, packet loss and retransmissions) that Bequant Automatic Congestion Management (ACM) uses to detect bottlenecks precisely and in real time. And when enforcing rate limits, the Bequant Active Queue Management (AQM) can leverage TCP's native flow-control mechanisms to slow the senders down at the source, keeping packet drops and retransmissions far lower than a traditional shaper or policer.

Bequant detects network congestion and manages the network capacity while improving the subscriber quality of experience. Keeping QoE constant, the network can accommodate more subscribers. Anibal Enríquez, CEO of Communicate, has reported 10% more users per Access Point.

Keeping the number of subscribers constant, the subscriber QoE is improved by reducing the network latency and losses. Cajun Networks has reported 83% reduction in latency and retransmissions going down from a high of 6,6% to an average of 0.32%.

Bequant traffic AI congestion mitigation (ACM) mitigates the congestion, active queue management (AQM) enforces rate limits in an optimal way and application-based traffic policies (DPI enabled) moderate the demand of heavy applications during peak hours.

More information can be found on Cajun Case study and Enhanced Subscriber QoE.

You can prevent routers from being overloaded by offloading subscriber plan enforcement to Bequant traffic management system. At the core of Bequant architecture sits an advanced queue management (AQM) built to scale. Bequant runs on off-the-shelf Intel or AMD servers and supports very high throughput, up to 400 Gbps in one single server.

More information can be found on BQN Platform.

Routers run an optimal packet fast path for core routing functionality and a less optimal path for ACL processing or QoS queues. In many models, just activating those features increases the CPU load and memory usage and reduces the throughput capacity. The impact becomes bigger as more entries are added.

In Bequant, our optimal path involves traffic classification, queues and policy enforcement because traffic management is our core functionality. Bequant incorporates technology such as DPDK to obtain optimal performance.

Bequant runs on off-the-shelf Intel or AMD servers, which typically have faster CPU and memory than router hardware. Thanks to superior hardware and implementation, Bequant supports very high throughput, up to 400 Gbps in one single server.

More information can be found on BQN Platform.

Bequant can enforce per-subscriber traffic policies with very little work by the network engineering team. Bequant can get policies or plans automatically from RADIUS signaling, interacting with the billing system (20 vendors so far and growing) or getting them via Bequant REST API.

Bequant AQM automatically detects and prioritizes interactive flows such as teleconferences or gaming, and its AI congestion mitigation (ACM) is self-learning and does not require manual fine tuning. The manual policies can be much simpler and stable, freeing engineering time.

Bequant improvement of subscriber QoS reduces support calls, further alleviating network engineering workload. Chadd Giles, CTO of Resound Networks, reports a 35% drop in support calls.

Yes, Bequant can enforce per-subscriber traffic policies interacting with the billing system. Bequant is currently integrated with 20 vendors so far and growing. The billing vendors include:

• Aradial
• Azotel
• Gaiia
• Gestfy
• HubSoft
• ISPCube
• ISP Gestión
• ISPSolution
• IXC
• Mikrowisp
• Phicus
• Powercode
• Sonar
• Splynx
• UISP
• Vision by Fibersmith
• Visp.net
• WISPControl
• Wisphub
• Wispro

More information can be found on BQN integrated Billing Systems.

Application-aware traffic shaping is a network management technique that identifies the type of application generating traffic (gaming, VoIP, streaming, large downloads, etc.) and then dynamically prioritizes the flows based on that identification. Bequant implements application-aware traffic shaping in two ways:

• Deep-packet-inspection (DPI): identifies the traffic to apply configured policies. One example is the identification of video streaming during peak hours to moderate its demand.
• Prioritization of interactive flows: Bequant automatically identifies interactive flows by their behavior and prioritizes them in the AQM. This requires no explicit policy configuration.

In most cases, additional network capacity will not deliver any gains, because the speed is mostly limited by the access network and the Wi-Fi last hop. Bequant traffic optimization features (TCPO, ACM, AQM, etc.) will reduce congestion in those network elements and improve the quality of experience in a way that an increase in network capacity will not deliver.

An example of this is Cajun Networks, who has reported 83% reduction in latency and retransmissions going down from a high of 6,6% to an average of 0.32% just deploying Bequant, without any network capacity expansion.

You can request a Bequant free trial on BQN Free Trial Application Form.

Managing subscriber policies in a centralized node scales better as traffic grows because a product like Bequant has bigger capacity, it is more economical to deploy and easier to maintain.

Regarding bigger capacity, Bequant uses Intel/AMD server with faster CPU and memory than router hardware and its internal architecture is optimized for policy management. An added benefit is that Bequant sees all traffic and can apply network-wide fair use policies using very simple rules.

Regarding deployment, one single powerful hardware can be much less expensive to acquire and roll out than many smaller ones, including the interaction with a RADIUS or billing system to get the policies from.

Finally, a distributed solution is cumbersome to maintain, because each element will have its own queue configuration to keep coherent with the rest.

You need to apply fair usage policies based on actual network bandwidth limitations. This will not degrade but improve the QoE for most users.

For example, if a segment of your network has a backhaul link that gets saturated during peak-hours, you can use Bequant traffic management to set a limit for the most bandwidth-hungry applications (e.g. video streaming), and a global rate limitation with scheduling that guarantees a minimum bandwidth for all users. As a result, the QoE of the customers served through that link will improve.

In the end, when well implemented, fair-usage policies allow you to share a limited bandwidth between many users while improving their quality of experience.

Deployed bandwidth capacity does not guarantee fast effective speeds: an advanced bandwidth management product is needed. What follows is a short description of the main causes of capacity underutilization and performance degradation and how Bequant tackles them:

• An application, for example a speed test or a video stream, runs at whatever rate its individual flows can sustain end to end, not at the rate of the core network. One of the causes is that TCP congestion control treats packet losses as congestion signals and backs off when it detects one, so a single impaired hop with random losses will cap the flow's rate while network capacity goes unused. Bequant addresses this with its transparent TCP optimization proxy (TCPO), which handles random losses far more efficiently.
• Network elements usually have internal queues to store traffic packets while they are processed. When the equipment is the bottleneck (e.g. a shaper enforcing the rate plan), the queues can grow too much and cause an excessive queue depth (bufferbloat). This increases the latency experienced by the subscriber, which can be quite noticeable in interactive applications like games and teleconferences. Bequant provides Active Queue Management (AQM) that prevents bufferbloat and minimizes delays in interactive applications, for an optimal customer experience.
• The speed a subscriber experiences is the speed of the bottleneck of the traffic path, which is not always their nominal rate plan speed. Common bottleneck examples include in-home Wi-Fi, a busy wireless sector or an oversubscribed OLT. The bottleneck location changes from time to time, as does the maximum speed at which it operates. Bequant Automatic Congestion Management (ACM) detects those bottlenecks in real time and activates the AQM rate limiting, to keep losses and latencies low during congestion episodes.

All three enforce or react to a traffic rate, but they treat the queue very differently:

• Traditional rate limiting (policing) drops every packet that exceeds the configured rate, without buffering. The plan is enforced, but the heavy induced loss makes transport protocols back off sharply, so throughput oscillates and the subscriber rarely sustains the nominal speed.
• Traffic shaping buffers excess packets in a queue and releases them at the configured rate. It is smoother than policing, but with a deep, unmanaged queue the queue depth grows during congestion and latency explodes (bufferbloat), which subscribers notice in games, video calls and teleconferences.
• Active queue management acts on the queue itself: it monitors queue depth and proactively drops or ECN-marks selected packets before the queue fills, so flows slow down smoothly instead of collapsing. The result is rate-plan enforcement with consistently low latency and packet loss.

Bequant implements an advanced AQM with per-subscriber policy enforcement, flow delay isolation, interactive flow prioritization and native TCP flow control. Bequant Automatic Congestion Management (ACM) dynamically detects bottlenecks and instructs the AQM to kick in, keeping latencies low and interactive services responsive in most network conditions.

Peak-hour congestion happens when the combined demand of many subscribers exceeds the capacity of a shared element in the traffic path, such as a backhaul link, a wireless sector or an OLT port. Queues build up, latency and packet losses increase and every subscriber behind that element sees a degraded experience. Adding capacity is not always possible, and it is often not effective, because the bottleneck is frequently in the access network or the in-home Wi-Fi rather than in the core. A solution is needed to detect and mitigate such congestion events.

Bequant is a well-rounded solution that manages peak-hour congestion with several complementary mechanisms. AI congestion mitigation (ACM) detects congestion symptoms per individual subscriber in real time from flow-level latency and loss measurements and activates AQM rate limiting only when and where congestion appears. Active queue management (AQM), with per-flow queues and TCP-native speed control, prevents bufferbloat and automatically prioritizes interactive flows such as gaming, VoIP and teleconferences, so they stay responsive while the link is busy. DPI-enabled application policies moderate the demand of the heaviest applications, such as video streaming, during peak hours, and fair usage policies on a congested segment can combine those application limits with a global rate limit and scheduling that guarantees a minimum bandwidth for every user.

The results are measurable: Cajun Networks reported an 83% reduction in latency, with retransmissions going down from a high of 6.6% to an average of 0.32%, without any network capacity expansion, and La Harpe Telephone used Bequant to successfully manage traffic surges during special events.

A flow monitor is not an active element, it does not act on the traffic, it gets flow information from the traffic devices and makes that information available for analysis. A traffic optimization platform such as Bequant, is deployed online and improves the traffic performance leveraging on the metrics it gets, so the metrics become actionable straight away, even when nobody in network engineering is analyzing the data.

To identify the applications with the biggest bandwidth demand, a traffic equipment with deep packet inspection (DPI) is needed. Bequant has DPI capabilities and it can identify the most popular applications of a specific subscriber, a group of subscribers or the network as a whole. Traffic management rules can be defined to manage those applications and prevent their congestion effects during peak times.

Find an example on how an ISP solved streaming overload in La Harpe Telephone Case study.

It helps by providing key insights into how the subscriber is using the Internet connection. For example, using a DPI enabled product such as Bequant, a technical support team can see what applications are being used at the home network and realize that the reported problem is related to that usage, for example to a big download in combination with streaming sessions by several members of the family.

The network metrics that can signal degradation of the quality of experience better are congestion, losses and latency. If any of these three or all of them combined increase and remain high for some time, there is a high probability that the customer is suffering from a degraded service and will soon complain, or worse, decide to leave without prior notice. This is the reason a product such as Bequant is important; to mitigate QoE issues and to give an ISP the network visibility it needs to proactively solve network issues before they cause churn.

See more information in this Case Study on churn reduction.

By using a product with network visibility such as Bequant, the ISP support team have access to key QoE indicators per subscriber and per network sections that will allow them to focus on impending network issues and address them proactively. Bequant has an AI Congestion Mitigation feature (ACM) that is constantly looking for congestion symptoms per individual subscriber. ACM automatically acts to reduce congestion, but it also reports the congestion found to the network visibility feature, so support teams can rate subscribers based on their experienced congestion. Bequant has tools to facilitate data analysis, such as an automatic ranking of the worst percentile of congestion-suffering subscribers.

Bequant also enables traffic captures and a live view of a subscriber connection at any given time. This real-time view of the network at different levels is what Victor de la Nuez, CEO of WiFi Canarias, highlights about Bequant.

Traffic optimization and network performance data can be combined in a product such as Bequant, that optimizes traffic at the same time it provides metrics of the resulting network performance.

The provided metrics leverage on the traffic optimization part, because Bequant TCP optimization provides accurate latency and losses values, obtained in flows under load, which is what end users experience. This accuracy is key for reliable congestion detection.

TCP Optimization also measures its own acceleration effect by leaving a small fraction of the traffic as a control group to compare the improvements against.

In the latency case, Bequant provides three separate latency metrics to give as much visibility as possible: an average of minimum values (a lower bound of latency values), the median and the percentile-80 (value not exceeded by 80% of the measured values).

The network metrics that can signal degradation of the quality of experience better are congestion, losses and latency. If any of these three or all of them combined increase and remain high for some time, there is a high probability that the customer is suffering from a degraded service. Latency affects the responsiveness of an application, critical in real time applications like teleconferences and gaming. Losses reduce the application available goodput (actual transfer rate of application content), decreasing the effective speed. Finally, congestion brings a degraded experience, with latencies and losses increasing while the available goodput collapses.

This is the reason a product such as Bequant is important; to mitigate QoE issues and to give an ISP the network visibility it needs to proactively solve network issues.

By using a traffic optimization and visibility product that has enough capacity to process all the network traffic from one single location. If geographically distributed locations are required, Bequant servers can be configured in a cluster where any location is accessible from any of the other sites. Of course, collocated servers can also be added to the cluster to collaborate seamlessly with the rest.

AI-driven network analytics speed up the identification of network problems by allowing the support team to express the goals of the issue search rather than following the steps needed to find them, which is more time consuming and requires knowledge of the product capabilities. An AI assistant is a quick way to obtain results for the most common cases, leaving direct interaction with the visibility data for the most advanced cases.

First, the network has deployed a traffic optimization element such as Bequant, because the first level of proactiveness is improving network performance without any human intervention, just using a closed control loop that is continuously processing the traffic, identifying issues and solving them as they occur.

The second level of proactiveness is for that traffic optimization product to provide network visibility that is accurate and detailed enough to detect QoE issues as soon as they happen, even before the affected customers complain.

The third level is importing the traffic optimization metrics into the network management system for analysis and notifications. Next generation management systems also integrate BSS and OSS system information to have a more holistic view of the subscriber satisfaction: the experienced network quality obtained from traffic optimization along with the history of interactions with customer support.

The benefits are manifold:

• Reduction in latency and packet loss. This appears across many deployments. Cajun Broadband achieved latencies below 10 ms and losses under 1%. InfoWest now consistently sees latencies between 15 ms and 28 ms, with losses below 0.4% — an 80% reduction.
• Fewer support calls. Bequant’s QoE improvements reduce the volume of customer support interactions. Resound Networks reports a 35% drop in calls related to network issues. Arbuckle has also seen a measurable decrease after deploying the solution.
• Congestion mitigation during peak hours. Bequant’s traffic optimization and application-aware policies help operators mitigate the effects of network congestion — whether in the network core or at the subscriber’s access link. La Harpe Telephone used Bequant to successfully manage traffic surges during special events. DigitalC is another example, resolving capacity constraints in one of their service areas.
• Opportunity to increase subscriber counts or plan speeds without degrading QoE. In some cases, the QoE gains can be reinvested to grow the subscriber base or raise plan speeds while maintaining service quality, avoiding the cost of a capacity upgrade. Arbuckle is a clear case: they were able to double the speed of some of their plans.

The financial return on investment (ROI) with Bequant is driven by key commercial and operational advantages: superior Quality of Experience (QoE) strengthens market reputation to accelerate subscriber acquisition while significantly reducing churn, while fewer support tickets and lower truck rolls directly cut operating expenses (OpEx) to deliver an immediate, positive impact on the ISP’s P&L.

Furthermore, Bequant’s DPI-based application shaping enables ISPs to save up to 20% on bandwidth. This directly reduces IP transit costs while unlocking existing network capacity, allowing providers to onboard new subscribers using their current infrastructure and deferring future CapEx investments.

As a result of these combined efficiency gains and cost reductions, Bequant ISP customers consistently achieve an ROI exceeding 30% starting from their very first month of operation.

ISPs sell Internet access, so the customer experience while using that service is central to why subscribers signed up and why they stay. With a traffic optimization product such as Bequant, network quality of experience becomes better and more consistent, reducing the events that can trigger dissatisfaction and, eventually, churn to a competitor.

Assessing churn risk requires customer-interaction data, such as previous complaints, but one key input is the customer’s quality of experience while using the network. With a product like Bequant, an ISP has the metrics needed to understand QoE and act proactively to solve quality issues before they trigger churn. Network visibility not only identifies churn risk from a network-quality perspective; it also gives the ISP the evidence needed to act before the customer complains.

See how Bequant can help reading the brief Reducing Customer Churn through Network KPIs.

From the subscriber’s perspective, good quality of experience means responsive gaming, stable teleconferences and VoIP calls, video streaming without stalling, fast downloads, and a network that performs consistently for the most common applications. From a business perspective, good quality of experience is the level of network performance that keeps customers satisfied, differentiates the ISP from competitors, and requires only a reasonable investment in network infrastructure. Bequant helps ISPs achieve those goals. Many real examples can be found in our Case Studies section.

A mid-size ISP can reduce customer support costs by improving network performance with a traffic optimization product like Bequant. Better network quality reduces the events that trigger calls to the support center or, worse, technician truck rolls to the customer premises. One real example is Resound Networks, which reported a 35% drop in customer calls related to network issues after deploying Bequant.

NPS scores are closely tied to subscriber satisfaction, and from a network perspective, quality of experience while using the service is what matters most. Improving network QoE with a traffic optimization product like Bequant is often the most straightforward and cost-effective way to improve customer satisfaction, especially compared with adding capacity where no capacity bottleneck exists or where other factors, such as in-home Wi-Fi limitations, are the real constraint. One example of home-network limitations and how Bequant can help is explained in the brief Solving the In-Home Wi-Fi Challenge in FTTH Networks.

The Bequant BQN Platform is a traffic optimization product that improves subscriber QoE by reducing latency, packet loss and congestion. It also provides visibility by subscriber, subscriber group and application.

The main differences between Bequant and open-source QoS/router-based solutions are:

• TCP optimization accelerates TCP traffic and provides the rest of the Bequant features with accurate metrics for loss, latency and congestion.
• AI-based congestion mitigation per individual subscriber.
• AQM with per-flow queues and TCP-native speed control, for maximum latency isolation and minimum packet drops.
• Up to 400 Gbps of throughput capacity per server.
• Clustering of several Bequant servers for seamless management.

Open-source QoS and router-based solutions typically:

• Do not have TCP optimization.
• Do not have AI-based congestion mitigation.
• Their AQM does not have per-flow queues; instead, flows share the same queue based on classification, and TCP-native speed control is absent because senders are slowed down by dropping packets.
• Capacity is typically below 100 Gbps per device.
• Do not support clustering.

The main differences between Bequant and Preseem are:

• TCP optimization accelerates TCP traffic and provides the rest of the Bequant features with accurate metrics for loss, latency and congestion. Preseem has no TCP optimization.
• Application awareness thanks to DPI. Preseem has no DPI feature.
• AQM with TCP-native speed control. Preseem AQM does not have TCP-native speed control and senders are slowed down by dropping packets.
• Up to 400 Gbps of throughput capacity per server. Preseem’s maximum capacity is 100 Gbps.
• Clustering several Bequant servers enables seamless management. Preseem has no clustering feature.

The most important evaluation criteria are:

• Active traffic optimization: the solution should improve network performance without manual intervention through TCP optimization, AI-driven congestion mitigation, active queue management and application-aware traffic prioritization.
• Network visibility: the solution should provide proactive insight into network behavior at multiple levels, including per-subscriber, per-subscriber-group, per-application and network-wide performance metrics.
• Scalability: the solution must scale with ISP traffic growth both vertically, through higher performance per server, and horizontally, by adding servers to a centrally managed cluster.
• Reliability: the solution should have a proven record of successful deployments with ISPs of similar size, technology and operational complexity.
• Deployment and integration: the solution should fit the existing network architecture with minimal router changes and integrate cleanly with RADIUS, billing systems, and operational workflows.
• Manageability: the solution should require little or no fine-tuning after deployment, and server clustering should allow occasional configuration changes to be made once and propagated across the cluster.
• Price: pricing should be proportional to the value delivered, including capacity savings, lower support workload, improved QoE and reduced churn risk.

Yes. For many small and mid-size ISPs, a dedicated traffic optimization platform is worth the investment. The investment is easiest to justify when the ISP can connect the traffic optimization platform to measurable outcomes such as avoided capacity expansion, fewer support calls, lower churn risk, improved QoE, higher plan speeds or the ability to serve more subscribers with the same infrastructure. In that sense, the platform should be evaluated not as an extra network cost, but as a way to protect margins while improving customer experience.

The strongest business case is usually better use of existing infrastructure. If congestion, latency, packet loss or overloaded access links are limiting the customer experience, a platform such as Bequant can improve it without requiring large router upgrades or manual per-device tuning.

For smaller teams, the operational value is also important. A centralized platform can simplify subscriber policy enforcement, integrate with RADIUS or billing systems, reduce manual queue configuration and give support teams subscriber-level visibility. That means fewer recurring troubleshooting cycles and faster identification of problems before customers complain.

The Bequant BQN Platform is deployed as an inline element on standard Intel or AMD servers. It is placed where it can see and manage subscriber traffic, usually between the subscriber access network and the upstream/core network.

It does not require replacing the ISP’s existing routers or redesigning the network architecture. In most deployments, Bequant complements the existing router infrastructure by offloading traffic optimization, subscriber policy enforcement and analytics from routers to a dedicated platform. This allows routers to focus on routing while BQN handles TCP optimization, active queue management, congestion mitigation, DPI-based policies and subscriber-level visibility.

The main network changes are normally limited to inserting the BQN server or cluster into the traffic path and integrating it with existing operational systems such as RADIUS and billing platforms when subscriber-specific policies are required. For multi-site or higher-capacity networks, multiple BQN servers can be clustered and centrally managed.

Bequant integrates with existing billing and subscriber management systems by importing subscriber-plan information and using it to enforce per-subscriber traffic policies automatically.

The BQN Platform can receive subscriber data through RADIUS signaling, direct interaction with supported billing systems, or the Bequant REST API. This allows the platform to map each subscriber to the correct service plan without requiring engineers to configure queues manually on routers.

Bequant already supports integrations with multiple ISP billing platforms — currently 20 vendors and growing — including systems such as Powercode, Sonar, Splynx, UISP, Wisphub, Wispro and others.

Bequant is licensed per server, with a monthly subscription based on the peak throughput processed. The subscription includes the right to use the software and access to support. Pricing is independent of the number of registered or concurrent subscribers.

Bequant has been in business for more than a decade and has been deployed successfully in more than 400 networks. We are a product company, 100% focused on improving our BQN traffic optimization platform and supporting our customers’ needs.

ISP customers describe Bequant as valuable because it improves real network performance, gives them better operational visibility and helps them get more from existing infrastructure.

In real deployments, customers report measurable gains such as lower latency, reduced packet loss and retransmissions, fewer support calls and better service consistency during peak traffic periods. For example, Cajun Broadband reported latency improvements and packet loss below 1%, InfoWest reported an 80% reduction in packet retransmissions, and Resound Networks reported a 35% drop in support calls related to network issues.

More customer examples can be found in Case Studies and Resound video testimonial.

The BQN Platform can support networks from small and mid-size ISPs to large multi-site operators. It runs on standard Intel or AMD servers and can scale both vertically and horizontally.

For throughput, a single BQN server can process up to 400 Gbps, depending on the hardware configuration and deployment requirements. When more capacity or geographic distribution is needed, multiple BQN servers can be deployed as a centrally managed cluster. This allows an ISP to expand capacity by adding servers while keeping policy management, analytics and traffic optimization consistent across the network.

Bequant uses DPI and subscriber-level data to make traffic management more precise while protecting customer privacy.

DPI is used to identify application categories and traffic behavior so the BQN Platform can apply application-aware policies. Subscriber-level data is used to map traffic to the correct service plan, enforce per-subscriber policies, identify congestion symptoms and help support teams troubleshoot performance issues faster.

From a privacy perspective, the platform stores only two pieces of information that require protection, and both are encrypted: the subscriber IP address and the subscriber identifier. Access to subscriber-level views is limited to authorized operational teams.

The upfront price of a one-time license is only part of the comparison; the bigger differences show up after deployment:

One-time (perpetual) licenses are CAPEX-heavy: capacity must be sized and paid for in advance, major version upgrades and maintenance are typically billed separately, and the vendor's incentive shifts to the next sale once the license is closed. Traffic optimization is also a moving target (encryption and new congestion-control algorithms keep changing the traffic mix) so a static product loses effectiveness over time.

Subscription licensing converts the cost to OPEX and scales it with actual usage: operators pay for the capacity they process today and grow the license as traffic grows. Software updates and support are included, and the vendor has to earn the renewal, which keeps development focused on the installed base rather than on the next sale.

Bequant follows the subscription model: the BQN Platform is licensed per server, with a monthly subscription based on the peak throughput capacity. The subscription includes the right to use the software, continuous updates and access to support, and pricing is independent of the number of subscribers, so costs track network growth instead of requiring an up-front capacity bet.

They act at different points of the network, and the right choice depends mostly on visibility and coverage:

Radio-based optimization runs in or next to the access equipment (the sector, AP or radio scheduler). It sees radio conditions directly (signal quality, airtime, modulation) and can react to them, but it only covers the subscribers behind that radio segment, it has to be deployed and tuned per site and per equipment vendor, and it cannot help when the bottleneck is elsewhere: backhaul, peering or in-home Wi-Fi.

Inline optimization in the packet core sits on the path of all subscriber traffic, so a single deployment covers every subscriber and every access technology (fixed wireless, fiber, DOCSIS, mobile). Because it works at the flow level, it detects congestion wherever it occurs from the flows' own behavior (packet loss, RTT growth) and applies per-subscriber policy enforcement without requiring any integration with the radio vendor.

The two approaches are complementary rather than exclusive: the radio scheduler manages spectrum efficiency inside its segment, while the inline platform manages end-to-end flow performance. If one has to be chosen, inline is better in networks with mixed access technologies and changing bottleneck locations.

Bequant BQN is deployed inline in the packet core, transparently: no changes to radios, CPE or subscriber devices. Its Automatic Congestion Management (ACM) detects congested sectors and other bottlenecks in real time from flow measurements and activates per-subscriber AQM rate limiting, keeping losses and latencies low wherever the congestion appears.

The key consideration is that the integration must be automatic, not manual, so the traffic optimizer enforcing the rate limits knows each subscriber's plan at all times.

When evaluating or switching billing systems, check that the optimizer can receive subscriber data through more than one path (RADIUS signaling, a direct billing-system integration or a REST API) so the same platform keeps working before, during and after the migration.

The BQN Platform supports all three methods and has off-the-shelf integrations with 20+ ISP billing vendors (including Powercode, Sonar, Splynx, UISP, Wisphub and Wispro) so each subscriber is mapped to the correct service plan automatically, without engineers configuring queues manually on routers.

A properly deployed traffic optimizer includes a bypass path that will be followed by the traffic in the event of a server failure.

Bequant BQN is deployed exactly this way. Because it is a transparent bump-in-the-wire, the bypass mechanism is very flexible, and it can be defined at the physical, link or routing levels:

• Physical level: external Niagara box or Silicom bypass network cards.
• Link level: LACP link aggregation, with bypass links not going through the BQN server wires.
• Routing level: OSPF dynamic routing, with a bypass path of higher cost than the one going through the BQN server.

Yes, none of the main techniques needs to see the payload:

TCP optimization operates at the transport layer: it improves loss recovery, acknowledgment behavior and congestion control of the connection itself, so it works the same whether the payload is plaintext or TLS-encrypted (HTTPS). Nothing is decrypted and end-to-end encryption is never touched.

QUIC encrypts its transport headers, so it cannot be proxied like TCP, but its flows still show measurable behavior on the wire: packet loss, RTT growth, rate changes. Active queue management and AI-based congestion detection act on those signals, so QUIC traffic gets the same bufferbloat control and congestion management as TCP traffic.

DPI does not depend on payload inspection either: it is based on TLS SNI and flow behavioral analysis. This is enough for application visibility and per-application policies.

Bequant BQN applies all three without decrypting subscriber data:

• TCP optimization.
• AQM and congestion management on any flow including QUIC.
• DPI-based application awareness of encrypted traffic.

Subscriber plan limits cap each subscriber individually, but they do not protect a shared element. An access point, a wireless sector or an OLT port is shared by many subscribers, and its capacity is usually lower than the sum of their plan speeds. During peak hours, every subscriber can be below their plan limit while the shared element is saturated: its queues grow, latency and losses increase, and all subscribers behind it see a degraded experience. Per-subscriber shaping alone cannot fix this, because no individual limit is being exceeded.

Topology-based shaping addresses the shared bottleneck. Subscribers are grouped by the network element that serves them (access point, sector, OLT port or backhaul link), and a rate limit matching the capacity of that element is applied to the combined traffic of the group, with scheduling that shares the available bandwidth fairly and guarantees a minimum for every subscriber. The enforcement is done in the traffic optimizer's sophisticated active queue management, instead of in the access equipment, so latency stays low and interactive flows remain prioritized.

Bequant complements this with AI congestion mitigation (ACM), which detects congestion symptoms per subscriber from flow latency and losses even when the location or capacity of the bottleneck is unknown or changing, and activates AQM rate limiting to keep losses and latencies low. Per-subscriber-group visibility shows which access points or OLT ports are suffering congestion, so the network team can decide where a topology-based policy or a capacity upgrade is needed.

Yes. Traffic shaping does not need to be a permanent restriction. A permanent limit below the plan speed penalizes subscribers during off-peak hours, when the network has spare capacity, while a congestion-triggered policy only acts when a shared element is actually close to saturation.

Bequant AI congestion mitigation (ACM) works this way. It continuously monitors per-subscriber latency and losses, obtained from flows under load, and when it detects congestion symptoms it activates AQM rate limiting to keep losses and latencies low during the congestion episode. When demand falls and the congestion symptoms disappear, the limitation is released and subscribers return to their full plan speed. Because ACM is self-learning, it does not require manual fine-tuning of thresholds for each network element.

Application-based policies can follow the same logic: DPI-based limits on bandwidth-hungry applications, such as video streaming, can be focused on peak hours to moderate their demand when the network needs it, leaving them unrestricted the rest of the day. Peak hours can be identified by time of day or, more flexibly, by the overall network throughput.

Application-based shaping uses deep packet inspection (DPI) to identify the applications generating the traffic and to apply policies to specific ones, for example limiting video streaming or large downloads. It is the right approach when a few bandwidth-hungry applications drive most of the peak demand and moderating them frees capacity for everything else.

Topology-based shaping applies a rate limit to the combined traffic of all subscribers served by a shared element, such as an access point, a sector, an OLT port or a backhaul link, matching the capacity of that element and sharing it fairly among its subscribers. It is the right approach when the problem is the shared bottleneck itself, whatever the application mix, because it moves the queue from the access equipment to the traffic optimizer, where it is actively managed.

The two are most effective when combined. On a backhaul link that saturates during peak hours, for example, Bequant can limit the most bandwidth-hungry applications and, at the same time, apply a global rate limit that guarantees a minimum bandwidth for every user. Bequant AQM additionally prioritizes interactive flows automatically, without explicit policy configuration, so gaming and video calls stay responsive under both approaches.

The answer depends on how the access point performs during peak hours, not on its nominal capacity, so the assessment should start with measured data. First, look at peak-hour utilization: the combined throughput of the subscribers served by the access point compared with the capacity it actually delivers in its radio conditions. Second, look at subscriber experience during those same hours: latency, packet losses, retransmissions and congestion. An access point with high utilization but low latency and losses still has headroom, one where latency and losses rise sharply at peak is already at its limit.

Bequant provides this data per subscriber and per subscriber group, so an ISP can group subscribers by access point and compare peak-hour throughput with QoE metrics. It provides an automatic list of the worst percentile of congestion-suffering subscribers, showing which access points are under pressure.

Additional load should then be validated, not assumed: add subscribers gradually and confirm that peak-hour latency, losses and congestion remain within target. Bequant congestion mitigation and AQM can increase the headroom of an existing access point by keeping latency and losses low as it approaches saturation. The reported gain is, for example, around 10% more users per access point, but it depends on each network, its radio conditions and its traffic mix. A Bequant free trial is the best way to measure it on your own network.