Protecting the Digital Gains: How a Government IXP Could Undermine Uganda’s Future

Why Market Forces Should Drive IXP Ecosystem Growth

Overview (TLDR)

Uganda’s Internet ecosystem has achieved remarkable progress over the past three decades. Sound regulation, private-sector investment, and community collaboration have created one of the most competitive markets in the region. Wholesale bandwidth prices have dropped from roughly $5,000 to $0.50 per megabit in just 15 years, and nearly all networks now interconnect locally through the Uganda Internet Exchange Point (UIXP), a neutral, non-profit service that enables fast, secure, and affordable traffic exchange.

Uganda’s well functioning industry has attracted investment from infrastructure developers, international wholesalers, global content providers, and local access networks – fostering competitive pricing dynamics and sophisticated commercial collaborations. Uganda also recently attracted its first carrier-neutral data center, Raxio, which is a critical facility for hosting large-scale content, facilitating interconnection, and strengthening Uganda’s position in the regional digital economy.

Recently, however, the National Information Technology Authority (NITA-U) announced plans to launch a government-run Internet Exchange Point (IXP) with an ambiguously worded requirement for "Mandatory Domestic Peering" which suggests that ISPs may be obligated to exchange domestic traffic at the new IXP, though the scope and enforcement of this requirement remain unclear. Separately, consultants hired by the Uganda Communications Commission (UCC) produced a market study which recommends the deployment of additional IXPs in secondary cities based on uncorroborated sources and questionable methodologies provided by ChatGPT.

While these initiatives may be well-meaning, they are not based on market demand. As a result, they could unintentionally oversupply and fragment the market, slow sector growth, and weaken the country’s prospects as an emerging regional Internet hub. In particular, the launch of a mandatory government IXP constitutes a state entry into a private-sector industry which could discourage investment, deter new entrants, and pave the way for a de jure national monopoly.

While there is a growing need for redundancy in Uganda's IXP ecosystem, international experience shows that this is best achieved through private-sector competition. If the government requires redundancy beyond what private IXP operators can provide, bilateral Private Network Interconnections (PNIs) are the best solution. PNIs, also known as cross-connects, are simple and cheap to deploy and maintain, and can be used for mandatory public/private interconnections without introducing the risks associated with a multilateral government IXP.

Uganda’s Internet ecosystem is already on a successful path. If we stay the course, the country will soon reach the scale where multiple private IXPs can coexist sustainably based on market demand, similar to the natural growth experienced in Kenya, Nigeria, South Africa, and other regional markets. The government can accelerate this outcome by reconsidering its plan to launch an IXP in favour of safer alternatives such as Private Network Interconnects (PNIs), by reducing taxes to stimulate demand, by making it easier to do business in Uganda, and by continuing to nurture the existing private-sector ecosystem.

A Note on Perspective

As Uganda’s only existing IXP operator, we recognise that some may view our position as self-interested. However, the UIXP was founded in 2001 as a non-profit community initiative and has actively defended open market access including the right for other IXPs to compete based on voluntary market dynamics. Our position is guided not by self-preservation, but by experience, data, and international best practice which we encourage readers to verify through the references provided.

Over nearly two decades, we have built a strong and respectful relationship with the Government of Uganda based on transparency, trust, and collaboration. Multiple government networks are now connected to our exchange and even pay to support the platform’s sustainability. Two of these networks (UTL & NITA) have partnered with us on initiatives such as the Google and Akamai caching projects. We also work closely with various government agencies on security and technical matters from time to time.

In that spirit, our hope is that this post can provide an evidence-based contribution that helps the government avoid actions that could unintentionally harm Uganda’s Internet ecosystem and national economy. 

 

How the Internet & IXPs Work

The Internet is not a single network, but a "network of networks," whereby thousands of independent networks operated by service providers, governments, and content companies interconnect to exchange data. These interconnections can occur in several ways, but the most efficient and scalable way is through Internet Exchange Points (IXPs) which are neutral switching platforms where multiple networks meet to exchange data traffic directly.

A simple economic rule underpins the Internet: "Distance = Cost." The closer content and users are to each other, the faster and cheaper their communication becomes.

In Uganda, most of the content and services people access are hosted abroad. This international traffic reaches the country via submarine cables landing in Mombasa and Dar es Salaam, then travels inland through cross-border fibre connections. Because this data must traverse long distances, it is relatively costly to deliver and slower to reach end users.

IXPs like the Uganda Internet Exchange Point (UIXP) make it easier and more cost-effective for networks and platform providers to host content and services locally. This allows them to serve users from Uganda rather than from abroad. This lowers the overall cost of Internet service delivery, improves Internet performance, reduces Uganda’s reliance on international links, and makes investment in data centers and digital infrastructure more attractive.

Beyond improving performance and affordability, IXPs strengthen national resilience and security by keeping domestic traffic within Uganda, ensuring that it can continue to flow even if international links are disrupted. However, IXPs are not the only way networks can interconnect: operators can also establish Private Network Interconnects (PNIs) or exchange traffic regionally through facilities in neighboring countries. A healthy Internet ecosystem supports multiple interconnection models while letting the market determine the most effective approach. 

 

The State of the Internet in Uganda

For three decades the government, industry, and local community have fostered healthy competition and growth. Today, Uganda’s Internet ecosystem is strong, resilient, and sophisticated:

• Multiple cross-border fibre routes connect Uganda to the global Internet via the Kenyan and Tanzanian coast, and inland to Rwanda, South Sudan, and the DRC.
  
  
• The backbone and metro fibre markets are open and competitive with robust infrastructure sharing and advanced service offerings including dark fibre.
  
  
• The access market is vibrant with a variety of local, regional, and global operators covering mobile, wireless, and fixed-line communications. Competitive dynamics have led to tower sharing and rapid retail price reductions and Uganda now boast some of the lowest access prices in the African region.
  
  
• Data center infrastructure is diversifying as evidenced by the 2021 launch of Raxio, the country’s first carrier-neutral colocation facility.
  
  
• The non-profit UIXP interconnects nearly all networks within the country, operates with excess capacity, and has strong community support. The UIXP is considered a model IXP within the region and plays a leading role in various pan-African institutions. Its success has attracted regional networks and global content providers including Akamai, Meta, and Netflix which now host their data locally, directly improving user experiences and keeping value within the country.
  
  

The results speak for themselves: wholesale bandwidth prices have fallen by more than 99% in just 15 years, while Internet accessibility and reliability continue to improve. These are the hallmarks of a healthy, competitive market that is steadily moving toward the scale needed to support multiple IXPs.

But Uganda does not operate in isolation; it competes directly with Kenya, a more established regional hub where the private sector has produced multiple IXPs and data centers. With coastal access and a larger economy, Kenya holds structural advantages, so Uganda must capitalise on its own strengths and avoid policy missteps that could undermine its attractiveness to investors, content providers, and other industry players.

 

But Uganda does not operate in isolation; it competes directly with Kenya, a more established regional hub where the private sector has produced multiple IXPs and data centres. Kenya's coastal access enables international operators to serve Uganda remotely if local conditions are unattractive. To compete as a secondary inland market, Uganda must aggressively capitalise on its strengths and avoid any missteps that could deter investors, content providers, and other Internet industry players.

What Is the Government Proposing?

In October 2025, NITA-U announced the creation of a government-run IXP called the National IP Peering Exchange (NIPX), with a "Mandatory Domestic Peering" requirement which states that networks would be "encouraged and, where applicable, required to exchange domestic traffic at the NIPX." The stated goals include improving performance, reducing international dependency, and strengthening digital sovereignty. Yet no industry or public consultation appears to have been conducted prior to the announcement, and the referenced National Peering Policy is not publicly available beyond the information included in the announcement.

 

While we appreciate the implication that this new IXP would operate in parallel to the UIXP (and other private IXPs which could theoretically enter the market in the future), we have a serious concerns about this announcement which we will elaborate below.

 

 

Separately, the UCC recently engaged consultants who generated an extensive market analysis report that recommends building two more IXPs in secondary cities based solely on population-based formulas cited from an uncorroborated ITU report. Alarmingly, this recommendation and the wider report appears to have been generated by ChatGPT, casting doubt on its methodological rigor. We understand that the UCC conducted a validation workshop for this report, but we were not included in that process and do not know if they intend to pursue this specific recommendation.

 

 

The screenshots above show page footers that clearly indicate that ChatGPT was used to generate the list of sources and corresponding recommendations in the UCC consultant's report.

These initiatives, though independent and presumably well intentioned, raise serious concerns because they would introduce significant structural changes to the market without clear evidence of demand or national benefit.

 

Exploring the Motivations for a Government IXP

The rationale stated in NITA’s IXP announcement sounds appealing — increased digital sovereignty, faster speeds, lower costs, and greater security — but in Uganda’s case, these issues are already addressed:

• Digital Sovereignty: Nearly all networks in Uganda already interconnect voluntarily via the UIXP and various Private Network Interconnects (PNIs), so the capacity to localise traffic and content already exists. Rather than duplicating this capacity, the government should build on it by creating an enabling environment that attracts more content and cross-border interconnection to existing private-sector facilities such in order to stimulate additional investment in the sector.
  
  
• Performance & Cost: Because the capacity to localise domestic traffic and content already exists, a new IXP would not materially improve national Internet performance unless it succeeds in attracting additional content and interconnection beyond what the existing ecosystem already enables. Since government-operated IXPs tend to under-perform and are typically less attractive to network operators than private IXPs, a new government IXP is unlikely to achieve this. Furthermore, because the cost of peering via the UIXP is a relatively insignificant component of overall service delivery costs, a government IXP would not materially reduce national Internet access costs even if it offered its service for free.
  
  
• Security: With a significant portion of domestic traffic already being localised via the UIXP and various PNIs, Uganda’s reliance on foreign transit networks and cable systems for internal communications is already reduced, significantly lowering the risk of external eavesdropping and traffic manipulation. When it comes to state security, the government has dedicated network infrastructure, encrypts its inter-network traffic, and has a wide range of tools to enforce its monitoring and censorship authority including regulatory compliance systems hosted inside Uganda's various access networks — a far more direct and effective point of control amid Uganda's increasingly complex web of interconnections.
  
  

In NITA's defense, such claims are typical of marketing in our industry and it is likely that any new private IXP would promote themselves similarly. The key difference is that a private IXP would need to prove its stated value proposition by competing for industry demand, whereas NITA can rely on its government authority to mandate connections regardless of market realities.

Beyond the stated objectives, some might wonder if the government's motivation includes a desire for financial income, but we think this would be misguided because most IXPs are not money-making ventures and the opportunities for profit in Uganda are extremely limited. The UIXP, for example, operated entirely on volunteer effort and donations from 2001 to 2017. Even today, with around 30 connected networks, it only earns enough income to cover operations, pay salaries, and ensure long-term sustainability. Simply put, there is no "gold mine" and a government-run IXP is more likely to become a recurring cost than a source of revenue.

In light of the above, we believe a mandatory government IXP would offer no clear commercial, public, or national benefit, and would introduce significant risks compared to the safer alternative of mandatory PNIs.

The Likely Impact on Uganda’s Digital Ecosystem

Regardless of any benefits, the imposition of a mandatory government IXP could have several unintended consequences:

• Premature Fragmentation: Forcefully splitting Uganda's demand for multilateral network interconnection across multiple facilities could unnecessarily dilute network effects which, in turn, could slow the growth of Uganda's data center ecosystem and hinder the country's development as an inland regional Internet hub.
  
  
• Higher Service Delivery Costs: Similarly, forcing ISPs to connect to a new IXP without corresponding market demand could artificially increase network complexity and inflate service delivery costs.
  
  
• New Market Barriers: Mandatory multilateral peering requirements would discourage some networks from entering Uganda. International networks are often unwilling to enter markets where they would be forced to interconnect with competitors or customers against their will.
  
  
• Reduced Investment: In addition to the above, top-down policy shifts, especially government forays into private markets, introduce uncertainty and risk in a sector that depends heavily on stability and predictability.
  
  
• Slower Economic Growth: Premature fragmentation, increased operational costs, new market barriers, and reduced investor confidence would slow market development, job creation, and broader economic growth.
  
  
• Wasted Public Funds: Building redundant infrastructure when private solutions already exist diverts resources from more pressing national priorities.

In summary, mandatory government IXPs can distort market dynamics, increase service delivery costs, discourage market participation, deter investment, and slow economic growth — outcomes that run counter to Uganda's development objectives. They also carry a significant risk of mandate expansion that could concentrate control over domestic interconnection.

In contrast, when a new private IXP enters the market, it offers services and succeeds only if there is genuine demand; otherwise the market naturally filters out unnecessary facilities. This voluntary, demand-driven model avoids premature fragmentation and maintains the coherence of the domestic interconnection environment.

In time, as Uganda’s economy grows and the regional Internet ecosystem expands, a second IXP will become both viable and healthy for the market — but it should emerge naturally from market demand, not through a top-down intervention.

 

What Should Be Done Instead

 

The Ugandan Internet ecosystem is fundamentally well configured as evidenced by various indicators including wholesale bandwidth pricing that is now among the lowest in Africa. To develop it further, the government should avoid unnecessary duplication and forays into private markets. Instead, it should:

• Mandate Private Network Interconnects (PNIs) between government and private networks to gain more redundancy and control over critical traffic flows without incurring all of the costs and risks associated with launching a government IXP.
  
  
• Reduce Internet taxes which now account for over 50% of the cost of access in order to improve accessibility, grow demand, and improve market competitiveness.
  
  
• Facilitate investment and competition by making it easier to do business in Uganda and by avoiding unpredictable and undesirable market interventions.
  
  
• Continue collaborations with community stakeholders including the UIXP in order to identify effective ways to further develop Uganda’s Internet ecosystem and strengthen its regional position.
  
  
• Promote carrier neutral data centers like Raxio by deploying government services (such as NIRA) inside these facilities, and by avoiding actions that would unnecessarily dilute their ecosystems.
  
  
• Authorise LEO satellite systems with in-country ground stations, traffic-localisation requirements, and other restrictions to prevent market distortions.
  
  
• Engage in open consultation with industry and civil society before undertaking significant actions or policy changes to ensure alignment with local and regional market realities.
  
  
• Promote broader economic development because the Internet industry cannot grow unless the overall economy does.

Conclusion

Uganda’s Internet industry has thrived on market-driven investment, competition, and collaboration. A second IXP could strengthen this ecosystem, but only where there is sufficient natural demand to sustain it. A mandatory government-run exchange, by contrast, would offer no clear benefit, risk premature fragmentation of the Internet ecosystem, and weaken Uganda’s long-term growth prospects.

We encourage the government to take a step back, consult openly with stakeholders, and base its next moves on evidence and market realities. Uganda’s incredible progress over the past two decades has been driven by enabling innovation, not imposing it, and this strategy remains key to future growth.

By staying true to these principles, Uganda can protect the gains that have made its Internet industry one of the most advanced in the region. 

Formal Submission

In order to ensure that our views are considered by the government, the UIXP made a formal submission to NITA-U and the Ministry of ICT & National Guidance on December 11, 2025. A copy of this submission can be found on our website.

Safeguarding Uganda's Digital Autonomy in a Low Earth Orbit Future

Low earth orbit (LEO) satellite Internet services are a hot topic in Uganda’s tech community, with services like Starlink frequently trending in online conversations as the government deliberates on whether to allow them in our market and under what conditions. 

The appeal of LEOs is clear: global coverage, fast deployment, and the promise of high-speed connectivity in even the most remote corners of the world. LEO operators such as Starlink, Amazon’s Project Kuiper, and SpaceSail represent a transformative shift in global telecommunications. They offer undeniable potential in markets where terrestrial infrastructure is limited or completely absent. However, in developing markets like Uganda, where the sector is evolving rapidly, we feel that it’s important to approach these developments with caution.

Uganda’s market has developed rapidly over the past 15 years. Wholesale bandwidth pricing has dropped from $4,000 to $1.50 per megabit. We now have multiple cross-border fibre backbones and good metro fibre coverage in most urban areas – with wireless and mobile networks extending beyond that. While there are still challenges, particularly around affordability due to taxes (which make up approximately 60% of the retail price), accessibility, and quality of service, the situation is improving rapidly, and the foundation for sustainable local growth and digital autonomy is in place.



The challenge with LEO services is that they bypass the terrestrial infrastructure that local operators and governments have spent years building. They send local Internet traffic into the sky, circumventing our fibre networks, Internet exchange points, and data centers. If these services were to gain significant traction, it could discourage future investment in domestic infrastructure, undermine the sustainability of local ISPs who employ Ugandans in large numbers, and create a significant digital dependency. Uganda would have
little leverage if such providers were to suddenly change their terms of service, restrict access, or raise prices, especially in a scenario where local alternatives have eroded.

A particular point of concern is the possibility that LEO services might not deploy a local IP node or ground station within Uganda. If user traffic is routed through the sky to neighbouring countries, or even further afield, it would encourage networks to host content and interconnect in those locations instead of Uganda. This would directly threaten Uganda’s emerging role as a regional network interconnection hub just as we are beginning to see momentum.

It is worth noting that Uganda offers some important potential advantages for LEO operators looking to establish ground stations. Our central geographic location within the East African Internet ecosystem positions us to deliver the lowest average latency to neighboring countries' networks. This makes Uganda a strategic choice not only for national connectivity but also as an inland regional hub for LEO infrastructure.

In summary, we’re not opposed to LEO satellite systems and believe they can play an important role, especially in under-served areas – but if they are to operate in Uganda, we feel that it is essential to ensure that their integration is done in a way that supports, rather than undermines, our local ecosystem by requiring them to:

• Deploy a local IP node and/or ground station and route local user traffic via Uganda;
• Only provide service to users in rural areas;
  
• Register a local operating company, or partner with one, and pay taxes accordingly;
• Comply with all other relevant national regulations, just like any other operator.
  
  

These measures would go a long way in ensuring a level playing field and ensuring that the economic and technical benefits of LEOs are shared more equitably. However, even these conditions may not fully mitigate the risks, and policymakers should proceed with a clear understanding of the potential long-term impacts.

LEO satellite services hold exciting potential, but without thoughtful integration, they risk undermining investment in terrestrial infrastructure – the metro networks, fibre backbones, data centers, and exchange points that form the foundation of Uganda’s digital economy. This infrastructure is not just about connectivity; it is the key to our national autonomy and influence in the global digital landscape.

By establishing clear rules for LEO operators, Uganda can embrace this promising innovation while safeguarding critical investment, protecting skilled jobs, and promoting long-term growth.

Endowment Established, Netflix Connected, Akamai Revived, New Services Launched

Message from the Executive Director

2024 was an important year for the UIXP as we took significant steps to secure our long-term sustainability and enhance our service offerings.

We recently achieved a major milestone with the launch of an endowment fund that utilises conservative interest-bearing investments to generate income. This is a key part of our long-term strategy to secure the UIXP’s future; to reduce our service pricing; and to ensure that the UIXP can remain relevant amid looming competition from international players which can offer subsidised services.

We also expanded our service offerings, deploying a Netflix cache with donated cache-fill from Lyca Mobile, and restoring Akamai's cache to service using bandwidth contributions from RENU. In addition, we recently launched a locally hosted Jitsi video conferencing service and an AI social media bot to help improve our online presence. In the background, our infrastructure remains stable and we’ve made a variety of improvements to both physical and digital security.

Despite this progress, the year was not without challenges. Google's worldwide withdrawal from remote peering sessions impacted our ecosystem, while local market conditions such as high taxes and the lingering Facebook ban continue to suppress growth.

Looking ahead, we aim to grow the endowment fund, implement a new ERP with a self-service portal, expand our portfolio of locally hosted services, and continue our involvement in regional Internet communities and initiatives.

I encourage everyone to read the full report below for more details. As always, we are grateful for everyone's support and look forward to working together for the good of the Internet in 2025!

Kyle Spencer,
Executive Director

Endowment Fund Launched

We are proud to announce that we have launched an endowment fund in order to promote the UIXP’s long-term sustainability and reduce its dependence on traditional service fees.

The endowment is based on conservative interest-bearing investments which, starting this year, will cover 25% of the UIXP's annual non-salary operating expenses. The initial installment was made possible by years of savings, conservative financial management, and multi-year prepayments from some of our larger peers, including Meta.

We aim to grow this fund over time with the hope that it will eventually cover 100% of our total annual expenses. This would allow us to offer free or near-free services which would both facilitate the growth of Uganda’s digital economy and help the UIXP stay relevant in the future as our region faces increasing competition from international IXPs that can offer subsidised services.

 

Free Ports & Raxio Discounts

We would like to remind everyone that we offer free 100 Mbps ports at Raxio with no cross-connect charges. We also offer a 6-month 25% discount on all new services at Raxio so, if you would like to connect, upgrade your port, or learn more, please contact us.

 

New Value Added Services 

• Encrypted Video Conferencing: We recently launched a free web-based open-source video conferencing system which offers features similar to Zoom and Google Meet. All user traffic is encrypted and routed via the UIXP which helps to protect users from mass surveillance. We encourage everyone to give it a try and send us feedback: https://kafunda.uixp.co.ug

• AI Marketing Bot: We recently introduced an AI marketing bot that posts occasional promotional updates on Twitter/X. This complements our manual posts and helps to ensure more consistent engagement with our community and the public.

 

Peering Updates 

• 
  Summary: We began and ended 2024 with 32 connected networks and approximately 40Gbps of peak traffic. There was some churn during this period with some networks disconnecting and others connecting. There was also a mid-year traffic depression following the disconnection of Google’s remote peering session with a subsequent boost in traffic in November caused by the restoration of the Akamai cache (see below for more details).
  

• Raxio: We now have six networks connected at the Raxio carrier neutral data center with another connecting soon. As a reminder, we offer free 100M ports and significant discounts on new services at this facility so please contact us if you would like to connect or migrate your network.

• IPv6 Adoption: 33% of all networks (10 out of 30) are peering with our route servers via IPv6 though most traffic is still exchanged via IPv4. We encourage all networks to transition to IPv6 and we are happy to provide support where possible. Click here for a live list of networks peering via IPv6 in Uganda.

• New Peers: Netflix and Kampala Siti Cable [AS328727] connected to the UIXP in 2024, though the Netflix cache is connected to the UIXP via the UIXP’s ASN [AS328998].

• Netflix: We deployed a Netflix cache this year, made accessible to all connected networks through our route servers, with donated cache-fill from Lyca Mobile. However, this cache requires manual activation. To connect, networks must append their BGP announcements to the UIXP route servers with the community string 40027:4000. For more detailed instructions, read the documentation on our website or contact us via e-mail.

• Akamai: After a period of downtime, the Akamai cache was restored on an experimental basis using donated cache-fill from RENU. However, its output is currently limited by the amount of donated cache-fill bandwidth that is available as well as the capacity of the cache cluster hardware.
  
  Traffic from this cache is automatically served to connected networks via the UIXP route servers. However, Akamai’s systems automatically moderate traffic distribution based on a variety of factors (including cache capacity) which can influence whether or not your particular ASN gets traffic from this cache cluster.
  
  We are working with Akamai to upgrade the cache cluster hardware and expect this to occur in Q2 2025. However, in order for the upgraded cluster to increase its output, we will need more cache-fill bandwidth. If we are unable to get additional donated bandwidth, we may try to implement a cache-fill cost-sharing model.

• Google: Due to a global policy change, Google withdrew from all remote peering sessions around the world. This impacted the UIXP as Google was connected via a long-distance fibre link to their nearest PoP in Mombasa. We believe this decision was largely driven by the cost and complexity of maintaining these links.
  
  We explored the possibility of either (a) inheriting the management and cost of the transport link to Mombasa in order to maintain Google’s remote peering session, or (b) replacing Google's remote peering session with a shared Google cache but, in this case, we would need to source cache-fill IPT. Both options were difficult logistically and/or financially. In either case, we would not be able to serve Google traffic for significantly less cost than our peers can already get it via wholesale IPT and, in the remote peering scenario, Google's traffic would still originate in Mombasa so there would be no latency benefit.
  

 

Facility Updates 

• Our switching infrastructure and data center facilities continue to operate reliably. We’ve also strengthened physical and digital security to ensure a safer and more resilient service environment.

• As a reminder, we send technical alerts about significant events, planned maintenance, and unplanned outages to the UIXP Techies mailing list. If your network is peering but is not a member of this list, please contact us to get added.

Accounting & Tax Updates 

• In 2023 we experienced a number of account delinquencies, particularly from smaller networks, which we attributed to challenging market conditions. Throughout 2024 we worked with the affected networks to reduce these balances and are happy to report that almost all have returned to good standing. We sincerely appreciate everyone for honouring their commitments and will do our best to ensure that we can continue to be accommodating should similar challenges arise in the future.

• Our efforts to migrate to a new open-source accounting system were delayed and did not happen in 2024 as we had hoped. However, this remains on our agenda and we hope to undertake this project in 2024. Our goal is to streamline accounting and operational processes through increased automation while providing networks with a convenient self-service portal.

• We remain in good standing with URA and have not faced any significant compliance issues during the past year.

 

Market Challenges

• The 2021 Facebook ban continues to suppress overall utilisation of the Internet and our interconnection service. It continuously inflates the cost of Internet service delivery by forcing users and networks to import Facebook traffic from international sources rather than our local cache. This makes Uganda’s Internet more expensive and its market less attractive for regional interconnection & colocation relative to neighbouring countries.

• We remain concerned by the increasingly high taxes levied on Internet services which now account for over 50% of the total cost of Internet access in Uganda. Like the Facebook ban, this hinders demand and deters investment in Uganda’s digital economy.

 

Community Engagement

• UIXP staff continue to play both leading and supporting roles within the local and regional Internet community via the African IXP Association (AFIX), the Africa Peering and Interconnection Forum (AfPIF), and the African Network Operators Group (AfNOG).

• The UIXP remains a member of the ICT Association of Uganda (ICTAU) and Internet Service Provider Association of Uganda (ISPAU). We work with both organisations to promote market development and policies that are beneficial for our community.

Free 100Mbps Ports, New CDNs, and Infrastructure Upgrades

Message from the Executive Director

2023 was a good year for the UIXP despite some significant challenges.

We started off facing three long-term CDN outages and reduced income due to low demand but finished with four live CDNs, two new peers, and a small surplus. Peak daily traffic increased from 10 Gbps to 45 Gbps while IPv6 adoption increased to 33% of connected networks.

Along the way we restored Google's remote peering link to Mombasa, successfully deployed a new Meta cache, and will soon enable a new Netflix cache in collaboration with Lyca Mobile. We also upgraded our back-end virtualization platform, implemented new cross-site functionality, and enhanced our data storage systems with N+1 redundancy. In addition, new security measures were implemented as part of ongoing efforts to improve physical and digital security.

On the sustainability front, our cash reserves remain untouched and we maintain good standing with URA despite the usual challenges. However, we observed a rise in account delinquencies among smaller networks which we attribute to challenging macroeconomic conditions. In response, we are working with stakeholders to promote progressive government interventions.

We are also working to implement a new open-source accounting system which would provide a number of benefits including improved accessibility and automation. As we progress, we may seek support for testing purposes.

Looking forward, we are thrilled to announce the introduction of free 100 Mbps ports at Raxio and Communications House. This was made possible by the growing utilization of 1 Gbps, 10 Gbps, and 100 Gbps ports. In line with this, any networks that use 100 Mbps ports will no longer be charged while any networks that use free 10 Mbps ports will be upgraded automatically.

I encourage everyone to read the full report below for more details. As always, we are grateful for everyone's support and look forward to working together for the good of the Internet in 2024.

Kyle Spencer,
Executive Director

Free Ports & Raxio Discounts!

We are excited to announce that we now offer free 100 Mbps ports and note that Raxio customers can still access our service without any cross-connect charges.

We believe this move will facilitate growth and competition in our market without imposing any significant risk to our sustainability thanks to the growing utilization of 1 Gbps, 10 Gbps, and 100 Gbps ports. Any networks that use 100 Mbps ports will no longer be charged while networks that use free 10 Mbps ports will be upgraded to 100 Mbps.

As a reminder, we also offer a 6-month 25% discount on all new services at Raxio. If you would like to connect, upgrade your port, or learn more about our services, please visit our website.

Utilization Report

• We began 2023 with 32 connected networks but only ~10Gbps of peak traffic due to cascading CDN outages involving Google and Akamai (in addition to the ongoing Facebook ban which disabled their legacy cache in 2021).
  
  
• Akamai's cache remains offline due to challenges related to IP transit. If anyone would like to donate IP transit for this cache, please contact us.
   
• We helped Google stabilise their remote peering link to Mombasa following months of severe disruption. The successful restoration of their link provided significant benefits to our members and lowered the cost of delivering Google traffic within our catchment area. However, we understand that Google may withdraw remote peering links in the future due to a global policy change so we encourage everyone to prepare for this possibility.
  
  
• Meta deployed a new cache at the UIXP within the Raxio data center. This was a fast and straightforward deployment thanks to Meta's full support for the project. This lowered the cost and improved the performance of traffic related to Meta's various services. Due to demand growth, we are already working with Meta to upgrade their capacity.
  
  
  
• We are in the final stages of deploying a Netflix cache in partnership with Lyca Mobile and hope to bring it online within the coming weeks. This project is somewhat experimental for Netflix since they typically deploy caches within ISPs and MNOs. Netflix's prefixes will be announced to the route servers by the UIXP's ASN [328998] when it goes live due to the nature of their cache's architecture. If you wish to establish a bilateral peering connection with our ASN, please contact us.
   
• IPv6: 33% of all networks (10 out of 30) are peering with our route servers via IPv6. This reflects growing support in our market though most traffic is still exchanged via IPv4. We encourage all networks to transition to IPv6 and we are happy to provide support where possible. Click here for a live list of networks peering via IPv6 in Uganda.

Infrastructure Updates

• We upgraded our server virtualization platform, implemented cross-site functionality, and upgraded our data storage systems with N+1 redundancy throughout. We also took steps to improve physical and digital security, though this is a continuous process and additional measures will be implemented in 2024.
  
  
• We continue to maintain our facility at Communications House. Power, air-conditioning, and security systems are all operating well. Annual uptime exceeds 99.99%.
   
• As a reminder, we send technical alerts about notable changes and foreseeable outages to the members-only techies@uixp.co.ug mailing list. If your network is peering but is not a member and would like to be, please contact us at support@uixp.co.ug.

Invoicing, Collections, and Taxes

• We started the year with no surplus and depressed income due to three long-term CDN outages. Despite this, we ended the year with a small surplus and our cash reserves remain untouched.
  
  
• In 2023 we observed a small but growing number of account delinquencies. We believe these delinquencies reflect increasingly difficult macroeconomic market conditions and we are working with various stakeholders to encourage government interventions that would relieve pressure on our industry. We are also doing our best to accommodate affected networks while balancing our need to maintain fairness and sustainability.
  
  
• We are attempting to migrate to a new open-source accounting system and, as part of this, automate certain processes. We may contact some of you in the near future to request support for testing purposes.
  
  
• We remain in good standing with URA despite the typical challenges of compliance.
  

Environmental Issues

• The 2021 Facebook ban continues to suppress overall utilization of the Internet and our interconnection service. It continuously inflates the cost of Internet service delivery by forcing users and networks to import Facebook traffic from international sources rather than our local cache. This makes Uganda’s Internet more expensive and its market less attractive for regional interconnection & colocation relative to neighbouring countries.
  
  
• We remain concerned by the increasingly high taxes levied on Internet services which now account for over 50% of the total cost of access. Like the Facebook ban, this makes the Internet more expensive and hinders demand for Uganda’s digital economy.

Community Engagement

• All UIXP staff continue to play voluntary leadership and support roles within the African IXP Association (AFIX), the Africa Peering and Interconnection Forum (AfPIF), and the African Network Operators Group (AfNOG).
  
  
• The UIXP remains a member of the ICT Association of Uganda (ICTAU) and Internet Service Provider Association of Uganda (ISPAU). We work with both organizations to promote community development and good policy outcomes.
  
  
• The UIXP continues to facilitate donations for the Uganda Network Operators Group (UGNOG) until such time that they can formalize and register a bank account. Any organizations that wish to sponsor UGNOG should notify us by sending an e-mail to accounts@uixp.co.ug CC isabel.odida@gmail.com.

 

Disaster Recovery: CDN Outages, Financial Challenges, and the Way Forward

Message from the Executive Director

2022 was difficult for the UIXP.

The year started off strong. In January we had 27 networks exchanging 35 Gbps of peak daily traffic. In February we connected our management network to our route servers in order to deliver new low-latency services. In March we deployed a dark fibre link between Raxio and Communications House to facilitate multi-site peering. In May we helped Google extend their global network to Uganda to facilitate the distribution of their traffic via the UIXP.

By June our peak daily traffic reached a record high of 50 Gbps and we planned to start offering free 100M ports – but then a series of external problems destabilized and eventually disabled both Google and Akamai’s peering amid a lingering national Facebook ban. This multifaceted environmental disaster reduced our peak traffic to 6 Gbps and negatively impacted our income immediately after we incurred new costs related to our multi-site expansion.

We are doing our best to help Google and Akamai resolve their issues while we work with Netflix and Meta to deploy new caching systems. We have also assumed a more conservative budgetary posture in an effort to maintain financial stability during this difficult period. We faced an unusually high number of delinquencies at the end of the year and we will be working to address this in the coming months.

In the background, we continue to contribute to community initiatives at the local, regional, and global level. This includes the Uganda Network Operators Group, ISP Association of Uganda, ICT Association of Uganda, the African Network Operators Group, the African Peering & Interconnection Forum, the African IXP Association, and Internet Exchange Federation.

I encourage everyone to read the full report for more details. As always, we are grateful for everyone's support and look forward to working together for the good of the Internet in 2023.

Kyle Spencer,
Executive Director

Raxio Promotional Discounts

As a reminder, significant discounts are available to all Raxio customers including a free cross-connect to the UIXP and a 6-month 25% discount on all exchange services. If you would like to connect, upgrade your port, or learn more about our services, please contact us.

New Members & Connections

Google (AS 15169), Raxio (AS 328821), Group Vivendi Africa (AS 36924), and the Uganda Internet Exchange Point management network (AS 328998) joined the exchange in 2022. This diverse array of content, enterprise, and access networks are all connected to our new node in the Raxio data center, bringing the total number of peers at that location to five.

Looking ahead, we intend to engage a broader range of Raxio customers in order to expand our membership. The more networks that connect, the more valuable the exchange is to everyone.

Technical Updates

We upgraded our inter-site link to 100 Gbps with geographic redundancy in March 2022 in order to accommodate the anticipated flow of Google traffic between Raxio and Communications House where the majority of our customers are connected. C-Squared is providing this link and we are grateful for their reliable and attentive service.

We experienced multiple broadcast storms in 2022 that significantly disrupted some of our member networks. In response, we are gradually implementing technical changes to reduce the possibility of future disruptions. Most notably, we now restrict all ports to a single MAC address.

As a reminder, we send technical alerts about notable changes and foreseeable outages to the members-only techies@uixp.co.ug mailing list. If your network is peering but is not a member and would like to be, please contact us.

Content Networks: Outages & Upcoming Deployments

Four content networks are connected to the UIXP but only YoTV is actively peering. Facebook, Google, and Akamai have all been disabled by different circumstances detailed below:

• The Facebook cache (AS 63293) stopped serving traffic in January 2021 after a government ban forced network operators to block access to the service. This reduced local bandwidth production by 30% and damaged Uganda’s reputation in the global telecommunications industry. Meanwhile, many users continue to access Facebook via virtual private networks (VPNs) which force network operators to import traffic from more expensive international sources. In other words, the Facebook ban increased the cost of service delivery, reduced overall performance, and steered investment to neighbouring countries which are perceived to have more attractive and stable policy environments.
  
  
• Google (AS 15169) established a remote peering session in May 2022 after decommissioning their prototype “self serve” cache (AS 36040). However, Google’s regional transport suppliers were unable to deliver a stable link to Mombasa despite six months of joint troubleshooting. The session was ultimately disabled in December 2022 due to extreme instability and poor quality of service. Google is now working to change transport suppliers and we are actively supporting them with this process.
  
  
• Akamai’s cache (AS 20940) stopped serving traffic when Google’s instability began because it forced the cache-fill donor, NITA-U, to re-route Google related user traffic via international links. This overloaded their capacity and the resulting congestion forced the Akamai cache offline. NITA-U is working to upgrade their international capacity but the procurement timeline is uncertain so we are searching for a new donor and may need to implement a paid access model until Akamai can justify paying for the cache-fill themselves.
  
  

On a more positive note, Netflix and Meta are shipping us new edge caches which we hope to receive next month. Our goal is to bring them online in the second quarter of this year:

• Meta plans to procure their own cache-fill and will peer with our route servers. Their new system will reportedly make it easier for networks to distinguish between traffic related to Facebook and other Meta properties such as Instagram, Oculus, and WhatsApp.
  
  
• Netflix requires a cache-fill donor but their requirements are relatively easy to fulfill due to the static nature of their content catalogue. If your network has 500 Mbps of idle capacity at night and would like to donate it to the UIXP community for this purpose, please contact us to discuss the requirements in more detail.
  
  
  

Environmental Challenges

The 2021 Facebook ban continues to suppress overall utilization of the Internet and demand for our regional interconnection service. It also increases the cost of Internet service delivery by forcing network operators to import Facebook traffic via international VPNs rather than sourcing it locally from our cache. In general, it makes the Internet more expensive and constricts Uganda’s commercial gravity relative to neighbouring telecommunications markets.

We are also concerned by the high taxes levied on Internet services which now account for over 50% of the total cost of access. Like the Facebook ban, this also makes the Internet more expensive, suppresses overall utilization, and reduces demand for our service.

Financial Challenges

We unfortunately ended the year without any surplus or savings due to an unusually high number of delinquencies. The problem is limited to a minority of networks but the loss of income is significant due to the thin margins of our “cost-recovery plus” sustainability model.

While we are sympathetic to the difficulties that network operators face in our market, it is in everyone’s best interest for everyone to keep their accounts up to date. Income instability impacts our ability to pay suppliers, maintain our infrastructure, and plan for the future.

That aside, our cash reserves remain intact and we remain in good standing with URA despite occasional conflicts caused by erroneous assessments, improper penalties, and onerous bureaucracy.

Community Engagement

UGNOG: The UIXP continues to facilitate donations for the Uganda Network Operators Group (UGNOG) until they can formalize and register a bank account. Any organizations that wish to sponsor UGNOG should notify us by sending an e-mail to accounts@uixp.co.ug CC isabel.odida@gmail.com.

ISPAU & ICTAU: The UIXP is a member of the ISP Association of Uganda (ISPAU) and ICT Association of Uganda (ICTAU) and works through both organizations to promote community development and good policy outcomes for our industry.

AFIX, AfPIF, & AFNOG: Our team remains deeply involved in various voluntary leadership and support roles in the African IXP Association (AFIX), the African Peering and Interconnection Forum (AfPIF), and the African Network Operators Group (AfNOG).

Nationalization in Uganda: A Looming Disaster

The Uganda Communications Commission (UCC) has proposed a new licensing framework that would effectively nationalize a core part of the country's telecommunications industry; Internet exchange points (IXPs).

Introduction to Internet Exchange Points (IXPs)

This news was communicated to us in a letter from the UCC dated June 7th, 2019. The letter included a draft of the licensing framework and a call for written feedback by July 5th, 2019 (click here to download a full copy).

Based on our analysis, the draft framework would establish a government controlled monopoly which all other market players would be subservient to. It would accomplish this with the following formula:

• Establish a “Designated National Internet Exchange Point” that all other IXPs will be required to connect to (9.j);


• Require government approval of contracts between IXPs and network operators (7.4.b);


• Allow the government to arbitrarily compel IXPs to make operational and technical changes (7.5.c)


• Allow the government to inspect, copy, or remove any data related to any IXP without a court order (7.5.b.i);


• Require all licensed network operators to connect to an IXP (8.2.a).

Technically speaking, this policy would merge all IXPs into a single national peering LAN, with each IXP merely acting as a heavily regulated access point for the enlarged infrastructure. The resulting entity would suffer from all of the classic symptoms of a monopoly as well as significant technical challenges and security risks derived from having multiple operators control access points that form part of the same LAN. As a centralized service, it would also inherently lack the resilience that a diverse array of independent IXPs would provide.

Here are two diagrams to help illustrate the concept:

In addition, the draft framework contains language which suggests that the government intends for the "Designated National Internet Exchange Point" to establish itself by expropriating an existing private operation; namely, ours.

This appears to confirm some of our worst fears about Uganda's new National Broadband Policy; a government strategy document, reportedly drafted in isolation, that seemingly calls for a large-scale nationalization and centralization of Uganda's Internet infrastructure under the guise of infrastructure sharing.

We (and others) have repeatedly warned that such policies would have severe socioeconomic consequences for Uganda and the wider East African region. In this particular case, the UCC's planned regulatory intervention in our otherwise healthy industry has no successful parallel anywhere in the world -- and global experts widely regard the other attempts as textbook examples of regulatory failure.

In light of the obvious risks and highly technical nature of this proposal, we feel that this limited survey of the local Internet community is insufficient validation. Accordingly, we strongly urge the UCC to defer any further work on this project until there is a clear rationale and its viability can be transparently proven with case studies and corroborating input from credible global experts.

Meanwhile, we are preparing to submit detailed feedback to the UCC and will post a copy of our submission here once it is ready. We encourage anyone else that would like to submit feedback to do so through us electronically. We will collect, manually submit, and (unless anyone objects) electronically publish all that we receive in order to promote transparency.

Please feel free to contact us here: board@uixp.co.ug

[UPDATE: A copy of our formal feedback to the UCC can be downloaded here. Our general position is that the creation of a de jure IXP monopoly would be bad for our industry, Uganda, and the region. We argue that the regulatory framework should instead seek to create an enabling environment for competition.

We have also uploaded a number of supporting submissions made by the Internet eXchange Federation (IX-F), the African Network Information Centre (AFRINIC), the ICT Association of Uganda (ICTAU), Liquid Telecom, and a personal submission by Diarmuid O'Briain. A copy of those submissions can be downloaded here.]

Liberalization in Uganda: A Looming Regression

On Thursday, April 26th, a concerned network operator forwarded us a copy of a survey they received from the national telecommunications regulator, the Uganda Communications Commission (UCC), which seeks input on how the UIXP should be governed, managed, and financially sustained.

We were not previously aware of this survey and are deeply concerned by its contents. Its text is inaccurate and misleading; its questions are poorly formed; and its broad dissemination indicates that the UCC is planning to attempt a top-down intervention in the UIXP's governance model and, by extension, Uganda's nascent network interconnection industry.

April 2018: Cover letter for UCC survey seeking input on the how the UIXP should be governed, managed, and financially sustained.

The UCC issued this survey with full awareness that the UIXP is actively and transparently working to advance its own governance reform process in collaboration with the network operator community; that the issues we need to address have not adversely impacted our growth and do not present an immediate risk to the continuity of our operations; and that the neutral non-profit governance model which the UIXP adopted in 2001 has become one of the most dominant and successful in Africa.

The UCC's justification for issuing this survey stems from a letter sent to them by two members of our Internet community that libelously accused the UIXP management team, Google, and Akamai of fraud and impropriety. The UCC quickly and conclusively learned that these claims were false but continued to interject in our affairs under the guise of a neutral intermediary intent on resolving a community conflict.

Meanwhile, in the background, we see a rising frequency of controlling and economically damaging government interventions in the telecommunications space. Recent examples include a controversial move to reclaim and prop up the terminally ill state owned telecommunications company; a proposal to limit the number of international gateway providers; an explicit order to tax social media users in order to curtail unfavourable on-line discourse; and a directive to block all online news providers that have not been granted a national license.

The government has also made multiple attempts to nationalize or directly involve themselves in the UIXP's governance in the past, including a 2014 vote by Parliament (which was never implemented) and other less formal/ethical efforts to achieve the same. They also launched a failed attempt to nationalize the .ug ccTLD and have displayed an increasing tendency to censor or block telecommunications services (including mobile money) during presidential elections and politically sensitive events.

In this context, it is hard to trust that the UCC's interest in the UIXP's governance model is benevolent. The nature of their current involvement, and this survey in particular, raises the specter of nationalization and other forms of government intervention that could ultimately deter or prohibit competition in Uganda's emerging network interconnection industry -- an outcome that would have significant long term consequences for Uganda's telecommunications market and national economy.

2014: Parliament adopts an ICT committee recommendation to nationalize the UIXP based on false information and without engaging UIXP management in any way.

The African network interconnection industry is evolving rapidly. There are now 42 IXPs in 32 countries which carry over 400 Gbps of Internet traffic on a daily basis -- up from only 160 Mbps in 2008. South Africa, Nigeria, and Angola all now have more than one IXP operator, with other countries soon to follow. In tandem, we are starting to see the deployment of carrier neutral datacenters, large-scale content, and cloud services.

In Uganda, the UIXP now interconnects 28 networks and carries over 6 Gbps of Internet traffic on a daily basis -- up from only 10 Mbps in 2008. This has made our market much more attractive to international content providers and carrier neutral datacenter investors. If we continue along this path, there will soon be enough demand for local network services to justify the entrance of Uganda's first carrier neutral datacenter and second IXP operator. This would mark a milestone in Uganda's telecommunications history and likely herald a golden era of Internet connectivity.

UIXP: Chart showing connected networks, traffic growth, and key governance events over time.

However, expropriating our private not-for-profit company, and nationalizing our nascent industry in the process, is a great way to make sure that never happens. It would significantly increase Uganda's investment risk profile; prevent new IXPs from forming; and make carrier neutral datacenters -- which rely on network interconnection for growth -- far less viable. As a result, large content providers would need to look elsewhere for hosting (e.g. Kenya) which, in turn, would ensure that Uganda must continue to pay other countries for access, thereby keeping end-user prices high and service quality low.

In our view this is a very real possibility that threatens everyone's interests. We therefore call on all network operators and other recipients of the UCC survey to take this into account when considering if, and how, to respond.

We further call on all network operators to more actively participate in our own sustainability and governance reform process. It should be clear to all by now that the only good way forward is to work together.