XiaoLi Network is visible, but visibility is not the same as commercial proof
XiaoLi Network is a useful case because it is not invisible. The public internet can see enough to know that something real is being routed. PeeringDB lists XiaoLi Network at https://www.peeringdb.com/net/11809 with ASN 201705, AKA XLI-NET, organisation Xinrun Li, network type "Educational/Research", an open peering policy, no IPv4 prefixes, forty IPv6 prefixes, and a reported traffic range of 1-5Gbps. RIPE RDAP for AS201705 at https://rdap.db.ripe.net/autnum/201705 shows an active autonomous system registered on 2026-02-04, tied to ORG-XL124-RIPE, with the name XLI-NET, a Guangzhou address, a maintainer, and a role contact called XiaoLi-NOC. BGP.tools at https://bgp.tools/as/201705 sees active IPv6 routing, community peers and several upstream paths. Hurricane Electric's BGP view at https://bgp.he.net/AS201705 reports only IPv6 prefixes, all observed originated prefixes covered by valid RPKI origin validation at the time of review, and eight internet exchange entries.
That is enough to make XiaoLi Network more than a domain name, a forum handle or a placeholder page. It has a public routing identity. It has contact points. It has a route record. It has a stated peering posture. It appears in the public databases that network operators use when deciding whether to treat a small network as real.
The harder question is what kind of reality the evidence supports. The PeeringDB network type is not "Cable/DSL/ISP" or "Content"; it is "Educational/Research". The RIPE organisation record at https://rest.db.ripe.net/ripe/organisation/ORG-XL124-RIPE.json names Xinrun Li, gives the organisation type as "OTHER", and states that the registration number is "Not Applicable". The same public record shows a sponsoring organisation, Cylix Pte. Ltd., rather than an ordinary China-based membership route into the registry. IPinfo at https://ipinfo.io/AS201705 classifies the network as hosting/cloud, shows zero IPv4 addresses, zero hosted domains, zero downstreams, and a handful of peers and upstreams. These are not faults. They are the public shape of a young, IPv6-only, sponsor-backed, technically active small network. But they do not by themselves prove that XiaoLi Network is a commercial regional ISP in the sense a mainland Chinese enterprise buyer, access customer, regulator, or transit counterparty would usually mean.
That distinction matters because the economic value in a small Chinese network is not merely the ability to announce a prefix. Announcing a prefix can now be achieved with an ASN, sponsored resources, a virtual server, tunnelled sessions, route servers and a set of patient upstreams. The scarce asset is being believed. A buyer has to believe that the network operator has the right to sell the service being described, can keep the route visible, can deal with abuse, can maintain paths through China Telecom, China Unicom, China Mobile and overseas carriers, and can support the customer when something breaks at night. A peer has to believe that the traffic exchange is worth maintaining. A regulator has to believe that the activity stays inside the relevant licensing and filing boundaries. A source of capital has to believe that the network is not just a routing lab dressed as a business.
XiaoLi Network therefore sits in the most interesting part of the market map: the zone where internet evidence is abundant but commercial evidence is thin. It has enough public routing substance to deserve attention. It also has enough ambiguity that the burden of proof becomes the story.
The identity trail points to a person-led network, not a conventional ISP brand
The most coherent identity trail begins with PeeringDB and RIPE rather than with a corporate website. PeeringDB's network page lists the organisation as Xinrun Li and gives the website as https://blog.vh.gs. The machine-readable PeeringDB organisation record at https://www.peeringdb.com/api/org/43287 also names Xinrun Li, country China, and connects the organisation to the XiaoLi Network entry. RIPE's REST record for the autonomous system at https://rest.db.ripe.net/ripe/aut-num/AS201705.json uses AS name XLI-NET, references ORG-XL124-RIPE, includes the sponsoring organisation ORG-CPL37-RIPE, and lists import and export policies with several other small networks. The RIPE role record at https://rest.db.ripe.net/ripe/role/XA565-RIPE.json gives a NOC role, a noc@vh.gs contact, and an abuse mailbox at abuse@vh.gs.
That is a usable public identity for network operations. It is also a notably personal identity. The linked web presence, https://blog.vh.gs/, is a Chinese personal blog titled lbyxiaolizi's blog, with categories covering VPS, BGP and other technical topics. The about page at https://blog.vh.gs/about.html reads like a long-running personal technical blog, not a carrier storefront. It describes blog history, hosting changes, overseas VPS usage and university context. The associated GitHub profile at https://github.com/lbyxiaolizi points to the same blog and shows the user identity lbyxiaolizi, a location or affiliation string of Sun Yat-sen University, and open-source activity.
For a small network this personal trail is not disqualifying. A large share of the serious BGP hobbyist, university-adjacent and small-transit world begins this way. The first signals are often a personal blog, a contact mailbox, a public maintainer, a few route records, a looking glass, and a PeeringDB page. The problem is that these same signals can support several very different interpretations. They can describe a research network, a personal lab, an early-stage hosting provider, a community peering project, a student project, an IPv6 learning exercise, a tiny transit resale venture, or a regional ISP in formation. The public record does not choose among those interpretations for the reader.
The ASN application post on the blog makes that ambiguity more valuable as evidence. In "BGP Player's first step - applying for an ASN" at https://blog.vh.gs/BGP/ASN.html, the author discusses the cost and process of obtaining an ASN, compares APNIC and RIPE paths, describes APNIC as the natural regional registry but relatively more expensive for the intended use case, and explains the attraction of RIPE-sponsored IPv6 resources. The post is not written like a corporate launch announcement. It is written like a technically literate operator entering the BGP world and documenting the economics of doing so. The tone is important: it supports competence, curiosity and cost awareness, but it does not support a claim that a regulated Chinese access business already exists.
That is the first credibility lesson. For XiaoLi Network, the public identity is strongest as a person-led network operations identity. It is weaker as a standalone commercial brand. If the business wants the market to read it as a regional ISP, the missing evidence is not another route record. It is a public bridge between the routed identity and a legally accountable service proposition: who is the contracting party, what service is being sold, where it is allowed to operate, how customers reach support, what infrastructure backs the service, and which parts are experimental.
ICP and licensing clues matter because China makes internet proof jurisdictional
China changes the credibility test for small networks because web presence, access service and hosting service are not just marketing questions. They are jurisdictional clues. The public ICP filing system at https://beian.miit.gov.cn/ exists because Chinese internet information services that operate within the mainland filing framework are expected to be traceable. The Ministry of Industry and Information Technology recordal rules, available through the Cyberspace Administration repost at https://www.cac.gov.cn/2005-02/09/c_1112147171.htm, say that non-commercial internet information services provided within the PRC territory must complete filing before providing the service, and that providers should display the filing number on the homepage with a link to the filing system. The same rules place obligations on access providers not to provide access to unfiled non-commercial services.
This does not mean every Chinese person-led technical blog must look like a mainland corporate website. The world is full of Chinese-language sites hosted overseas, using non-mainland domains, overseas DNS and overseas mail. XiaoLi's visible web footprint fits that pattern. The public blog uses vh.gs, a .gs domain. The site itself does not present, on the visible pages reviewed, a commercial ISP order flow, a mainland access offer, a published ICP footer, a telecom licence number, a product catalogue, a service-level agreement, a public status page or a customer support process. Host.io's domain view for https://host.io/vh.gs points the apex domain at Cloudflare infrastructure and Microsoft mail protection, while a direct public DNS lookup during review placed blog.vh.gs on an address outside AS201705. Those details are not a criticism of the blog. They simply show that the public website is not itself proof that XiaoLi Network is serving customers from its own routed footprint.
The licensing side is even more important. A Beijing government guide to telecom business licence renewal at https://banshi.beijing.gov.cn/pubtask/task/1/110000000000/f6364769-0bfd-42a3-98f7-9cf6e9f37213.html?locationCode=110000000000&serverType=1002 summarises core conditions from the telecom business licensing framework: the operator must be a legally established company, have funding and personnel suited to the business, have credibility or ability for long-term service, have necessary sites, facilities and technical plans, and not be listed for serious telecom dishonesty. MIIT's telecom business market platform at https://tsm.miit.gov.cn/dxxzsp/ is the public-facing environment for licence-related information and notices. MIIT's 2017 clean-up notice, available through the State Council Information Office at https://www.scio.gov.cn/xwfb/gwyxwbgsxwfbh/wqfbh_2284/2017n_8019/2017n07y25r/wjxgzc_8498/202207/t20220715_202287.html, explicitly targeted unlicensed operation, operation outside licensed scope and transfer or lease of telecom licences in the internet access market.
Against that context, XiaoLi Network's public record supports a conservative reading. It shows a real network identity and a personal technical operation. It does not show a mainland telecom service licence, a public ICP-linked service site, or a legal company registration connected to a commercial ISP offer. Absence from open sources is not proof that such documents do not exist. Some operators have private contracts, limited-scope services, or names that do not match their public network brand. But the visible record should discipline the market's language. The strongest public claim is that XiaoLi Network is a China-associated, person-led, IPv6-only educational or research network with real routing presence. A stronger commercial ISP claim would require separate evidence.
This is why ICP and licence clues are economic signals, not bureaucratic trivia. In China, compliance evidence affects customer trust. It affects whether a buyer can explain the service internally. It affects whether a host can connect a domain, whether a service can be sold to mainland customers, whether a carrier partner can justify a path, and whether a provider can survive scrutiny if traffic patterns become sensitive. For a small network, the cost of producing this proof can exceed the cost of the route itself.
The routing record is real, but it is narrower than the word "ISP" implies
The strongest technical evidence for XiaoLi Network is the routing record. AS201705 is not merely a name in one database. RIPE RDAP and RIPE REST show the registration. PeeringDB shows the network profile and exchange points. BGP.tools observes active routing and peers. BGP.he.net observes IPv6 prefixes and exchange presence. IPIP.NET at https://whois.ipip.net/AS201705 mirrors key registry details and describes AS201705 as XLI-NET - Xinrun Li, registry RIPE, country China, with no IPv4 and a small number of IPv6 prefixes. IPinfo at https://ipinfo.io/AS201705 supplies another independent view: no IPv4, a very large IPv6 address count by address-space mathematics, zero hosted domains, one pingable IPv6 address in its public probe set, no downstreams and several upstreams.
The record is especially consistent on one point: this is an IPv6-only public footprint. PeeringDB says IPv4 prefixes are zero. BGP.he.net says the same. IPinfo says zero IPv4 addresses. BGP.tools tags the network as IPv6 only. That matters because IPv6-only is both technically forward-looking and commercially limiting. In a lab, a research network, a community peering project or a transit experiment, IPv6-only can be completely rational. In a mass access or hosting business, IPv4 scarcity still shapes customer expectations, support tickets, content reachability and commercial packaging. A customer can admire IPv6 hygiene and still ask where the IPv4 plan is.
The route records also show dependence on sponsored and delegated resources rather than a large native allocation. RIPE's organisation record for ORG-CPL37-RIPE at https://rest.db.ripe.net/ripe/organisation/ORG-CPL37-RIPE.json identifies Cylix Pte. Ltd. as the sponsoring LIR. The RIPE search view for 2a14:7dc0::/29 at https://rest.db.ripe.net/search.json?query-string=2a14:7dc0::/29&source=ripe&type-filter=inet6num shows a Singapore-registered allocation associated with Cylix and a geofeed URL. A separate RIPE search for AS201705 route6 entries at https://rest.db.ripe.net/search.json?query-string=AS201705&source=ripe&inverse-attribute=origin&type-filter=route6 returns several IPv6 route records using AS201705 as origin, including 2a14:7dc0:510::/48, 2a14:7dc0:511::/48, 2a14:7dc0:512::/48, 2a14:7dc0:514::/48, 2a14:7dc0:515::/48, 2a14:7dc0:516::/48 and 2a14:7dc0:517::/48, along with other records visible in the search result. That is a visible operational footprint, but it is not the same as a deep physical access footprint in China.
RPKI is a point in XiaoLi Network's favour. BGP.he.net's AS page showed originated prefixes with valid RPKI status and no invalids at review time. Cloudflare Radar's routing and RPKI pages for AS201705, including https://radar.cloudflare.com/routing/rpki/as201705, provide another public place where the network's routing and origin-validation posture can be inspected. For a small network, having valid route origin authorisations is a credibility asset. It says the operator is not merely throwing prefixes into BGP; there is some attention to routing hygiene.
Yet the route record remains narrow. BGP.tools saw a small number of active IPv6 prefixes and identified the network as personal and IPv6 only. IPinfo showed zero downstreams. BGP.he.net observed twenty-nine IPv6 peers, while BGP.tools saw a larger community of known peers; the exact count differs because public monitors see different parts of the internet at different times. The important point is not which counter is "right". The important point is that all of them describe a small, young network whose public importance comes from being seen in the routing system, not from showing a base of customers, a large access plant, a content platform or a visible hosting estate.
For a network like XiaoLi, route visibility is a cost centre and a marketing signal at once. It proves technical competence. It lets the operator join public peering fabrics, test policy, learn failure modes, and appear in the same databases as much larger networks. But every new public trace invites a commercial question: what does the trace stand for? If the answer is education and research, the record is enough to be interesting. If the answer is regional ISP, the record is only the opening paragraph.
Internet exchanges can buy presence, but not the trust normally attached to facilities
PeeringDB's most impressive line for XiaoLi Network is its exchange presence. The network page and the netixlan API at https://www.peeringdb.com/api/netixlan?asn=201705 show eight public exchange entries: Better CNIX, BGP.Exchange Amsterdam, BGP.Exchange Hong Kong, BGP.Exchange Singapore, BGP.Exchange Tokyo, Protocol 7 Hong Kong, Protocol 7 Tokyo and CXIX Great Lakes. Several of these entries are 1G ports, with 10G entries at Protocol 7 Hong Kong and CXIX Great Lakes. The geography looks broad: China, Hong Kong, Singapore, Tokyo, Amsterdam and North America.
The commercial interpretation needs care. In the modern small-network world, a wide PeeringDB map does not necessarily mean owned routers in all those places, colocation contracts in all those facilities, staff in each market or direct customer access in each country. BGP.Exchange and Protocol 7 are explicitly attractive to small networks because they lower the cost of peering visibility. They support virtual or tunnelled exchange participation and route-server learning in ways that let hobbyists, students, research networks and small providers appear in multiple locations without building a traditional carrier estate. CXIX and other community exchange fabrics play a similar role in making peering more accessible.
That accessibility is good for the internet. It lets technically serious operators learn by doing. It gives IPv6-only networks a way to discover route policy, failover, filtering, communities and abuse workflows. It also creates a new proof problem. The map can look larger than the business. A PeeringDB exchange entry proves that the network wants and may be able to exchange traffic at that fabric. It does not prove customer density, a local sales base, paid transit revenue, owned access facilities, Chinese carrier interconnection, or the ability to meet a customer's latency and support expectations.
The Better CNIX entry is particularly interesting because it is the most China-shaped exchange signal in the public record. PeeringDB lists a Better CNIX connection with an IPv6 address and 1G speed. That matters more than a generic overseas tunnel because the target lens is China. Still, one exchange entry does not settle mainland operating status. A serious commercial buyer would want to know where the port is physically or virtually delivered, which upstreams carry traffic into the mainland, what the contractual chain looks like, and whether the service claim is access, transit, hosting, research, community peering or something else.
This is the second credibility lesson. Internet exchanges can make a small network visible faster than the business can become trustworthy. That speed is powerful. It allows a network like XiaoLi to build a public BGP resume in months. It also separates visible routing from durable commercial operations. If route presence is the main asset, then the operator must explain the difference between peering reach and service capability before someone else explains it less generously.
The blog reinforces this reading. The Tailscale tunnel post at https://blog.vh.gs/BGP/tailscale-tunnel.html discusses how to use Tailscale and WireGuard-like tunnelling to connect servers near BGP sessions and announce addresses. It is a practical technical note, not a sales page. It shows the operator understands the tooling small networks use to stretch their control plane across cheap infrastructure. It also reminds the reader that in this tier of the market, the visible BGP edge may be assembled through rented servers, tunnels and upstream goodwill. That can be a legitimate engineering choice for a lab or research network. It is not the same proof as a carrier facility.
Carrier dependency is the real Chinese bottleneck
For a small network associated with China, the hardest market test is not whether the ASN appears in RIPE. It is how traffic behaves when the route touches the Chinese access world. China Telecom, China Unicom and China Mobile dominate the user access environment. International paths are shaped by commercial interconnects, regulatory boundaries, congested or premium routes, and a long local vocabulary of route quality: CN2, CTG, 9929, 10099, CMI, SoftBank transit, Japan and Hong Kong handoff, and the difference between a route that works on a test and a route that survives evening traffic.
The operator behind XiaoLi Network appears to understand that vocabulary. The VPS review at https://blog.vh.gs/vps/evoxt-my-premium.html discusses Malaysian VPS routing and names CTG GIA, CN2 GIA, China Unicom's 10099 and 9929 paths, China Mobile via SoftBank and Japan, and performance from mainland carriers. A NodeSeek search snippet for lbyxiaolizi comments also shows route-path literacy, including a correction about China Unicom paths. These are not verified service claims for XiaoLi Network. They are useful market intelligence about the operator's knowledge environment. The blog is written by someone who thinks about Chinese carrier routes in the way Chinese hosting and VPS buyers think about them.
That matters because Chinese small-network credibility is carrier credibility in disguise. A customer buying from a tiny network does not usually care that a route record exists somewhere in RIPE. The customer cares whether users on China Telecom in Guangdong, China Unicom in Shandong, China Mobile in Jiangsu, or a university network in Beijing can reach the service predictably. The customer cares whether an overseas path collapses during evening congestion. The customer cares whether an abuse incident causes a cheap upstream to withdraw service. The customer cares whether the operator can explain packet loss without blaming the internet in general.
XiaoLi Network's public record does not yet show direct carrier depth. IPinfo lists upstreams such as MoeDove, WJY, YunZheng, Johannes Ernst and Cyberverse in its public view. BGP.tools similarly shows a small-network upstream set, including AS216211 Cyberverse, AS212895 Johannes Ernst, AS204921 YunZheng and AS44324 MoeDove. These names may be perfectly reasonable for a small IPv6 research network. They are not the same as public direct carrier arrangements with the Chinese access networks that define mainland delivery. The result is a dependency stack: XiaoLi depends on upstreams and exchanges; those upstreams depend on other transit; the end user depends on the major Chinese access networks; and the buyer depends on XiaoLi's ability to manage a chain it does not fully control.
Every small network faces dependency. The Chinese version is more severe because the market is huge and carrier-controlled. Xinhua's summary of the 2025 communications industry statistical bulletin at https://app.xinhuanet.com/news/article.html?articleId=4052292841fb8ebb8dd8c96189aa4dc6 reports telecom business revenue of about RMB 1.75 trillion, emerging businesses including cloud computing and data centres at RMB 450.8 billion, and 691 million fixed broadband users across China Telecom, China Mobile and China Unicom. Xinhua's report on the 56th CNNIC Statistical Report at https://www.xinhuanet.com/government/20250722/fff6ad77dce44fcbbd5fff2668031284/c.html put China's internet user base at 1.123 billion by June 2025, with 79.7% penetration and 4.55 million 5G base stations. In a market that large, a small IPv6 network cannot credibly sell scale. It has to sell specificity: a research purpose, a niche peering use case, a technical community, a special route, a learning environment or a trusted operator contact.
This is where XiaoLi Network's proof burden becomes economic rather than merely technical. It does not need to prove it is China Telecom. It does need to prove what kind of dependency a counterparty is accepting. If the service is a lab, say so. If it is transit, show the upstream chain and route policy. If it is hosting, show the facility, licence posture and support model. If it is community peering, show the rules. The carrier bottleneck cannot be wished away by an ASN.
Hobby-network signals are not commercial proof, but they are not noise
The public record around XiaoLi Network contains several signals that would be easy to dismiss as hobbyist chatter. That would be a mistake. In small-network economics, community signals are part of the credibility market. They show whether the operator understands the game, whether peers have seen the handle before, whether the person asks practical questions, whether the claimed network identity matches the surrounding behaviour, and whether the operation is likely to respond when something breaks.
The most important hobby signal is the blog's ASN post. It openly frames the process as a "BGP player" first step. It explains why an Asia-based operator might look at APNIC and then choose a RIPE-sponsored path instead. APNIC's own fee schedule at https://www.apnic.net/about-apnic/corporate-documents/documents/membership/member-fee-schedule/ and its membership cost page at https://www.apnic.net/get-ip/apnic-membership/how-much-does-it-cost/ show why cost matters: the sign-up and member fee structure is meaningful for a student, hobbyist or tiny provider. RIPE's 2026 charging scheme at https://www.ripe.net/publications/docs/ripe-848/ lists LIR contributions and separate charges for independent resources and ASN assignment. Sponsored access can therefore be the rational low-cost route into the global routing table.
That cost logic is not evidence of weakness. It is evidence of market segmentation. Large carriers buy regulatory certainty, access facilities, staff, IPv4 holdings, support operations and interconnect contracts. Small networks buy just enough registry and routing presence to do something specific. They optimise for learning, visibility, control and affordability. XiaoLi's public writing fits that second market.
The LowEndTalk comment at https://lowendtalk.com/discussion/218509/deluxhost-net-summer-offer-f-z-x-new-q-series-intel-amd-vps-double-bonus-ned is another useful but limited signal. In the thread, a user under the same public handle asks whether a BGP session can be set up after waiting four months. That is not proof of a XiaoLi customer service failure, and it should not be read as a verified commercial claim about XiaoLi Network. It is useful because it shows the operational reality of the small BGP market: cheap hosting offers, ticket queues, delayed BGP sessions, dependence on providers, and the patience needed to assemble a public routing footprint.
The GitHub profile is similarly modest evidence. It does not show a telecom company. It shows a technical person with repositories, public identity and a link to the same blog. The blog's technical posts show a learning trail. Forum snippets show route literacy. PeeringDB shows participation. Together, these signals make XiaoLi Network more credible as a real small-network operator than a bare database entry would be. They also make it less credible as an already mature commercial regional ISP, because the public posture is personal, experimental and community-led.
This distinction is important for readers who buy or peer with small networks. Hobbyist does not mean unserious. Many serious network engineers began as hobbyists. Some personal networks become resilient community infrastructure. Some small ASNs are run more carefully than larger commercial networks. But hobby signals change the kind of proof one should demand. The correct question is not "is this fake?" The correct question is "what claim is this evidence strong enough to support?"
For XiaoLi Network, the evidence supports technical presence, route learning, IPv6 operation and community visibility. It does not yet support customer scale, licensed mainland access operations, physical network depth, or a conventional regional ISP business.
Customer trust depends on evidence XiaoLi has not yet made public
The customer trust gap is the largest commercial gap in the public record. There is no obvious public product page for XiaoLi Network. There is no price list. There is no status page. There are no visible customer case studies. There is no public statement of service scope. There is no facility list beyond exchange points. There is no support process beyond NOC and abuse contact records. IPinfo's page showed zero hosted domains. PeeringDB showed no facility count even while listing exchange points. That profile can be acceptable for an educational or research network. It is a problem if the market is expected to infer a commercial access or hosting service.
Trust in small networks is built by reducing unknowns. A customer wants to know whether the operator controls the prefix, whether RPKI is maintained, whether route filters are sane, whether upstreams are diverse, whether abuse reports are answered, whether maintenance is communicated, whether the operator can survive a payment dispute with an upstream, and whether the service depends on a single VPS in a single city. A Chinese customer adds more questions: what is the ICP posture, what licence scope applies, whether the server or service is inside or outside mainland China, whether cross-border connectivity is legally and operationally stable, and whether the route is usable from the carrier that matters to that customer.
XiaoLi Network answers some of these questions indirectly. RIPE and PeeringDB show contactability. RPKI status appears clean in BGP.he.net's view. PeeringDB shows open peering. The blog demonstrates technical interest. Route records show real work. But these answers are still indirect. The public record is strongest for operators who already know how to read routing databases. It is much weaker for customers who need a commercial promise in ordinary language.
That weakness has economic consequences. The network may have to discount more heavily. It may have to rely on friends, forum reputation or community peers rather than mainstream buyers. It may have to spend time proving facts that larger providers can outsource to brand and licence recognition. It may lose deals not because the route is bad but because the buyer cannot document the risk. The proof burden becomes an operating cost.
There is a second economic consequence: the network's own best market may not be the market implied by the category label. If XiaoLi Network remains an educational or research network, the right audience may be peers, students, IPv6 labs, tunnel users, route-policy learners and small operators who value competence and transparency over formal procurement. If it wants to act as a regional ISP, the audience changes. The proof package must become more formal, less personal and more compliance-aware. It would need service descriptions, licence and filing clarity, an entity able to contract, customer terms, a network map that distinguishes physical facilities from virtual exchange sessions, and a support promise.
The point is not that XiaoLi Network must become corporate to be valuable. The point is that different kinds of value require different kinds of evidence. A small research ASN can be excellent without a sales page. A commercial ISP cannot ask the market to infer the sales page from BGP.
The market context rewards small technical networks but punishes vague claims
China's internet market gives small networks a strange opportunity. On one hand, the main access market is overwhelmingly shaped by large state-linked carriers and national infrastructure plans. Xinhua's MIIT statistics show hundreds of millions of broadband users, huge telecom revenue and accelerating cloud and data-centre business. CNNIC's reports show enormous domain, IPv6 and mobile internet scale. No tiny network can compete with that through size.
On the other hand, the same scale creates niche demand. Developers, gamers, VPS buyers, students, exporters, content operators, cross-border teams and network hobbyists all care about path quality, latency, packet loss, route stability and special IPv6 behaviour. A tiny network can matter if it solves one specific problem, creates a trusted community, offers a transparent lab, or provides a route that larger providers do not prioritise.
XiaoLi Network's public evidence is closer to that niche logic than to mass access. It is IPv6-only. It is present on community-oriented exchange fabrics. It is described as educational or research. Its linked blog talks about VPS routing, ASN registration and tunnels. It has no public downstream base. These are not the attributes of a carrier trying to sign millions of broadband users. They are the attributes of a small network building credibility among people who already understand why a personal ASN can be interesting.
The danger is category inflation. In public databases, every ASN can start to look like a network company. A page label can make a tiny research network appear comparable to a regional ISP. A broad exchange map can make a tunnelled setup look like a physical footprint. A Guangzhou address can make overseas resources look like mainland infrastructure. A few valid RPKI records can make the entire operation look mature. None of those readings is necessarily malicious; they are common errors in interpreting internet evidence.
For investors, partners and peers, XiaoLi Network should be read through a more precise lens. It appears to be a technically active, China-linked, person-led IPv6 network with a RIPE-sponsored ASN and community exchange visibility. Its commercial credibility depends on whether it can produce evidence beyond that: customers, service scope, compliance posture, carrier paths, uptime history, support behaviour and contractable identity. Without that evidence, the safe interpretation is small-network capability, not proven ISP business.
This is also why non-official signals must be handled carefully. The blog and forums are valuable because they reveal the operator's vocabulary, incentives and learning path. They should not be converted into verified claims of service quality or customer scale. A forum complaint about waiting for a BGP session is evidence of the surrounding market's friction, not evidence that XiaoLi fails to deliver. A VPS route review is evidence of carrier literacy, not evidence that XiaoLi controls those routes. A GitHub profile is evidence of a technical person, not evidence of corporate capacity.
In a low-information market, the temptation is to over-read every clue. The disciplined view is more useful: accept that XiaoLi Network is real where the evidence is strong, and keep the commercial claims narrow where the evidence is weak.
What would change the view
The view on XiaoLi Network would improve quickly if the public proof package changed. A current, clearly linked service website would matter. So would a visible ICP filing for a mainland-facing web service, if the service is in scope for filing, or a clear statement that the service is overseas and not offered as mainland access. A telecom business licence or licence-linked partner structure would matter if XiaoLi is selling access, IDC, VPN, CDN, cloud or hosting services that fall inside regulated categories. A company registration tied to the service brand would matter. A status page, looking glass, route policy document, maintenance history and abuse-response process would all help.
Network evidence could also change the view. Direct or well-documented upstream contracts, carrier arrangements or reputable transit providers would make the service proposition stronger. Public route stability over several quarters would matter more than a burst of new route records. RPKI should remain valid. ASPA, where supported, would be another route-security signal. A clear distinction between physical locations, virtual exchange participation and tunnelled sessions would reduce confusion. Public latency and reachability data from Chinese carriers would be useful if presented carefully and kept current.
Commercial evidence would be decisive. Named customers, case studies, public terms of service, support commitments and clear product boundaries would transform the interpretation. The market does not need a small network to become large before trusting it. It needs to know what it is buying. A small IPv6 research network can be trusted as a research network. A regional ISP can be trusted only when it supplies the evidence a regional ISP normally supplies.
The negative facts would also matter. Stale RIPE records, invalid RPKI, repeated withdrawal of exchange presence, abandoned contact mailboxes, unresolved abuse reports, uncorrected licence or ICP claims, route leaks, unexplained customer complaints, or a blog and PeeringDB page that stop being maintained would all weaken credibility. For small networks, neglect is often more damaging than size. A tiny network that updates records, answers mail and states its limits can be more credible than a larger-looking network that leaves its public evidence to decay.
The current evidence therefore supports a middle position. XiaoLi Network is not a ghost. It is visible in the routing system, has a coherent personal identity trail, maintains public network records, appears to understand Chinese route economics, and has enough RPKI and exchange evidence to be taken seriously as a small IPv6 network. But the public evidence does not yet justify reading it as a proven Chinese regional ISP business. The missing asset is proof: proof of service scope, proof of compliance posture, proof of carrier dependency, proof of customers, and proof that the public routing footprint corresponds to a durable economic function.
That is the lesson beyond XiaoLi Network. In the Chinese small-network fringe, the cost of being believed may be higher than the cost of being routed.
Public evidence register
PeeringDB's XiaoLi Network page at https://www.peeringdb.com/net/11809 and its API records at https://www.peeringdb.com/api/net/11809 and https://www.peeringdb.com/api/netixlan?asn=201705 support the core identity, ASN, network type, open peering policy, IPv6-only profile and exchange presence. The PeeringDB organisation API at https://www.peeringdb.com/api/org/43287 ties the network entry to Xinrun Li.
RIPE's RDAP page at https://rdap.db.ripe.net/autnum/201705, REST aut-num record at https://rest.db.ripe.net/ripe/aut-num/AS201705.json, organisation record at https://rest.db.ripe.net/ripe/organisation/ORG-XL124-RIPE.json, and role record at https://rest.db.ripe.net/ripe/role/XA565-RIPE.json support the AS name XLI-NET, registration timing, Guangzhou address, contacts, maintainer and abuse mailbox. The Cylix organisation record at https://rest.db.ripe.net/ripe/organisation/ORG-CPL37-RIPE.json supports the sponsored-resource reading. RIPE search results for 2a14:7dc0::/29 and AS201705 route6 records support the IPv6 resource and route-record analysis.
BGP.tools at https://bgp.tools/as/201705, Hurricane Electric at https://bgp.he.net/AS201705, IPinfo at https://ipinfo.io/AS201705, IPIP.NET at https://whois.ipip.net/AS201705, Cloudflare Radar at https://radar.cloudflare.com/traffic/as201705 and Cloudflare Radar RPKI at https://radar.cloudflare.com/routing/rpki/as201705 support the live-routing, IPv6-only, peer, upstream and RPKI discussion. These public monitors differ in counts because they observe from different vantage points; the article uses the shared direction of the evidence rather than treating any single counter as complete.
The official or operator-linked web trail comes from https://blog.vh.gs/, the about page at https://blog.vh.gs/about.html, the ASN post at https://blog.vh.gs/BGP/ASN.html, the tunnel post at https://blog.vh.gs/BGP/tailscale-tunnel.html, the VPS route review at https://blog.vh.gs/vps/evoxt-my-premium.html, and the GitHub profile at https://github.com/lbyxiaolizi. These support the interpretation of a person-led, technically active, BGP-aware operation. The LowEndTalk thread at https://lowendtalk.com/discussion/218509/deluxhost-net-summer-offer-f-z-x-new-q-series-intel-amd-vps-double-bonus-ned is used only as community-market context for BGP-session procurement friction.
The China regulatory and market frame comes from the ICP filing portal at https://beian.miit.gov.cn/, the MIIT recordal rules reposted at https://www.cac.gov.cn/2005-02/09/c_1112147171.htm, the Beijing telecom licence renewal guide at https://banshi.beijing.gov.cn/pubtask/task/1/110000000000/f6364769-0bfd-42a3-98f7-9cf6e9f37213.html?locationCode=110000000000&serverType=1002, MIIT's telecom business market platform at https://tsm.miit.gov.cn/dxxzsp/, the 2017 MIIT clean-up notice at https://www.scio.gov.cn/xwfb/gwyxwbgsxwfbh/wqfbh_2284/2017n_8019/2017n07y25r/wjxgzc_8498/202207/t20220715_202287.html, Xinhua's 2025 communications-industry summary at https://app.xinhuanet.com/news/article.html?articleId=4052292841fb8ebb8dd8c96189aa4dc6, and Xinhua's CNNIC 56th report summary at https://www.xinhuanet.com/government/20250722/fff6ad77dce44fcbbd5fff2668031284/c.html.
APNIC's fee schedule at https://www.apnic.net/about-apnic/corporate-documents/documents/membership/member-fee-schedule/, APNIC's membership cost explainer at https://www.apnic.net/get-ip/apnic-membership/how-much-does-it-cost/, and RIPE's 2026 charging scheme at https://www.ripe.net/publications/docs/ripe-848/ support the article's discussion of why small networks optimise for sponsored, low-cost routes into public routing visibility.