llms.txt
Compare with other sites' llms.txt files →# Routing Lab 路由实验室
> A free, browser-based network lab built by network engineers for the people who work around networks — product managers, testers, software engineers and students — so you can see the whole picture from layer 1 to layer 7 by actually running the protocols: MPLS, SR-TE, BGP, OSPF, IS-IS, EIGRP, RIP, VLAN, STP, DNS, load balancing, VPN, TLS, RDMA/InfiniBand, spine-leaf, OTN/ODF optics and IDC topology. 21 modules, English and Chinese, nothing to install. 一群网络工程师做的免费实验室:让网络产品经理、测试、软件工程师和在校学生在浏览器里把协议真的跑起来,从一层到七层看清整张网。覆盖 MPLS、SR-TE、BGP、OSPF、IS-IS、EIGRP、RIP、VLAN、STP、DNS、负载均衡、VPN、TLS、RDMA/InfiniBand、Spine-Leaf、OTN/ODF 光网络与 IDC 拓扑。21 个模块,中英双语,不用安装。
## Labs 实验
- [MPLS Label Switching · MPLS 标签交换](https://ipnetmaster.com/mpls): How an MPLS label switched path is signalled and how packets are forwarded by label instead of by a destination lookup at every hop. Watch the label stack get pushed, swapped and popped hop by hop, including penultimate hop popping. MPLS 的标签交换路径是怎么建立的、报文怎么按标签转发而不用每跳查路由表:逐跳看标签栈的压入、交换、弹出,以及次末跳弹出(PHP)。
- Push / swap / pop: what each LSR does to the label stack, and why the transit router never looks at the IP header. 压入/交换/弹出:每台 LSR 对标签栈做了什么,以及中间路由器为什么根本不看 IP 头。
- Penultimate hop popping — why the last-but-one router removes the label instead of the egress. 次末跳弹出:为什么是倒数第二跳而不是出口路由器摘掉标签。
- FEC and the label forwarding table (LFIB): how a packet is classified once at the ingress. FEC 与标签转发表(LFIB):报文为什么只在入口分类一次。
- [SR-TE Segment Routing · SR-TE 段路由](https://ipnetmaster.com/srte): Segment Routing traffic engineering: the path is carried in the packet as a label stack, so there is no per-hop signalling state to maintain. Compare an SR-TE explicit path against plain IGP shortest path and see which links actually carry the traffic. 段路由流量工程:路径以标签栈的形式写在报文里,中间节点不用维护每条路径的状态。把 SR-TE 显式路径和 IGP 最短路对照着看,流量到底走了哪几条链路。
- Prefix SID vs adjacency SID: the two building blocks of every SR path. 前缀 SID 与邻接 SID:所有 SR 路径都由这两种段拼出来。
- Why SR-TE needs no RSVP-TE state in the core, and what that costs you at the head end. SR-TE 为什么不需要核心保存 RSVP-TE 状态,代价又落在头端的哪里。
- Steering traffic off the shortest path on purpose — and watching the label stack grow as you do it. 刻意把流量从最短路上引开,同时看着标签栈一层层变长。
- [eBGP · eBGP 跨自治系统路由](https://ipnetmaster.com/ebgp): How autonomous systems interconnect with external BGP: session establishment, route advertisement, and how AS_PATH prevents loops between providers. Includes the attribute decision process that picks the best path. 自治系统之间怎么用 eBGP 互通:会话怎么建立、路由怎么通告、AS_PATH 怎么在运营商之间防环,以及选路时属性的比较顺序。
- AS_PATH loop prevention: an AS drops any route that already contains its own number. AS_PATH 防环:路由里只要出现自己的 AS 号就直接丢弃。
- The best-path decision order (weight, local pref, AS_PATH length, origin, MED …) applied to real routes. 选路顺序(weight、local pref、AS_PATH 长度、origin、MED…)在真实路由上逐条走一遍。
- Why an eBGP peer is normally one hop away, and what changes when it is not. eBGP 邻居为什么通常直连一跳,不直连时又要改什么。
- [iBGP · iBGP 自治系统内部路由](https://ipnetmaster.com/ibgp): Carrying externally learned routes inside an AS: why iBGP does not re-advertise routes between peers, why that forces either a full mesh or a route reflector, and how next-hop-self and the IGP underlay fit together. 外部路由怎么在 AS 内部传递:iBGP 为什么不在邻居之间转发路由、由此为什么必须全互联或者上路由反射器,以及 next-hop-self 和底层 IGP 的配合。
- The iBGP split-horizon rule and the n(n-1)/2 full-mesh problem it creates. iBGP 的水平分割规则,以及它带来的 n(n-1)/2 全互联问题。
- Route reflectors: cluster ID and originator ID as the replacement loop-prevention mechanism. 路由反射器:用 cluster ID 和 originator ID 顶替原来的防环机制。
- Why a route can be in the BGP table but not installed — the unresolved next hop. 为什么路由在 BGP 表里却装不进路由表——下一跳不可达。
- [OSPF Link State · OSPF 链路状态路由](https://ipnetmaster.com/ospf): Flood LSAs to build an identical map of the network on every router, then run Dijkstra on it. Step through adjacency states, LSA types, area design and the SPF run that produces the routing table. 泛洪 LSA 让每台路由器拼出同一张全网地图,再各自跑 Dijkstra:邻接状态机、LSA 类型、区域划分,以及算出路由表的那次 SPF。
- Adjacency state machine: Down → Init → 2-Way → ExStart → Exchange → Loading → Full. 邻接状态机:Down → Init → 2-Way → ExStart → Exchange → Loading → Full 每一步在干什么。
- Why every router in an area must hold an identical LSDB, and what happens when they do not. 同一区域内为什么必须持有完全相同的 LSDB,不一致时会出什么问题。
- DR/BDR election on a broadcast segment, and the flooding traffic it saves. 广播网段上的 DR/BDR 选举,以及它省下的泛洪流量。
- Change one interface cost and watch the SPF tree and routing table change with it. 改一条接口开销,看着 SPF 树和路由表跟着变。
- [IS-IS Link State · IS-IS 链路状态路由](https://ipnetmaster.com/isis): The other link-state IGP, the one carriers actually run: L1/L2 levels instead of areas, TLV-encoded LSPs that extend without a protocol revision, and a topology that rides directly on layer 2. 运营商网络里更常见的那个链路状态协议:用 L1/L2 两级代替区域、用 TLV 编码的 LSP 不改协议就能扩展,而且直接跑在二层上。
- L1 / L2 / L1-L2 routers: how the two-level hierarchy replaces OSPF's area 0. L1 / L2 / L1-L2 三种角色:两级层次结构怎么顶替 OSPF 的骨干区域。
- TLV encoding is why IS-IS carried IPv6 without a new protocol version. TLV 编码就是 IS-IS 不改版本号就支持 IPv6 的原因。
- The attached bit and the default route it pulls L1 routers toward. ATT 位,以及它把 L1 路由器引向的那条默认路由。
- [EIGRP Distance Vector · EIGRP 距离矢量路由](https://ipnetmaster.com/eigrp): DUAL in action: successors and feasible successors are computed before anything breaks, so failover is a table lookup rather than a re-convergence. See exactly when a router has to go active and query its neighbours. DUAL 算法实况:后继和可行后继在故障之前就算好了,切换只是查一次表而不是重新收敛。看清楚什么情况下路由器不得不进入 active 状态去查询邻居。
- The feasibility condition: why only some backup paths are safe to pre-install. 可行性条件:为什么只有一部分备份路径可以提前装进表里。
- Successor vs feasible successor, and the moment the route goes active. 后继与可行后继,以及路由转入 active 的那一刻。
- The composite metric (bandwidth and delay by default) and what actually moves it. 复合度量(默认只看带宽和时延),以及真正能改变它的因素。
- [RIP Distance Vector · RIP 距离矢量原型](https://ipnetmaster.com/rip): The simplest distance vector protocol, kept here because it is the clearest way to watch the count-to-infinity problem happen — and to see exactly which of split horizon, route poisoning and hold-down timers stops it. 最简单的距离矢量协议。留着它是因为「计数到无穷」在这里看得最清楚,也最容易看出水平分割、毒性逆转、抑制计时器分别挡住了哪一步。
- Watch count-to-infinity happen step by step, then turn split horizon on and watch it stop. 一步步看着「计数到无穷」发生,再打开水平分割看它停下来。
- Why the metric maxes out at 16 hops, and why that is a feature here. 度量为什么到 16 跳就算不可达,以及这个上限在这里为什么是优点。
- Periodic updates vs triggered updates, and the convergence delay each one causes. 周期更新和触发更新,各自带来多长的收敛延迟。
- [Data Center Spine-Leaf · 数据中心 Spine-Leaf 架构](https://ipnetmaster.com/dc): How a fat-tree data center fabric is actually dimensioned: oversubscription ratio, ECMP hashing across spines, and what an elephant flow does to a hash-based load spread. 胖树数据中心网络到底怎么算容量:收敛比、跨 spine 的 ECMP 哈希,以及一条大象流会把基于哈希的负载分担压成什么样。
- Oversubscription: count the uplinks and the server ports, then decide if the ratio is honest. 收敛比:把上联口和服务器口数出来,再看这个比例撑不撑得住。
- ECMP hashes per flow, not per packet — which is why one big flow cannot be spread. ECMP 按流哈希而不是按包,所以单条大流根本分不开。
- Every leaf is two hops from every other leaf: the property the whole design exists to buy. 任意两台 leaf 之间都是两跳——整个架构就是为了买下这个性质。
- [OTN/ODF Optical Network · OTN/ODF 光网络](https://ipnetmaster.com/otn): How data centers actually interconnect over the physical optical layer: which devices a layer-1 circuit passes through from router to router, how a ROADM decides to add, drop or express a wavelength, and how ring protection survives a fiber cut. IDC 之间在物理光层到底是怎么互联的:一条一层电路从路由器到路由器经过哪些设备、ROADM 怎么决定一个波长是上路、下路还是直通,以及环网保护怎么扛住一次光缆挖断。
- Add / Drop / Express: the only three things that can happen to a wavelength at a ROADM, and the mechanism that decides which path a circuit takes. 上路、下路、直通:一个波长在 ROADM 上唯一可能发生的三件事,也是决定电路走哪条路的机制。
- A transponder is the only device that does optical-electrical-optical conversion — everywhere else on the path is either a passive patch (ODF) or all-optical (ROADM express, OLA). 转发器是唯一做光电光转换的设备——路径上其余地方要么是无源跳纤(ODF),要么全光通过(ROADM 直通、OLA 放大)。
- 1+1 ring protection lights both directions at once, so a fiber cut is a millisecond switch, not a recomputed path. 1+1 环网保护是两个方向同时点亮,光缆被挖断只是毫秒级切换,不是重新算路。
- [IDC Topology · IDC 拓扑](https://ipnetmaster.com/idc): How a region's data centers are laid out and related: core, aggregation, business, satellite, interconnect, scrubbing, PoP and colocation sites, active-active vs active-standby pairs, fiber distance and RTT computed from real coordinates, order-of-magnitude traffic, and a failover drill — take a DC or a backbone span down and watch who gets isolated, promoted or detoured. Modeled on AWS global regions and China's multi-region layout. 一个 region 里的数据中心怎么布局、怎么相连:核心、汇聚、业务、卫星、三方互联、高防、PoP、托管,双活与主备关系,按真实经纬度算出的光缆距离和往返时延,数量级的流量,以及故障演练——打掉一个机房或一段骨干,看谁被孤立、谁备升主、出口怎么绕行。以 AWS 全球 region 和中国多 region 为蓝本。
- Eight kinds of facility in one region, each with a position and a job — and the aggregation layer is the one that must never be a single point. 一个 region 里八种机房各有位置和职责,其中汇聚层是最不能单点的那个。
- Distance decides the relationship: RTT <= 2ms (100-200 km) allows synchronous active-active; anything farther is asynchronous, so RPO is no longer zero. 距离决定关系:往返 ≤ 2ms(100~200 公里)才能同步双活,再远只能异步,RPO 不再是 0。
- Failover is a derivation, not a script: take any site or span down and the same rules explain who is isolated, who is promoted, and where egress and replication detour. 故障演练不是剧本:打掉任何机房或骨干,同一套规则推出谁孤立、谁备升主、出口和复制绕到哪。
- [Layer 2 Switching · 网络二层交换](https://ipnetmaster.com/l2): What a switch actually does with a frame: learn the source MAC, look up the destination, flood when it is unknown. Includes ARP resolution and the broadcast domain that ties it all together. 交换机拿到一个帧到底做了什么:学源 MAC、查目的 MAC、查不到就泛洪。顺带把 ARP 解析和广播域这条主线串起来。
- MAC learning is source-based: the table fills up from traffic the switch happens to see. MAC 学习看的是源地址:表是被路过的流量填满的。
- Unknown unicast flooding — the normal behaviour that looks like a bug. 未知单播泛洪:看起来像故障的正常行为。
- ARP: how a host turns an IP address into the MAC it needs to build the frame. ARP:主机怎么把 IP 地址问成组帧需要的那个 MAC。
- [STP Spanning Tree · STP 生成树协议](https://ipnetmaster.com/stp): Redundant links plus a flat layer 2 equals a broadcast storm. STP computes a loop-free tree and blocks the rest: root bridge election, port roles and states, and what a topology change actually costs. 二层冗余链路等于广播风暴。STP 算出一棵无环的树,把多余的链路阻塞掉:根桥选举、端口角色和状态,以及一次拓扑变更真正的代价。
- Root bridge election by bridge ID — and why the default priority picks the wrong switch. 按桥 ID 选根桥,以及默认优先级为什么经常选中不该选的那台。
- Root / designated / blocked: how each port ends up with its role. 根端口/指定端口/阻塞端口:每个端口的角色是怎么定下来的。
- Without STP, one loop is enough to melt the segment — watch the frame multiply. 没有 STP 时一个环就足以打垮网段——看着帧一份份复制出去。
- [Layer 3 Static Routing · 网络三层静态路由](https://ipnetmaster.com/l3): Longest prefix match decided hop by hop, plus the classic trap that catches everyone once: routing is one-way, so a reachable destination with no return route still fails. 逐跳的最长前缀匹配,外加那个人人都会踩一次的经典坑:路由是单向的,目的地可达但没有回程路由照样不通。
- Longest prefix match: the most specific route wins, regardless of the order rules were entered. 最长前缀匹配:最精确的那条赢,跟配置顺序无关。
- Routing is one-way. Configure only the forward path and the ping still fails. 路由是单向的:只配去程,ping 照样不通。
- Default route vs a specific route, and when the packet falls through to 0.0.0.0/0. 默认路由和具体路由,报文什么时候才落到 0.0.0.0/0 上。
- [DNS · DNS 域名系统](https://ipnetmaster.com/dns): Walk a name resolution the way a resolver does: root → TLD → authoritative, one referral at a time. Recursive vs iterative queries, the roles of authoritative, recursive and forwarding servers, and where caching and TTL change the answer. 像解析器那样走一遍域名解析:根 → 顶级域 → 权威,一次一个委派。递归查询与迭代查询、权威/递归/转发三种角色,以及缓存和 TTL 在哪一步改变了结果。
- Recursive vs iterative: who does the walking, the client or the resolver. 递归还是迭代:到底是客户端在走,还是解析器在走。
- The delegation chain from the root down, one NS referral at a time. 从根开始的委派链,一次一个 NS 转介。
- TTL and caching: why a record change does not take effect everywhere at once. TTL 与缓存:改了记录为什么不会到处同时生效。
- [Load Balancing · 负载均衡](https://ipnetmaster.com/lb): Three decision granularities side by side: L3 sends a client to a server, L4 pins a connection, L7 can route every single request. See what each one can and cannot see, and where session persistence has to come from. 三种决策粒度并排看:三层把客户端指到某台服务器,四层锁定一条连接,七层可以每个请求单独调度。看清各自能看到什么、看不到什么,以及会话保持只能从哪一层来。
- L3 / L4 / L7: per client, per connection, per request — the whole difference in one line. 三层/四层/七层:按客户端、按连接、按请求——差别一句话就说完了。
- Health checks decide correctness; algorithms only decide fairness. 健康检查决定对不对,算法只决定均不均。
- Session persistence: why a stateless algorithm breaks a stateful application. 会话保持:无状态的算法为什么会把有状态的应用弄坏。
- [RDMA · RDMA 远程直接内存访问](https://ipnetmaster.com/rdma): Zero copy and kernel bypass: the NIC writes straight into remote memory, so the CPU and the kernel network stack are not on the path at all. That is also why RDMA traffic needs a lossless fabric and does not tolerate the drops TCP shrugs off. 零拷贝加内核旁路:网卡直接写进对端内存,CPU 和内核协议栈根本不在路径上。这也正是 RDMA 需要无损网络、受不了 TCP 满不在乎的那点丢包的原因。
- The data path with and without the kernel, drawn side by side. 走内核和不走内核的两条数据路径,并排画出来。
- Why RDMA needs PFC / ECN: one drop costs far more than it does for TCP. RDMA 为什么需要 PFC/ECN:一次丢包的代价远比 TCP 大。
- Queue pairs and completion queues: the interface that replaces sockets. 队列对与完成队列:顶替 socket 的那套接口。
- [InfiniBand · InfiniBand 高性能互联](https://ipnetmaster.com/ib): Ethernet vs RoCE vs InfiniBand, compared three ways. Inside the host the verbs interface is identical — every real difference lives in the network: credit-based flow control, a centralized subnet manager, and link-level reliability instead of end-to-end retries. 以太网、RoCE、InfiniBand 三路对比。主机内部的 verbs 接口完全一样,差别全在网络这一侧:基于信用的流控、集中式子网管理器,以及链路级可靠而不是端到端重传。
- Credit-based flow control: a sender that cannot overrun the receiver never needs to drop. 基于信用的流控:发送方压根超不了接收方,也就不需要丢包。
- The subnet manager: routing decided centrally instead of by a distributed protocol. 子网管理器:路由是集中算好的,不是分布式协议跑出来的。
- Same verbs, three fabrics — what actually changes when you swap the network underneath. 同一套 verbs、三种网络:把底下的网络换掉,到底有什么变了。
- [VLAN Layer 2 Isolation · VLAN 二层隔离](https://ipnetmaster.com/vlan): 802.1Q tagging: many independent broadcast domains sharing one physical wire. Access ports, trunk ports, the native VLAN, and why two hosts on the same switch with the same subnet still cannot reach each other. 802.1Q 标签:一根线上跑多个互不打扰的广播域。接入口、trunk 口、native VLAN,以及为什么同一台交换机上同网段的两台主机照样不通。
- Access vs trunk: exactly where the tag is added and where it is stripped. 接入口与 trunk 口:标签到底在哪儿加上、在哪儿摘掉。
- The native VLAN goes untagged — the mismatch that quietly bridges two VLANs. native VLAN 不打标签——配错时会悄悄把两个 VLAN 桥接起来。
- Same subnet, different VLAN, no connectivity: isolation is doing its job. 同网段、不同 VLAN、就是不通:隔离本来就是这个效果。
- [VPN Encrypted Tunnels · VPN 加密隧道](https://ipnetmaster.com/vpn): Site-to-site and remote access side by side: how the tunnel is negotiated, what the encapsulated packet looks like on the wire, and which traffic the split-tunnel policy actually sends through it. 站点互联和远程接入并排看:隧道怎么协商出来、封装后的报文在线路上长什么样,以及分离隧道策略实际把哪些流量送了进去。
- The packet before and after encapsulation, header by header. 封装前后的报文,一层头一层头对照着看。
- Site-to-site vs client-to-site: same crypto, very different traffic selection. 站点互联与客户端接入:加密是一套,选流量的方式完全不同。
- Split tunnel: the policy decides what is protected, not the tunnel. 分离隧道:决定什么被保护的是策略,不是隧道本身。
- [SSL/TLS · SSL/TLS 加密](https://ipnetmaster.com/tls): The handshake in full: how the certificate chain is validated up to a trusted root, how the session key is agreed without ever sending it, and what actually changed in TLS 1.3. 完整的握手过程:证书链怎么一级级验到可信根、会话密钥怎么在从不传输的前提下协商出来,以及 TLS 1.3 到底改了什么。
- Certificate chain validation, from leaf to intermediate to trusted root. 证书链校验:从叶子证书到中间 CA 再到可信根。
- Key exchange: how both sides end up with the same secret without it crossing the wire. 密钥交换:双方怎么得到同一个密钥,而它从没在网上传过。
- TLS 1.3 cut a round trip and dropped the legacy ciphers — what that means in the trace. TLS 1.3 少了一个往返、砍掉了历史算法——在抓包里具体是哪一步没了。
- [Network Products by Layer · 网络产品总览 L1-L7](https://ipnetmaster.com/overview): A vendor-neutral map of networking products from layer 1 to layer 7: what capability lives at each layer, which vendors worldwide sell it, and the order of magnitude of price and performance. 不分厂商的网络产品地图:一到七层每一层有哪些能力、全球哪些厂商在做、价位和性能的数量级。
- Every capability is placed at the layer where it actually operates, not where marketing puts it. 每项能力放在它真正工作的那一层,而不是营销说的那一层。
- For each capability: the vendors that sell it, the form factor, and a price / performance order of magnitude. 每项能力下:谁在卖、什么形态、价位和性能的数量级。
- A reference for comparing across vendors before you read any single vendor's page. 先看这一页,再去看任何一家厂商的页面。
- [Security Products by Layer · 安全产品总览 L1-L7](https://ipnetmaster.com/secoverview): Security capabilities placed by layer: firewall, IPS, WAF, DDoS scrubbing, zero trust, DNS security, and which vendors offer each. 安全能力按层归位:防火墙、IPS、WAF、DDoS 清洗、零信任、DNS 安全,各由谁提供。
- Where each security control sits in the stack, and what it can and cannot see from there. 每种安全控制在协议栈的哪一层,从那一层能看到什么、看不到什么。
- Vendors and form factors per capability, hardware and SaaS alike. 每项能力下的厂商与形态,硬件和 SaaS 都有。
- Read with the firewall, VPN and TLS labs on this site. 配合本站的 VPN 与 TLS 实验一起看。
- [Network Equipment Vendors L1-L7 · 网络设备厂商 L1-L7](https://ipnetmaster.com/equipment): Cisco, Arista, Juniper, Huawei, H3C, Ruijie, Ciena, Nokia, F5, A10, Extreme, Aruba, Ubiquiti, NETGEAR, TP-Link, Calix, Accton / Edgecore: each vendor's product lines from optical transport to application delivery, with company background and flagship layers. 思科、Arista、Juniper、华为、H3C、锐捷、Ciena、Nokia、F5、A10、Extreme、Aruba、Ubiquiti、NETGEAR、TP-Link、Calix、Accton:各家从光传输到应用交付的产品线、公司背景与主力层。
- Each vendor is read the same way: L1 to L7, with the layers its business actually depends on starred. 每家厂商用同一把尺子看:L1 到 L7,真正撑起它营收的层打星。
- Company profiles: origins, acquisitions, industry impact, and who they partner with. 公司简介:起源、并购、行业影响、和谁合作。
- Price and performance as orders of magnitude, so vendors can be compared, not quoted. 价位与性能按数量级给出,用来比较,不用来报价。
- [Networking Silicon L1-L7 · 网络芯片 L1-L7](https://ipnetmaster.com/netsilicon): Broadcom, NVIDIA, Marvell and the other chip makers inside the switches and NICs: what merchant silicon changed about the equipment business, and which layers each vendor's chips implement. Broadcom、NVIDIA、Marvell……交换机和网卡里面那颗芯片的厂商:商用芯片改变了设备业的什么,各家芯片实现的是哪几层。
- A switch ASIC is the thing that implements L2 forwarding — the vendor above it mostly adds software. 交换芯片才是真正实现二层转发的东西,上面的设备厂商主要加的是软件。
- Merchant silicon vs in-house silicon: the split that explains Arista, white boxes and SONiC. 商用芯片 vs 自研芯片:这道分水岭解释了 Arista、白盒和 SONiC。
- NICs, DPUs and optics: the silicon at the host edge, where AI clusters spend their money. 网卡、DPU 和光模块:主机侧的芯片,AI 集群的钱花在这里。
- [Carriers L1-L7 · 运营商 L1-L7](https://ipnetmaster.com/carriers): AT&T, Verizon, NTT, Comcast, Cogent, Colt, Lumen, Telstra, China Telecom and two dozen more: what a carrier sells at each layer, from dark fiber and wavelengths to private lines, IP transit and cloud on-ramps. AT&T、Verizon、NTT、Comcast、Cogent、Colt、Lumen、Telstra、中国电信等二十多家:运营商在每一层卖什么——从暗光纤、波长到专线、IP 中转和云互联。
- The same product has different names at every carrier; this page lines them up by layer. 同一种产品在每家运营商叫法都不一样,这一页按层把它们对齐。
- Tier-1 transit, regional carriers and cable operators read very differently once you look at L1. 一级中转商、区域运营商和有线电视运营商,看到 L1 就知道差别在哪。
- Where a carrier's product ends and the interconnect site's begins. 运营商的产品到哪里为止、互联点的从哪里开始。
- [Data Center Operators L1-L7 · 数据中心运营商 L1-L7](https://ipnetmaster.com/datacenters): Equinix, Digital Realty, CoreSite, Colt DCS and others: what a data center operator actually provides — spa