分类
网络
本分类下的所有工具与文章，集中在一处。
工具
Cable run planner
Speed, distance, environment, PoE in - every compliant cable and fiber option out, each citing the TIA-568 or 802.3 rule that governs it, with honest exclusions for the rest.
CIDR / 子网计算器
将任意 IPv4 CIDR 块拆解为网络地址和广播地址、可用主机范围、主机数量和子网掩码。完全在你的浏览器中运行。
dig output explainer
Paste real dig output and get a decoded, explained breakdown: the header and flags, the EDNS OPT pseudo-section, every record in each section, and the query stats. Parsed entirely in your browser, nothing is resolved or sent anywhere.
HTTP methods comparison
Enter one to four HTTP methods (try "get vs query") to see each one's registry facts: safe, idempotent, cacheable, request-body semantics, CORS safelist membership, HTML-form support, and the defining RFC - plus exactly which properties differ. Includes QUERY, the RFC 10008 method registered in June 2026.
HTTP status code explainer
Paste status codes (404, 301 302 307) or families (5xx) and get each one decoded: family, registered name, documented meaning, and the operational notes - including the protocol's own fallback rule for codes it has never met.
IPv6 工具箱
解析一个 IPv6 地址或前缀，查看其规范（RFC 5952）形式和完全展开形式、特殊用途分类、前缀运算、内嵌的 EUI-64 MAC（若存在）以及其 ip6.arpa 反向 DNS 名称。完全在你的浏览器中运行。
MTU / MSS calculator
Enter a link MTU and your encapsulation stack (vxlan, gre, pppoe, vlan, mpls, +N custom) to get the inner MTU and TCP MSS, the Ethernet frame sizes, the underlay MTU an overlay needs, and the wire efficiency of standard versus jumbo frames.
nslookup output explainer
Paste real nslookup output and get a decoded, explained breakdown: the resolver it used, whether the answer is authoritative, every record (with MX / SRV / SOA field breakdowns), and any failures. Parsed entirely in your browser, nothing is resolved or sent anywhere.
OUI / MAC vendor lookup
Look up the manufacturer behind a MAC address, and read whether it is unicast or multicast and universally or locally administered. All in your browser.
p0f passive TCP/IP signature explainer
Decode a p0f v3 SYN fingerprint into its eight fields and see what each reveals about the sender's OS, network path, and stack.
ZscalerPAC file explainer and validator
Paste a Proxy Auto-Config file and it reads back the proxy directives it returns, the PAC helper functions it uses (with the DNS-consulting ones flagged), structural and correctness lints, and whether it is a Netskope explicit-proxy steering file. Never runs the file.
NetskopeRegistered domain (eTLD+1)
Find the public suffix (eTLD) and registered domain (eTLD+1) for any hostname, using the Public Suffix List.
Syslog PRI decoder + encoder
Decode a syslog PRI such as 134 into its facility and severity, or encode them back, all in your browser.
VOSS / EXOS command translator
A side-by-side reference between VOSS (Fabric Connect / SPBM) and EXOS. Search a task or command to see how each CLI expresses it - and where EXOS has no equivalent.
VOSS fabric identifier decoder
Decode an Extreme SPBM fabric identifier - a 24-bit I-SID, a 20-bit nickname (X.XX.XX), or a system-id / B-MAC. Auto-detected, all in your browser.
文章
IPv4 地址如何运作
每一个点分四段地址背后的那 32 位，以及私有、回环与特殊地址段意味着什么。
阅读The Syslog PRI: One Number, Two Meanings
Every syslog message starts with a PRI, a number in angle brackets that packs a facility and a severity into a single value. The formula is small and the arithmetic is easy once you have seen it: PRI equals facility times eight plus severity.
阅读Syslog Facilities and Severities, Explained
Syslog defines 24 facilities and 8 severities. The severities are a clean urgency scale from emergency down to debug; the facilities are a mix of genuinely useful categories and historical Unix leftovers, plus eight local slots that network devices lean on heavily.
阅读子网划分基础
如何把一个网络分割为更小的子网，以及为何借用主机位便是全部的诀窍。
阅读CIDR 记法详解
192.168.1.0/24 中的那道斜杠究竟意味着什么，以及一个前缀长度如何界定一个 IP 地址的块。
阅读Syslog on Network Devices: Which Facility Does What
Firewalls, load balancers, and switches almost all log to the local facilities, but each vendor picks a different default. Knowing that FortiGate defaults to local7, Cisco ASA to local4, and F5 BIG-IP to local0 turns a wall of PRI numbers into a map of which box said what.
阅读Syslog Message Formats: RFC 3164 vs RFC 5424
The PRI is the same everywhere, but what follows it is not. Legacy BSD syslog (RFC 3164) has a loose, year-less format, while the modern format (RFC 5424) is precise and structured. Knowing which one you are looking at explains missing timestamps, ambiguous fields, and why parsers disagree.
阅读VLSM：把一个块划分为不等的子网
如何在不浪费空间的情况下把一个地址块切成不同大小的子网，以及让一切井然有序的最大优先规则。
阅读How Syslog Travels: UDP, TCP, and TLS
Syslog can ride over plain UDP, over TCP, or over TLS, and the choice decides whether messages can be silently lost, reordered, or read in transit. This covers the three transports, the ports involved, and why anything you rely on for audit should not be sent over UDP.
阅读一个从头到尾求解的 VLSM 分配
为一个现实网络做的完整可变长度子网分配：为每个网段确定大小、从大到小排序、分配真实地址，并对剩余空间进行记账。
阅读The First Hour: Hypothesis-Driven Fault Isolation
The difference between a two-hour incident and a two-day one is usually decided in the first hour, and it is rarely decided by tools. It is decided by method: aligning onset with change, isolating by scope, reading layer signatures, and treating every explanation as a hypothesis that evidence must support or weaken before anyone acts on it.
阅读超网与路由聚合
相邻前缀如何合并为一个更短的前缀，决定两个块能否合并的对齐规则，以及精确聚合与覆盖它们的单一超网之间的区别。
阅读Change Windows That Do Not Become Incidents
Most self-inflicted outages are changes that went wrong with no clean way back. The difference between a change and an incident is rarely the change itself; it is the runbook around it - what you verified before, how you sequenced it, what would make you stop, and whether the way back was written down and tested before you needed it.
阅读路由汇总
为什么一条汇总路由可以替代许多具体路由，它所依赖的连续且对齐的分配，以及汇总一个你并不完全拥有的范围时的黑洞风险。
阅读Root Cause Is a Verb, Not a Noun
The phrase root cause invites a single villain and a tidy ending. Real incidents rarely have one; they have contributing factors, and the honest work is structuring the candidates and the evidence that would confirm or rule out each - not naming a culprit before the evidence is in.
阅读子网重叠与空隙
两个前缀重叠或一个包含另一个意味着什么，为什么最长前缀匹配使某些重叠是有意为之的，以及如何找出地址规划中未分配的空隙。
阅读Blast-Radius Thinking Before You Change Anything
Before a change, the question is not only will this work but if it goes wrong, how far does the damage reach. Blast radius is the shape of that reach - target, neighbours, dependents, people - and thinking about it in tiers is how you decide what to contain before you touch anything.
阅读私有 IPv4 地址空间与 RFC 1918
三个私有范围、它们为何在互联网上不可路由，以及 CIDR 工具会标记的其他特殊用途块。
阅读TAC Cases That Get Triaged Fast
The slowest support cases are rarely the hardest problems; they are the ones that opened without the diagnostic the vendor needs. A case that arrives complete - clear problem, exact error, the diagnostic bundle, the impact - skips the round trips and starts with an engineer actually working it.
阅读理解 IPv6 编址
一个 128 位的 IPv6 地址如何被构造与书写、把它规范地压缩的规则、那些地址类型与作用域意味着什么，以及接口标识符与反向 DNS 如何运作。
阅读IPv6 主机如何取得地址：SLAAC 与 DHCPv6
一台 IPv6 主机如何自链路本地向上自我配置、路由器通告决定了什么，以及 SLAAC、隐私地址与 DHCPv6 之间的差别。
阅读Capture Points Before Packets
The instinct under pressure is to start tcpdump somewhere convenient and stare at the flood. The senior habit is the opposite: decide where to observe, in what order, and what each observation would mean - before a single packet is collected. Evidence is designed, not fished for.
阅读IPv6 子网划分与 /64 边界
为何 IPv6 子网划分关乎结构、而非稀缺，为何单个子网几乎总是一个 /64，以及前缀委派如何发放地址空间。
阅读Public suffixes and the registered domain (eTLD+1)
What a public suffix (eTLD) and a registered domain (eTLD+1) are, why you cannot compute them by taking the last two labels, how the Public Suffix List algorithm resolves them, and where the boundary matters: certificate rate limits, cookies, and same-site.
阅读Map the Path Before You Troubleshoot
The most expensive troubleshooting sessions share one flaw: nobody actually knew the path. The senior habit is refusing to reason about a failure until the chain is modeled - resolution included, return leg separate, every rewrite and TLS boundary named, and every unknown left visibly unknown.
阅读邻居发现：IPv6 如何取代 ARP
IPv6 如何在一条链路上以 ICMPv6 与多播、而非广播式 ARP 来寻得邻居，那五条邻居发现消息，以及广播为何消失。
阅读Baselines Before You Need Them
Every comparison is only as honest as its weaker side. The discipline is captured baselines before changes, declared confidence instead of remembered health, observation windows that let convergence speak, churn literacy for the counters that always move - and never calling a change successful on green components alone.
阅读让 IPv6 与 IPv4 共存：双栈与转换
互联网如何在两个互不兼容的地址族之间架桥：双栈、Happy Eyeballs、NAT64，以及那些使之运转的、内嵌 IPv4 的地址。
阅读What Is an OUI, and How MAC Addresses Are Assigned
How a MAC address is structured, what the OUI (the manufacturer prefix) is and who hands it out, the difference between universally and locally administered addresses, and why a randomized Wi-Fi MAC has no vendor at all.
阅读Fabric Connect and SPBM: Why VOSS Retires Spanning Tree
What Extreme's Fabric Connect actually is - Shortest Path Bridging MAC (SPBM, IEEE 802.1aq) with an IS-IS control plane and a MAC-in-MAC data plane - and why collapsing the core to a single link-state protocol replaces spanning tree and the usual overlay stack.
阅读Reading dig Output From Top to Bottom
A dig answer has a fixed shape: a version line, the header, the flags line, the OPT pseudo-section, the four sections, and the query stats. Once you know what each block is, you can read any response at a glance and spot the one line that explains a resolution problem.
阅读The DNS Header: Opcode, Status, and Flags
The header line and the flags line hold the message-level facts: what kind of query this is, whether it succeeded, and seven single-bit flags (qr, aa, tc, rd, ra, ad, cd) that tell you who answered and how. Reading them correctly is the difference between a two-minute diagnosis and an hour of guessing.
阅读The I-SID: How VOSS Replaces VLAN Stretching
Why Extreme's Fabric Connect provisions services at the edge instead of trunking VLANs hop by hop, what the 24-bit I-SID is, and how Layer 2 VSN, Layer 3 VSN, and IP Shortcuts all ride the same mechanism over an SPBM MAC-in-MAC core.
阅读IS-IS, Nicknames, and B-MACs: The VOSS Control Plane
How SPBM uses IS-IS as its single link-state control plane on fabric links only, what a 20-bit node nickname is and why it must be unique, and how the system-id / backbone MAC drives MAC-in-MAC forwarding.
阅读Reading the Records in a dig Answer
Every record in a dig section is five columns: name, TTL, class, type, and rdata. This walks the columns and then the rdata of the record types you actually meet, from A and CNAME to MX, SOA, SRV, and CAA, so a wall of records reads as plain facts.
阅读EDNS and the OPT Pseudo-Section
The OPT pseudo-section is not a record and not something you queried: it is EDNS(0) metadata that dig surfaces near the top of an answer. It carries the UDP payload size, the DO flag that requests DNSSEC, and options like COOKIE, and it quietly explains a whole class of resolution failures.
阅读Fabric Attach: Auto-Provisioning the Edge (Where VOSS Meets EXOS)
How Fabric Attach lets an edge device signal the service it needs so the fabric provisions the I-SID automatically, the FA Server / Proxy / Client roles, how it rides LLDP (Link Layer Discovery Protocol), and how an EXOS switch attaches to a VOSS fabric without running SPBM itself.
阅读DNSSEC Records in dig Output
Add +dnssec and a dig answer grows a new family of records: RRSIG, DNSKEY, DS, and the NSEC or NSEC3 denial records. This explains what each one is, how they chain from the root down to a zone, and what the ad flag really certifies.
阅读VOSS vs EXOS: Two Extreme Operating Systems
Extreme ships universal hardware that boots either EXOS or VOSS (Fabric Engine). This is what actually differs - a traditional-Ethernet OS with an intuitive CLI versus a fabric-native OS built on SPBM - and the three boundaries at which they interconnect.
阅读Reading nslookup Output
nslookup prints a Server / Address header for the resolver it used, an optional Non-authoritative answer marker, and then the answer in a per-type prose format. Knowing that shape lets you read any result quickly and see at a glance whether it succeeded, where it came from, and what it means.
阅读SMLT and vIST: Dual-Homing a Fabric Edge
How Split MultiLink Trunking dual-homes an edge device to a pair of switches with active-active links and no spanning tree, how virtual IST runs the inter-switch trunk through the SPBM fabric itself, and the smlt-peer-system-id and smlt-virtual-bmac that make the cluster one logical node.
阅读nslookup vs dig: Which to Use
nslookup and dig both query DNS, but nslookup is terser and hides the header flags and TTLs that dig shows in full. This maps one output onto the other and gives a simple rule for which to reach for.
阅读How nslookup Prints Each Record Type
Instead of dig's fixed columns, nslookup labels each record in prose: mail exchanger =, canonical name =, nameserver =, and a multi-line block for SOA. A short guide to reading each type's line.
阅读Authoritative vs Non-Authoritative Answers
The Non-authoritative answer marker in nslookup means the result came from a resolver's cache, not from a server that actually holds the zone. This explains the difference, why it is usually fine, and how to get an authoritative answer when you need one.
阅读nslookup Errors and What They Mean
When a lookup fails, nslookup prints a line like ** server can't find NAME: CODE. The code is the whole diagnosis. This covers NXDOMAIN, SERVFAIL, REFUSED, and timeouts, what each one tells you, and the first thing to check for each.
阅读Reverse DNS Lookups with nslookup
Reverse DNS maps an IP address back to a name through PTR records that live under in-addr.arpa for IPv4 and ip6.arpa for IPv6. nslookup does this automatically when you hand it an address. This covers how the special reverse name is built, why mail servers care, and why the forward and reverse can legitimately disagree.
阅读nslookup Interactive Mode
Run nslookup with no arguments and it drops into an interactive prompt where you can switch resolvers, change the record type, turn on debug output, and look up many names in one session. This covers the handful of commands worth knowing and when interactive beats a one-shot query.
阅读dig Query Options and Output Control
dig's real power is its options: choosing the server to ask, the record type, and exactly how much of the answer to print. This covers the handful you will actually use every day, from @server and -t to +short and the +noall +answer combination that trims dig down to just the records.
阅读Following Delegation with dig +trace
dig +trace resolves a name the way the internet actually does it: starting at the root, following the delegation to the TLD, and then to the domain's own authoritative servers, printing each hop. It is the single best way to see where resolution breaks and to understand how DNS is stitched together.
阅读Reading a curl Command
A curl command is a shell command: the word curl, a set of options, and a URL. Reading it means seeing how the shell splits the line first (quotes, backslashes, line continuations) and then how curl reads short, long, and clustered flags.
阅读curl Data Flags and the Content-Type Trap
curl has several ways to attach a body, and they differ in encoding and default Content-Type. The big surprise is that -d defaults to form encoding, not JSON, so a JSON body can be mislabeled and rejected.
阅读Translating curl to fetch()
The browser fetch API and curl describe the same request differently. Method, headers, and body map across cleanly, but a couple of differences (implicit form Content-Type, cookies, and TLS verification) need care.
阅读Headers, Authentication, and Cookies in curl
Headers, auth, and cookies are how a request identifies and authorizes itself. -H adds headers, -u is HTTP Basic, a bearer token is just a header, and -b/-c handle cookies. All of them are sensitive.
阅读How curl Infers the HTTP Method
curl does not always need -X to choose a method. Body data implies POST, -I implies HEAD, -G forces GET, and an explicit -X always wins. Knowing the rules tells you at a glance what a request will do.
阅读curl Flags That Change Security Posture
A few curl flags change how safe a request is: -k disables TLS verification, http sends everything in clear text, and credentials in the URL can leak. None make a request malicious, but each is worth reading before you run or share a command.
阅读The TCP Proxy: What a Layer 4 Middlebox Does and Does Not See
A TCP proxy terminates the client's TCP connection and opens a separate one to the server, splicing two independent flows together at Layer 4. It rewrites addresses and ports, can pool and reuse connections, and sees nothing of the application payload above the transport header. This explains full-proxy versus packet-forwarding, why the source IP disappears, and how the Proxy Protocol puts it back.
阅读HTTP Proxies: Forward vs Reverse, Explicit vs Transparent
An HTTP proxy parses requests at Layer 7, so it can route by URL, rewrite headers, and enforce policy on content a TCP proxy cannot see. Two axes describe every deployment: forward vs reverse (which side it works for) and explicit vs transparent (whether the client knows it is there). This covers the CONNECT method, X-Forwarded-For and Via, and where each combination is used.
阅读The 27 Protocols curl Speaks
curl is known as an HTTP tool, but the current tool speaks 27 URL schemes: file transfer over FTP, SFTP and SMB, mail over SMTP, POP3 and IMAP, MQTT publish-subscribe, LDAP lookups, and relics like Gopher, DICT and Telnet. Knowing the map, and which schemes start in cleartext, changes how you use it.
阅读HTTP/0.9 vs 1.0 vs 1.1 vs 2 vs 3: Five Versions of the Web's Protocol
From a one-line GET that could only fetch HTML to a multiplexed protocol riding QUIC: what each HTTP version added, why it was needed, and which RFC defines it today - including the 2022 reorganization that split HTTP semantics (RFC 9110) from the per-version wire syntax.
阅读AltaVista: built to break a processor, and lost by becoming a portal
AltaVista started as a benchmark. A researcher wanted a workload chaotic enough to stress DEC's new Alpha chips, so he pointed a crawler at the entire web. The result led search for years and then was dismantled by its own owners, which makes it the clearest case study in the industry of losing a market you already have.
阅读GPON: How One Fiber Serves a Whole Neighborhood
A Gigabit Passive Optical Network runs one strand of glass from the exchange, splits it with unpowered prisms, and shares it among dozens of homes. How the light is divided, why upstream needs a schedule, and where the decibel budget goes.
阅读Jumbo frames: when 1500 bytes stops being enough
Why Ethernet settled on 1500 bytes, what 9000-byte jumbo frames actually buy, where they shine, and how an MTU mismatch turns into a silent black hole.
阅读HTTP QUERY: the read that finally carries a body
RFC 10008 (June 2026) gave HTTP its first new method since 2010: QUERY is safe, idempotent, and cacheable like GET, but carries a request body like POST. What it fixes, how its caching and discovery work, why 'safe' is not 'harmless', and what every layer of infrastructure must check before QUERY traffic arrives.
阅读The OSI Model in Practice: Mapping Real Traffic to Seven Layers
The Open Systems Interconnection model earns its place not as trivia but as a shared address system for problems: a MAC address is Layer 2, an IP address Layer 3, a TCP or UDP port Layer 4, and the application on top. How to map real artifacts to layers, where the model bends against TCP/IP reality, and why troubleshooting conversations go faster when everyone points at the same floor.
阅读Switch, Router, Firewall: Who Does What on the Path
Three device roles carry most of every network diagram: switches forward frames inside a Layer 2 domain, routers move packets between networks, and firewalls decide what is allowed to pass. The function and purpose of each, how an application delivery controller relates to all three, and how to read a network diagram without getting lost.
阅读ARP and MAC Addresses: How IP Finds Ethernet
Every packet on a local network is delivered by MAC address, and ARP is the directory service that maps an IP to one. The one-to-one mapping, reading ARP output to prove resolution worked, gratuitous ARP, and MAC masquerading - the failover trick that makes an address move without the switches noticing.
阅读Routing Tables and the Default Gateway: How a Packet Picks Its Next Hop
Why a route is needed at all, what a hop is, and the longest-prefix-match rule that decides which table entry wins - worked through the way the retired fundamentals exam asked: given a destination and a routing table, identify the route to be used. Plus the default route, the entry that answers when nothing else does.
阅读NAT Explained: Source, Destination, and Why the Internet Still Works
Network Address Translation rewrites addresses in flight - source NAT lets a thousand private hosts share one public address, destination NAT publishes an inside service on an outside address, and the translation table is what makes the return traffic find its way home. The function and purpose, the flavors, and the troubleshooting consequences of each.
阅读DHCP: the Lease Lifecycle Behind Automatic Addressing
The Dynamic Host Configuration Protocol hands out addresses, gateways, and resolvers so hosts arrive configured instead of blank. The discover-offer-request-acknowledge exchange, what a lease means and how renewal works, relays that carry requests across subnets, and the failure signatures - including the self-assigned address that says no server ever answered.
阅读The TCP Connection Lifecycle: Handshake, Teardown, and Why Connections Fail
Every TCP conversation has three acts: the SYN handshake that creates it, the data flow that justifies it, and the FIN or RST that ends it. The difference between polite close and abort, the possible reasons a connection terminates, and the short list of causes when one fails to establish at all - the literacy behind every capture you will ever read.
阅读VPN Fundamentals: What Tunnels Protect, and What They Don't
A virtual private network wraps traffic in an encrypted tunnel so it crosses untrusted networks as if it never left home. The rationale - privacy, encryption, and the limits of anonymity - the valid uses from site-to-site links to remote access, and the honest boundaries: what a tunnel genuinely protects, and the claims it cannot keep.
阅读The Last Mile: From POTS to Always-On
The story of residential connectivity is the story of one stubborn stretch of infrastructure - the last mile - being reinvented four times: the plain old telephone service that carried voices, the ADSL trick that made the same copper pair carry data full-time, the cable plant's DOCSIS second life, and fiber to the home. Why 1999's 256 kbps mattered more than its speed, what always-on actually changed, and how one house can end up served by copper's ghost, coax, glass, and the sky at once.
阅读Pagers and Paging Networks: Reachable Before Real-Time
Before the phone in your pocket, there was the bip on your belt: a one-way radio receiver on a simulcast broadcast network, fed - in its classic form - by a human operator who took your caller's words and typed them into the air. How paging networks actually worked, from two-tone beeps to POCSAG and FLEX, why the operator-relay model meant strangers read your life aloud, and why hospitals kept pagers long after everyone else moved on.
阅读LEO Constellations: Why the New Satellite Internet Is a Different Animal
Geostationary satellites solved coverage and lost latency: parked at 35,786 km, physics alone charges half a second round trip. Low Earth Orbit constellations invert the deal - satellites a few hundred kilometers up, latency in the tens of milliseconds, at the price of needing thousands of moving satellites, phased-array terminals that track them, and constant handoffs. The geometry, the engineering it forces, inter-satellite laser links, and the honest trade-offs.
阅读LoRa and LPWAN: Kilometers of Range on a Coin Cell
There is a corner of networking where the requirements invert everything broadband optimizes for: send a few bytes, a few times an hour, kilometers through a city, from a battery that must last years. LPWAN is that corner; LoRa is its best-known citizen - Semtech's chirp spread spectrum radio trading bitrate for astonishing link budgets, and LoRaWAN, the LoRa Alliance's network layer, organizing gateways, device classes, and the duty-cycle etiquette of shared spectrum.
阅读HTTP Methods: The Verbs of the Web
GET, HEAD, POST, PUT, DELETE, PATCH, OPTIONS, TRACE, CONNECT - and now QUERY. What each method promises, why 'safe' and 'idempotent' are the two properties that actually matter (to caches, retries, proxies, and crawlers), why HTML forms only ever learned two verbs, and how to read an API's soul from the methods it accepts.
阅读HTTP Status Codes: The Five Families
Three digits, and the first one does most of the work: 1xx continues, 2xx succeeds, 3xx redirects, 4xx blames the client, 5xx confesses for the server. The family logic, the codes an operator actually meets (200, 204, 301 vs 302 vs 307, 304's cache dance, 401 vs 403, 404 vs 410, 429, 500 vs 502 vs 503 vs 504), the famous curiosities, and why an unknown code's first digit is always enough to act on.
阅读HTTP Headers: The Anatomy of the Metadata
Everything HTTP knows about a message that isn't the message travels in headers: name-colon-value lines with case-insensitive names, folded into four working roles - request context, response context, representation metadata, and payload plumbing. The end-to-end vs hop-by-hop split that proxies live by, the Host header that made virtual hosting possible, content negotiation, conditionals, and why header order became a fingerprint.
阅读HTTP Cookies: State Over a Stateless Protocol
HTTP forgets you after every request - by design. Cookies are the retrofit that lets it remember anyway: the server writes a note with Set-Cookie, the browser returns it with Cookie, and everything else - scope, lifetime, security - is rules about when that note travels. Domain and Path scoping, session vs persistent lifetimes, why the server never sees what the browser knows, and where the security flags article picks up.
阅读AJAX, XHR, and fetch: When Pages Learned to Talk Back
For its first decade the web had one move: click, blank screen, new page. XMLHttpRequest gave pages a second one - request data in the background, update in place - and 'AJAX' named the revolution (which promptly dropped the X for JSON). How XHR worked, what fetch fixed (promises, streams, a sane API), what stayed the same underneath (it is all still HTTP), and the boundary every background request answers to: same-origin, with CORS as the negotiated exception.
阅读HTML, CSS, and the DOM: The Page as a Living Tree
Three technologies, one division of labor: HTML declares structure, CSS declares presentation, and the DOM is what actually exists at runtime - the tree the browser built from your HTML, the only thing scripts can touch, and the reason 'view source' and 'inspect element' show different worlds. How markup becomes a tree, how selectors address it (the same selectors CSS styles with and scripts query with), and why the DOM is where XSS happens and where CSP stands guard.
阅读What Is an Algorithm? A Working Primer
An algorithm is a finite, unambiguous recipe that turns input into output - and the engineering questions are always the same three: is it correct, how does its cost grow, and what does it trade away. Big-O as the grammar of growth, why constants and asymptotes both matter, the core families you already operate (search, sort, hash, graph, state machines), and where each one is already running inside this site's own tools.
阅读The IEEE 802 family, group by group
802 is not one standard but a committee of working groups, each owning a slice of the lower two layers. Here is the map: who owns Ethernet, who owns Wi-Fi, what the numbers after the dot mean, and which groups are alive, hibernating, or long dead.
阅读Structured cabling: the system behind the wall jack
TIA-568 turns a building's wiring from improvisation into a system: horizontal runs, telecom rooms, the 100-meter channel, categories, pinouts, and the fire ratings inspectors actually check. The map from wall plate to backbone.
阅读BGP: how the internet decides where traffic goes
The Border Gateway Protocol is the internet's routing brain - and it works nothing like the routing inside your network. A practical primer: autonomous systems, the path-vector idea, the attributes that encode business policy, why convergence is slow on purpose, and what RPKI fixes.
阅读OSPF: the map, the math, and area 0
Open Shortest Path First is the standard interior routing protocol: every router gets the whole map and runs Dijkstra on it. A practical primer: link-state flooding, areas and the backbone rule, cost and the reference-bandwidth trap, DR elections, and the design habits that keep OSPF boring.
阅读IS-IS: the routing protocol hiding under the internet
OSPF's link-state twin from the OSI world runs the carrier backbones your packets cross daily - and most engineers have never typed its name. A practical primer: the layer-2 trick, NET addresses, levels instead of areas, TLV extensibility, and why the big networks never left.
阅读MPLS: labels, stacks, and the BGP-free core
Push a label at the edge, swap it hop by hop, pop it before the exit - and suddenly the core needs no routing table, VPNs isolate thousands of customers, and traffic goes where engineering says. A practical primer on label switching, L3VPNs, and the Segment Routing present.
阅读Peer-to-peer, from Napster to the swarm: the architectures that outlived the piracy
Napster, Gnutella, Kazaa, eMule, BitTorrent - four years of file-sharing produced five distinct network architectures, each solving the weakness the last one died from. The centralized index, unstructured flooding, the supernode, the distributed hash table, and the swarm - what each actually was, how the courtroom shaped the topology, and why these ideas quietly power Skype, streaming CDNs, and blockchains today.
阅读The Brazilian market reserve, in theory and in practice
For eight years Brazil legally reserved its computer market for domestically-owned companies. The theory was an infant industry that would grow up and compete. The practice was clones, smuggling, a trade war with Washington, and a generation of engineers who learned on machines the policy created. Both halves are true.
阅读Digital transformation: what actually changed, and how to read what comes next
Transformation is not the technology arriving. It is the moment a capability stops being remarkable and becomes assumed. A look at what genuinely changed across money, work, health, and the state, and then an honest method for reading predictions, including a deadline that is real and moving at the same time.
阅读How the internet gets shut down, and who can actually do it
There is no off switch, but there are chokepoints: cables, routing, exchanges, and the small number of companies most traffic passes through. A tour of the mechanisms, what each state can genuinely do with them, and the uncomfortable conclusion that watching everything is easier than switching anything off.
阅读Who actually governs the internet in Brazil
Brazil built one of the world's most distinctive internet governance models: multistakeholder before that was a word, with a research foundation running the country's first connection and a committee that is not a ministry. The organs, the laws, the state computing backbone underneath it all, and the 2025 ruling that rewrote platform liability.
阅读Amateur radio: callsigns, modes, and why it still matters
A callsign is an identity issued by treaty, readable anywhere on earth. What the letters and the digit mean, how PY2 says São Paulo, the difference between voice and CW and digital, what a repeater actually does, and why a hobby older than broadcasting is still the fallback when everything else fails.
阅读The data communications window, 1968 to 1972
A regulator opened a market. The Carterfone decision let equipment AT&T had not built attach to the telephone network, and within four years an industry of modem and multiplexer companies existed that could not have existed before. The trigger, the window, the products that defined it, and why it closed.
阅读The networking window, 1979 to 1982
Corporations had bought computers for a decade and now needed them to talk to each other inside a building. The window that opened around 1979 produced local area networks and the data PBX, settled a contest between two answers to the same problem, and closed once the answer was obvious.
阅读The radio spectrum: which frequencies go how far, and why
Why a shortwave signal crosses an ocean on 100 watts while your Wi-Fi struggles through a wall. The bands from VLF to EHF, the three propagation modes that explain nearly everything, and the trade that governs all radio: reach or capacity, never both.
阅读The internetworking window, 1984 to 1988
The previous window's success created this one's problem: buildings full of local area networks that could not reach each other. Bridges and routers answered it, a standards war ran alongside, and two trade shows in 1988 made the outcome visible before the argument was formally over.
阅读The two shows of 1988
One trade show put competing vendors' equipment on a single live network in public, where failure would be seen. The other showed an ecosystem that was largely described rather than shipped. The standards argument was formally undecided and a buyer walking both floors did not need it settled.
阅读The founders who kept founding
Reading this industry as a sequence of companies misses that it is substantially the same population of people, recombining. Ungermann left Intel for Zilog and Zilog for Ungermann-Bass; Metcalfe left Xerox for 3Com. What that pattern explains, and what it does not.
阅读BFD: when a link is up and dead at the same time
Routing protocols detect failure with their own timers, measured in tens of seconds. BFD detects it in milliseconds, and exists because the most dangerous link failure is the one where the interface stays up. What it does, what it does not do, and where it goes wrong.
阅读First-hop redundancy: VRRP, HSRP, and the gateway that is a fiction
A host knows one default gateway and cannot fail over. VRRP and HSRP solve that by making the gateway address belong to a role rather than to a router. How they work, why the protocol choice matters less than people think, and the failure modes that make a redundant pair worse than a single router.
阅读Multicast: what it costs to not flood
Multicast sends one copy where unicast would send thousands, and the saving is real. What it buys in bandwidth it pays for in state, and the failure modes are not the ones people expect: a network that floods multicast everywhere is usually working exactly as configured.
阅读Public DNS resolvers: what you are actually choosing
Cloudflare, Google, Quad9, OpenDNS and the rest differ far less in speed than in policy. What each one does with your queries, which ones block by default, why anycast makes 'nearest' complicated, and the trade nobody states: the resolver that protects you also sees everything you ask for.
阅读FortiGate as a DNS server: four modes, and where the filter applies
A FortiGate can relay DNS, answer from its own database, or resolve from the root itself. The four modes behave differently under failure, and the DNS filter profile does not apply to all of them - which is the configuration that looks protected and is not.
阅读