[{"id":"networkingadvanced-001","courseId":"networking-advanced","domain":"ip-layer","topic":"Anycast addresses","prompt":"An IPv6 address is configured on interfaces in several data centers, and each packet sent to it should reach only one of them. Which TWO statements about this kind of address are correct?","hint":"Consider how many interfaces receive one packet, and whether the address format differs from a normal address.","explain":"An address normally names one interface. The same address can also sit on several interfaces, and the routing system then decides which of them a packet reaches. Think about how many copies a sender's packet makes.","source":"https://www.rfc-editor.org/rfc/rfc4291.html","interaction":{"type":"multiple","tellCount":true},"options":[{"id":"a","text":"Packets sent to it are routed to the nearest interface that holds it, by the routing protocols' measure of distance.","why":"Routing picks one interface holding the address: the one nearest by the routing protocols' distance measure."},{"id":"b","text":"Its bits are syntactically the same as those of an ordinary unicast address.","why":"Anycast addresses are allocated from unicast space and use the unicast formats, so they look like unicast addresses."},{"id":"c","text":"Every interface holding it receives a copy of each packet sent to it.","whyNot":"That is multicast delivery; an anycast packet goes to only one of the interfaces that hold the address."},{"id":"d","text":"It must come from a reserved prefix that unicast addresses can never use.","whyNot":"Anycast addresses use any of the defined unicast formats, so no separate reserved prefix is needed."},{"id":"e","text":"Nodes spot the shared address by seeing duplicates, with no configuration at all.","whyNot":"A node must be explicitly configured to know that an address is anycast."},{"id":"f","text":"Every router on the Internet must hold a separate host route for it, wherever members sit.","whyNot":"The host route is required only within the region of the prefix; outside it the address may be aggregated, except in the worst case of a global set."}],"answer":"ab","explanation":"An anycast address is assigned to several interfaces, and each packet goes to the nearest one by routing distance; its format is an ordinary unicast format (RFC 4291, Section 2.6).","concept":"An anycast address is assigned to several interfaces, and each packet goes to the nearest one by routing distance; its format is an ordinary unicast format (RFC 4291, Section 2.6).","rev":1,"updatedAt":"2026-10-26","kind":"scenario","takeaways":["Anycast is a routing choice made per destination, not a special address format."]},{"id":"networkingadvanced-002","courseId":"networking-advanced","domain":"ip-layer","topic":"IPv6 router fragmentation","prompt":"A route to a server crosses three links with MTUs of 9000, 1500 and 1400 bytes. Under the IPv6 definitions, what is the path MTU of this route?","hint":"Think about the one link that limits every packet on the whole route.","explain":"Each link can carry packets only up to its own size limit, and a route must work for every link it crosses. Consider which single link sets the size a sender can rely on across the whole route.","source":"https://www.rfc-editor.org/rfc/rfc8200.html#section-4.5","interaction":{"type":"single"},"options":[{"id":"a","text":"9000 bytes","whyNot":"The largest link is not the constraint; a packet that big would not fit the smaller links on the route."},{"id":"b","text":"1500 bytes","whyNot":"That is only the first link's MTU, but the path MTU considers every link on the route."},{"id":"c","text":"1400 bytes","why":"The path MTU is the minimum link MTU across all links between source and destination, which is the 1400-byte link."},{"id":"d","text":"3967 bytes","whyNot":"An average of the links means nothing, since a packet must fit every link it crosses."}],"answer":"c","explanation":"IPv6 defines path MTU as the minimum of the link MTUs along the path between a source and a destination node (RFC 8200, Section 2).","concept":"IPv6 defines path MTU as the minimum of the link MTUs along the path between a source and a destination node (RFC 8200, Section 2).","rev":1,"updatedAt":"2026-10-26","kind":"scenario","takeaways":["Path MTU is set by the smallest link on the route, not by the first or largest one."]},{"id":"networkingadvanced-003","courseId":"networking-advanced","domain":"ip-layer","topic":"IPv6 minimum link MTU","prompt":"For TCP over IPv6, the maximum segment size is computed as the maximum packet size minus 60 octets.","hint":"Think about how the size of the network header differs between IP versions.","explain":"An upper-layer protocol must leave room for the network header before sizing its data. Header lengths differ between IP versions, so one largest packet carries a different amount of transport data.","source":"https://www.rfc-editor.org/rfc/rfc8200.html#section-5","interaction":{"type":"true-false"},"options":[{"id":"a","text":"True","why":"The IPv6 header is 20 octets longer than the IPv4 header, so the MSS subtracts 60 octets, not 40."},{"id":"b","text":"False","whyNot":"Subtracting the IPv4 figure of 40 ignores the longer IPv6 header, so the segments would be too big."}],"answer":"a","explanation":"Over IPv6 the MSS must be the maximum packet size minus 60 octets, since the minimum IPv6 header is 20 octets longer than the minimum IPv4 header (RFC 8200, Section 8.3).","concept":"Over IPv6 the MSS must be the maximum packet size minus 60 octets, since the minimum IPv6 header is 20 octets longer than the minimum IPv4 header (RFC 8200, Section 8.3).","rev":1,"updatedAt":"2026-10-26","kind":"scenario","takeaways":["TCP over IPv6 needs a 60-octet header allowance, not the IPv4 figure of 40."]},{"id":"networkingadvanced-004","courseId":"networking-advanced","domain":"ip-layer","topic":"Happy Eyeballs connection delay","prompt":"A client looks up a host name and gets an A answer first, while the AAAA answer is still pending. How long should it wait for the AAAA answer before starting the IPv4 connection?","hint":"A short window, long enough for a second answer to arrive.","explain":"Two lookups for one name can come back in either order. The client should prefer one family, but a slow answer must not hold up the connection for long.","source":"https://www.rfc-editor.org/rfc/rfc8305.html#section-5","interaction":{"type":"single"},"options":[{"id":"a","text":"50 milliseconds","why":"RFC 8305 recommends a Resolution Delay of 50 milliseconds before the client proceeds on IPv4 alone."},{"id":"b","text":"2 seconds","whyNot":"Two seconds is the Last Resort Local Synthesis Delay, a much longer timer for hostnames with broken AAAA records."},{"id":"c","text":"Not at all; start IPv4 at once","whyNot":"The client should wait briefly so a late AAAA answer can still take the preference for IPv6."},{"id":"d","text":"Until the AAAA answer arrives, however long it takes","whyNot":"The wait is bounded by the Resolution Delay, not open-ended."}],"answer":"a","explanation":"Happy Eyeballs v2 sets the Resolution Delay at 50 milliseconds: if an IPv4 answer arrives first, the client waits that long for the IPv6 answer (RFC 8305, Section 3).","concept":"Happy Eyeballs v2 sets the Resolution Delay at 50 milliseconds: if an IPv4 answer arrives first, the client waits that long for the IPv6 answer (RFC 8305, Section 3).","rev":1,"updatedAt":"2026-10-26","kind":"scenario","takeaways":["Give the preferred answer a short head start, never an open-ended wait."]},{"id":"networkingadvanced-005","courseId":"networking-advanced","domain":"ip-layer","topic":"Address family interleaving","prompt":"A client has resolved several IPv4 and IPv6 addresses for one host and has sorted them into an order. Why should the ordered list keep addresses from both families, rather than only the top one of each?","hint":"Consider which candidate addresses the client knows about when it begins connecting.","explain":"A client opens connections one at a time, pausing briefly before every new one, and stops at the first success. How far it can get depends on which candidates were placed in its sequence beforehand.","source":"https://www.rfc-editor.org/rfc/rfc8305.html#section-4","interaction":{"type":"single"},"options":[{"id":"a","text":"So staggered attempts race across the whole ordered list, not just its top entries from each family.","why":"Racing the whole list gives the staggered effect across every candidate, not only the first address of each family."},{"id":"b","text":"So the client can skip Destination Address Selection for the rest of the list.","whyNot":"Sorting must still happen first, using Destination Address Selection, before any attempt starts."},{"id":"c","text":"So IPv4 entries are always tried before any IPv6 entry.","whyNot":"Nothing about keeping both families puts IPv4 ahead; the client's preference policy favors IPv6."},{"id":"d","text":"Because a DNS answer never holds more than one address per family.","whyNot":"Answers can hold several addresses, which is why the client sorts everything it has received so far."}],"answer":"a","explanation":"The client sorts all addresses received so far, and staggered racing then covers the whole sorted list across both families, not just the first of each (RFC 8305, Section 4).","concept":"The client sorts all addresses received so far, and staggered racing then covers the whole sorted list across both families, not just the first of each (RFC 8305, Section 4).","rev":1,"updatedAt":"2026-10-26","kind":"scenario","takeaways":["Keeping both families in the ordered list lets racing reach past the first address of each."]}]