Today: Internet Services. Internet Services. Typical Workload (Web Pages) Web caches (proxy server)

Transcription

1 Today: Internet Services Internet Services COMP5213: Computer and Network Organiza;on Web Caching Domain Name System Content Distribu;on Network Web Services Instructors: Javid Taheri 1! 2! Typical Workload (Web Pages) Web caches (proxy ) Mul;ple (typically small) objects per page File sizes Heavy- tailed Pareto distribu;on for tail Lognormal for body of distribu;on Goal: satisfy client request without involving origin! user sets browser: Web accesses via cache browser sends all HTTP requests to cache object in cache: cache returns object else cache requests object from origin, then returns object to client client HTTP request HTTP response HTTP request HTTP response Proxy HTTP request HTTP response origin client origin 3! 4!

2 More about Web caching Caching example (1) Cache acts as both client and Cache can do up- to- date check using If-modified-since HTTP header Issue: should cache take risk and deliver cached object without checking? Heuris;cs are used. Typically cache is installed by ISP (university, company, residen;al ISP) Why Web caching? Reduce response ;me for client request. Reduce traffic on an ins;tu;on s access link. Internet dense with caches enables poor content providers to effec;vely deliver content Assump;ons average object size = 100,000 bits avg. request rate from ins;tu;on s browser to origin serves = 15/sec delay from ins;tu;onal router to any origin and back to router = 2 sec Consequences u;liza;on on LAN = 15% u;liza;on on access link = 100% total delay = Internet delay + access delay + LAN delay = 2 sec + minutes + milliseconds institutional network public Internet 1.5 Mbps access link 10 Mbps LAN origin s institutional cache 5! 6! Caching example (2) Caching example (3) Possible solu;on increase bandwidth of access link to, say, 10 Mbps Consequences u;liza;on on LAN = 15% u;liza;on on access link = 15% Total delay = Internet delay + access delay + LAN delay = 2 sec + msecs + msecs o_en a costly upgrade institutional network public Internet 10 Mbps access link 10 Mbps LAN origin s Install cache suppose hit rate is.4 Consequence 40% requests will be sa;sfied almost immediately 60% requests sa;sfied by origin u;liza;on of access link reduced to 60%, resul;ng in negligible delays (say 10 msec) total delay = Internet delay + access delay + LAN delay =.6*2 sec +.6*.01 secs + milliseconds < 1.3 secs institutional network public Internet 1.5 Mbps access link 10 Mbps LAN origin s institutional cache institutional cache 7! 8!

3 Problems Caching Proxies Sources for Misses Over 50% of all HTTP objects are uncacheable why? Not easily solvable Dynamic data à stock prices, scores, web cams CGI scripts à results based on passed parameters Obvious fixes SSL à encrypted data is not cacheable Most web clients don t handle mixed pages well à many generic objects transferred with SSL Cookies à results may be based on passed data Hit metering à owner wants to measure # of hits for revenue, etc. Capacity How large a cache is necessary or equivalent to infinite On disk vs. in memory à typically on disk Compulsory First ;me access to document Non- cacheable documents CGI- scripts Personalized documents (cookies, etc) Encrypted data (SSL) Consistency Document has been updated/expired before reuse Conflict No such misses 9! 10! Today: Internet Services Web Caching Domain Name System Content Distribu;on Network Web Services DNS: Domain Name System People: many iden;fiers: name, passport # Internet hosts, routers: IP address (32 bit) - used for addressing datagrams name, e.g., - used by humans Q: map between IP addresses and name? Domain Name System: distributed database implemented in hierarchy of many name s applica1on- layer protocol host, routers, name s to communicate to resolve names (address/name transla;on) note: core Internet func;on, implemented as applica;on- layer protocol complexity at network s edge 11! 12!

4 Hierarchical Name Space Hierarchical Name Space (cont) shade.ug.cs.usyd.edu.au! root org cn net edu com au gov edu com net unsw usyd uts anu cs ee ug Each node in hierarchy stores a list of names that end with same suffix Suffix = path up tree E.g., given this tree, where would following be stored: cnn.com mit.edu cpu0.cs.usyd.edu.au shade.ug.cs.usyd.edu.au! root cn edu com au org net gov edu com net unsw usyd uts anu cs ee ug Zone = con;guous sec;on of name space E.g., Complete tree, single node or subtree A zone has an associated set of name s Must store list of names and tree links Subtree shade 13! shade Single node Complete Tree 14! DNS Design Zones are created by convincing owner node to create/ delegate a subzone Records within zone stored mul;ple redundant name s Primary/master name updated manually Secondary/redundant s updated by zone transfer of name space Zone transfer is a bulk transfer of the configura;on of a DNS uses TCP to ensure reliability Example: CS.USYD.EDU.AU created by USYD.EDU.AU administrators Who creates USYD.EDU.AU? DNS name s Why not centralize DNS? single point of failure traffic volume distant centralized database maintenance doesn t scale! no has all name- to- IP address mappings local name s: each ISP, company has local (default) name host DNS query first goes to local name authorita;ve name : for a host: stores that host s IP address, name can perform name/address transla;on for that host s name 15! 16!

5 DNS: Root name s contacted by local name that can not resolve name root name : contacts authorita;ve name if name mapping not known gets mapping returns mapping to local name e NASA Mt View, CA f Internet Software C. Palo Alto, CA b USC-ISI Marina del Rey, CA l ICANN Marina del Rey, CA a NSI Herndon, VA c PSInet Herndon, VA d U Maryland College Park, MD g DISA Vienna, VA h ARL Aberdeen, MD j NSI (TBD) Herndon, VA k RIPE London i NORDUnet Stockholm m WIDE Tokyo 13 root name s worldwide! Typical Resolu;on Client Local DNS NS metro.ucc.usyd.edu.au" NS crux.cs.usyd.edu.au" A www=ipaddr root & au DNS metro.ucc.usyd.edu.au DNS crux.cs.usyd.edu.au DNS 17! 18! Typical Resolu;on Steps for resolving Applica;on calls gethostbyname() (RESOLVER) Resolver contacts local name (S 1 ) S 1 queries root (S 2 ) for ( S 2 returns NS record for usyd.edu.au (S 3 ) What about A record for S 3? This is what the addi;onal informa;on sec;on is for (PREFETCHING) S 1 queries S 3 for S 3 returns A record for Can return mul;ple A records! what does this mean? Lookup Methods Recursive query: Server goes out and searches for more info (recursive) Only returns final answer or not found Itera;ve query: Server responds with as much as it knows (itera;ve) I don t know this name, but ask this Workload impact on choice? Local typically does recursive Root/distant does itera;ve local name dns.eurecom.fr requesting host surf.eurecom.fr root name iterative query intermediate name metro.ucc.usyd.edu.au 5 6 authoritative name crux.cs.usyd.edu.au shade.ug.cs.usyd.edu.au 19! 20!

6 DNS: caching and upda;ng records DNS records once (any) name learns mapping, it caches mapping cache entries ;meout (disappear) aher some ;me update/no;fy mechanisms under design by IETF RFC 2136 hjp:// 21! DNS: distributed db storing resource records (RR) Type=A! RR format: (name, value, type, class, ttl) n name is hostname! n value is IP address! Type=CNAME! n name is alias name for some canonical (the real) name!! is really Type=NS east.backup2.ibm.com name is domain (e.g. foo.com) value is IP address of authorita;ve name n value is canonical name! for this domain! Type=MX! n value is name of mail associated with name!! 22! DNS protocol, messages DNS protocol, messages DNS protocol : query and reply messages, both with same message format msg header! identification: 16 bit # for query, reply to query uses same #! flags:! n query or reply! n reply is authoritative! n recursion desired! n recursion available! Name, type fields for a query RRs in response to query records for authoritative s additional helpful info that may be used 23! 24!

7 Workload and Caching Subsequent Lookup Example What workload do you expect for different s? Why might this be a problem? How can we solve this problem? DNS responses are cached Quick response for repeated transla;ons Other queries may reuse some parts of lookup NS records for domains DNS nega;ve queries are cached Don t have to repeat past mistakes E.g. misspellings, search strings in resolv.conf Cached data periodically ;mes out Life;me (TTL) of data controlled by owner of data TTL passed with every record Client ftp.it.usyd.edu.au Local DNS ftp.it.usyd.edu.au ftp=ipaddr root & au DNS usyd.edu.au DNS it.usyd.edu.au DNS 25! 26! Reliability Prefetching DNS s are replicated Name service available if one replica is up Queries can be load balanced between replicas Try alternate s on ;meout What s a good value for a ;meout? Hard to tell à what are the tradeoffs? Bejer be conserva;ve! Exponen;al backoff when retrying same Why do we need this? Same iden;fier for all queries Don t care which responds Name s can add addi;onal data to any response Typically used for prefetching CNAME/MX/NS typically point to another host name Responses include address of host referred to in addi;onal sec;on 27! 28!

8 DNS: Summary Today: Internet Services Mo;va;ons! large distributed database Scalability Independent update Robustness Hierarchical database structure Zones How is a lookup done Web Caching Domain Name System Content Distribu;on Network Web Services Caching/prefetching and TTLs 29! 30! Components in a CDN Content distribu;on networks (CDNs) aaa.com! bbb.com! ccc.com! Cache! Backend Servers!!! Geographically distributed surrogate s!! The content providers are the CDN customers. Content replica;on CDN company installs hundreds of CDN s throughout Internet in lower- ;er ISPs, close to users CDN replicates its customers content in CDN s. When provider updates content, CDN updates s origin in North America CDN distribution node! Redirectors!! Clients! CDN in S. America CDN in Europe CDN in Asia 31! 32!

9 CDN example origin distributes HTML Replaces: Origin CDNs authoritative DNS Nearby CDN with ruth.gif HTTP request for DNS query for HTTP request for CDN company! cdn.com! distributes gif files! uses its authoritative DNS to route redirect requests! More about CDNs rou;ng requests not just Web pages CDN creates a map, indica;ng streaming stored audio/video distances from leaf ISPs and CDN nodes streaming real- ;me audio/video when query arrives at authorita;ve DNS : CDN nodes create applica;on- layer overlay network determines ISP from which query originates uses map to determine best CDN 33! 34! How Akamai Works How Akamai Works Clients fetch html document from primary E.g. fetch index.html from cnn.com URLs for replicated content are replaced in html E.g. <img src= hjp://cnn.com/af/x.gif > replaced with <img src= hjp:// a73.g.akamaitech.net/7/23/cnn.com/af/x.gif > Client is forced to resolve axyz.g.akamaitech.net hostname How is content replicated? Akamai only replicates sta;c content Modified name contains original file name Akamai is asked for content First checks local cache If not in cache, requests file from primary and caches file 35! 36!

10 How Akamai Works Akamai Subsequent Requests cnn.com (content provider) DNS root Akamai cnn.com (content provider) DNS root Akamai 11 Get index.ht ml End- user Get foo.jpg Get /cnn.com/foo.jpg Akamai high-level DNS Akamai low-level DNS Nearby matching Akamai Get index.h tml 1 2 Akamai high-level DNS End- user Get /cnn.com/foo.jpg 10 Akamai low-level DNS Nearby matching Akamai 37! 38! Today: Internet Services Web Caching Domain Name System Content Distribu;on Network Web Services Web History Consider a future device for individual use, which is a sort of mechanized private file and library. It needs a name, and to coin one at random, "memex" will do. A memex is a device in which an individual stores all his books, records, and communications, and which is mechanized so that it may be consulted with exceeding speed and flexibility. It is an enlarged intimate supplement to his 1945: memory.! Vannevar Bush, As we may think, Atlan;c Monthly, July, Describes the idea of a distributed hypertext system. A memex that mimics the web of trails in our minds. 39! 40!

11 Web History Web History (cont) 1989: Tim Berners- Lee (CERN) writes internal proposal to develop a distributed hypertext system. Connects a web of notes with links. Intended to help CERN physicists in large projects share and manage informa;on 1990: Tim BL writes a graphical browser for Next machines NCSA released 26 WWW s worldwide 1993 Marc Andreessen releases first version of NCSA Mosaic browser Mosaic version released for (Windows, Mac, Unix). Web (port 80) traffic at 1% of NSFNET backbone traffic. Over 200 WWW s worldwide Andreessen and colleagues leave NCSA to form Mosaic Communica;ons Corp (predecessor to Netscape). 41! 42! Internet Hosts Web Servers Clients and s communicate using the HyperText Transfer Protocol (HTTP) Client and establish TCP connec;on Client requests content Server responds with requested content Client and close connec;on (eventually) Current version is HTTP/1.1 RFC 2616, June, HTTP request Web! client! (browser)! HTTP! TCP! IP! HTTP response (content) Web content Streams Datagrams Web!! How many of the 2 32 IP addresses have registered domain names? 43! 44!

12 Web Content Sta;c and Dynamic Content Web s return content to clients content: a sequence of bytes with an associated MIME (Mul;purpose Internet Mail Extensions) type Example MIME types text/html HTML document text/plain Unformajed text application/postscript Postcript document image/gif Binary image encoded in GIF format image/jpeg Binary image encoded in JPEG format The content returned in HTTP responses can be either sta1c or dynamic. Sta4c content: content stored in files and retrieved in response to an HTTP request Examples: HTML files, images, audio clips. Request iden;fies content file Dynamic content: content produced on- the- fly in response to an HTTP request Example: content produced by a program executed by the on behalf of the client. Request iden;fies file containing executable code Borom line: All Web content is associated with a file that is managed by the. 45! 46! URLs Each file managed by a has a unique name called a URL (Universal Resource Locator) URLs for sta;c content: Iden;fies a file called index.html, managed by a Web at that is listening on port 80. URLs for dynamic content: Iden;fies an executable file called proc, managed by a Web at that is listening on port 8000, that should be called with two argument strings: and 213. How Clients and Servers Use URLs Example URL: Clients use prefix ( to infer: What kind of to contact (Web ) Where the is ( What port it is listening on (80) Servers use suffix (/index.html) to: Determine if request is for sta;c or dynamic content. No hard and fast rules for this. Conven;on: executables reside in cgi-bin directory Find file on file system. Ini;al / in suffix denotes home directory for requested content. Minimal suffix is /, which all s expand to some default home page (e.g., index.html). 47! 48!

13 HTTP Requests HTTP Requests (cont) HTTP request is a request line, followed by zero or more request headers Request line: <method> <uri> <version> <version> is HTTP version of request (HTTP/1.0 or HTTP/1.1) <uri> is typically URL for proxies, URL suffix for s. A URL is a type of URI (Uniform Resource Iden;fier) See hjp:// <method> is either GET, POST, OPTIONS, HEAD, PUT, DELETE, or TRACE. 49! HTTP methods: GET: Retrieve sta;c or dynamic content Arguments for dynamic content are in URI Workhorse method (99% of requests) POST: Retrieve dynamic content Arguments for dynamic content are in the request body OPTIONS: Get or file ajributes HEAD: Like GET but no data in response body PUT: Write a file to the! DELETE: Delete a file on the! TRACE: Echo request in response body Useful for debugging. Request headers: <header name>: <header data> Provide addi;onal informa;on to the. 50! HTTP Versions Major differences between HTTP/1.1 and HTTP/1.0 HTTP/1.0 uses a new connec;on for each transac;on. HTTP/1.1 also supports persistent connec4ons mul;ple transac;ons over the same connec;on Connection: Keep-Alive HTTP/1.1 requires HOST header Host: Makes it possible to host mul;ple websites at single Internet host HTTP/1.1 supports chunked encoding (described later) Transfer- Encoding: chunked HTTP/1.1 adds addi;onal support for caching 51! HTTP Responses HTTP response is a response line followed by zero or more response headers. Response line: <version> <status code> <status msg> <version> is HTTP version of the response. <status code> is numeric status. <status msg> is corresponding English text. 200 OK Request was handled without error 301 Moved Provide alternate URL 403 Forbidden Server lacks permission to access file 404 Not found Server couldn t find the file. Response headers: <header name>: <header data> Provide addi;onal informa;on about response Content-Type: MIME type of content in response body. Content-Length: Length of content in response body. 52!

14 GET Request to Apache Server From Firefox Browser URI is just the suffix, not the entire URL! GET /~bryant/test.html HTTP/1.1 Host: User-Agent: Mozilla/5.0 (Windows; U; Windows NT 6.0; en-us; rv: ) Gecko/ Firefox/ Accept: text/html,application/xhtml+xml,application/xml;q=0.9,*/ *;q=0.8 Accept-Language: en-us,en;q=0.5 Accept-Encoding: gzip,deflate Accept-Charset: ISO ,utf-8;q=0.7,*;q=0.7 Keep-Alive: 115 Connection: keep-alive CRLF (\r\n) GET Response From Apache Server HTTP/ OK Date: Fri, 29 Oct :48:32 GMT Server: Apache/ (Unix) mod_ssl/ OpenSSL/0.9.7m mod_pubcookie/3.3.2b PHP/5.3.1 Accept-Ranges: bytes Content-Length: 479 Keep-Alive: timeout=15, max=100 Connection: Keep-Alive Content-Type: text/html <html> <head><title>some Tests</title></head> <body> <h1>some Tests</h1>... </body> </html> 53! 54! Serving Dynamic Content Serving Dynamic Content (cont) Client sends request to. If request URI contains the string /cgi-bin, then the assumes that the request is for dynamic content. GET /cgi-bin/env.pl HTTP/1.1 Client Server The creates a child process and runs the program iden;fied by the URI in that process Client Server env.pl fork/exec! 55! 56!

15 Serving Dynamic Content (cont) Issues in Serving Dynamic Content The child runs and generates the dynamic content. The captures the content of the child and forwards it without modifica;on to the client Client Content! Server env.pl Content! How does the client pass program arguments to the? How does the pass these arguments to the child? How does the pass other info relevant to the request to the child? How does the capture the content produced by the child? These issues are addressed by the Common Gateway Interface (CGI) specifica;on. Client Request! Content! Server Content! Create! env.pl 57! 58! CGI Because the children are wriren according to the CGI spec, they are o_en called CGI programs. The add.com Experience input URL! host! port! CGI program!args! Because many CGI programs are wriren in Perl, they are o_en called CGI scripts. However, CGI really defines a simple standard for transferring informa;on between the client (browser), the, and the child process. Output page! 59! 60!

16 Serving Dynamic Content With GET Serving Dynamic Content With GET Ques;on: How does the client pass arguments to the? Answer: The arguments are appended to the URI Can be encoded directly in a URL typed to a browser or a URL in an HTML link adder is the CGI program on the that will do the addi;on. argument list starts with? arguments separated by & spaces represented by + or %20 URI o_en generated by an HTML form <FORM METHOD=GET ACTION="cgi-bin/adder"> <p>x <INPUT NAME="n1"> <p>y <INPUT NAME="n2"> <p><input TYPE=submit> </FORM> URL: cgi-bin/adder?n1=15213&n2=18243 Result displayed on browser: Welcome to add.com: THE Internet addition portal. The answer is: > 33456! Thanks for visiting!! 61! 62! Serving Dynamic Content With GET Addi;onal CGI Environment Variables Ques;on: How does the pass these arguments to the child? Answer: In environment variable QUERY_STRING A single string containing everything aher the? For add: QUERY_STRING = n1=15213&n2=18243 General SERVER_SOFTWARE SERVER_NAME GATEWAY_INTERFACE (CGI version) Request- specific SERVER_PORT REQUEST_METHOD (GET, POST, etc) QUERY_STRING (contains GET args) REMOTE_HOST (domain name of client) REMOTE_ADDR (IP address of client) CONTENT_TYPE (for POST, type of data in message body, e.g., text/ html) CONTENT_LENGTH (length in bytes) 63! 64!

17 Even More CGI Environment Variables In addi;on, the value of each header of type type received from the client is placed in environment variable HTTP_type Examples (any - is changed to _ ) : HTTP_ACCEPT HTTP_HOST HTTP_USER_AGENT THANK YOU Ques;ons? 65! 66!