HTTP 1.0

• TCP-Port 80
• PULL - Protocol
• Request / Response
• Stateless Protocol
• Non-Persistent Connections
• Supported Methods (GET, HEAD, POST)

Operation

Simplified Operation

General Operation

Using Caches

Request format

Response FORMAT

Example - Requests I

//Request
GET /wireshark-labs/HTTP-wireshark-file1.html HTTP/1.0
Host: gaia.cs.umass.edu

//Response
HTTP/1.0 200 OK
Date: Tue, 05 Nov 2013 09:37:56 GMT
Server: Apache/2.2.3 (CentOS)
Last-Modified: Tue, 05 Nov 2013 09:37:01 GMT
Content-Length: 128
Content-Type: text/html


Congratulations.  You've downloaded the file 
http://gaia.cs.umass.edu/wireshark-labs/HTTP-wireshark-file1.html!

Example requests II

//Request
HEAD /wireshark-labs/HTTP-wireshark-file1.html HTTP/1.0
Host: gaia.cs.umass.edu

//Response
HTTP/1.0 200 OK
Date: Tue, 05 Nov 2013 09:37:56 GMT
Server: Apache/2.2.3 (CentOS)
Last-Modified: Tue, 05 Nov 2013 09:37:01 GMT
Content-Length: 128
Content-Type: text/html

Questions / Did i miss something?

HTTP/1.0

HTTP/1.1

What has changed?

• Persistent connections
• New Methods
• HTTP-Pipelining
• Transfer Encoding
• Partial GET
• Better control over caches
• Validators
• 1** - Status Codes are used now

Persistent connections I

Create TCP-Connection ONCE and use it for several request/response pairs. (Reduces Load on Network and gets Results faster)

Usage of Header 'Connection: keep-alive' in request indicates the intention to use a persistent connection

Servers have the last word, only if Header 'Connection: keep-alive' is included in request ==> connection is persistent now

PERSISTENT CONNECTIONS II

GET /wireshark-labs/HTTP-wireshark-file1.html HTTP/1.1
Host: gaia.cs.umass.edu
Connection: keep-alive
Accept: text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,*/*
User-Agent: Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/30.0.1599.114 Safari/537.36
Accept-Encoding: gzip,deflate,sdch
Accept-Language: en-US,en;q=0.8
If-None-Match: "8734d-80-c649c940"
If-Modified-Since: Tue, 05 Nov 2013 09:37:01 GMT    
    

HTTP/1.1 200 OK
Date: Tue, 05 Nov 2013 09:59:49 GMT
Server: Apache/2.2.3 (CentOS)
Last-Modified: Tue, 05 Nov 2013 09:59:01 GMT
ETag: "8734d-80-14f76340"
Accept-Ranges: bytes
Content-Length: 128
Keep-Alive: timeout=10, max=100
Connection: Keep-Alive
Content-Type: text/html; charset=UTF-8

Congratulations.  You've downloaded the file http://gaia.cs.umass.edu/wireshark-labs/HTTP-wireshark-file1.html!

persistent connection III

//Request
GET / HTTP/1.1
Host: www.orf.at
Connection: keep-alive
Accept: text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,*/*
User-Agent: Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/30.0.1599.114 Safari/537.36
Accept-Encoding: gzip,deflate,sdch
Accept-Language: en-US,en;q=0.8    

//Response
HTTP/1.1 200 OK
Date: Tue, 05 Nov 2013 10:08:50 GMT
Server: Jetty(6.1.22)
X-Cache: HIT from localhost
ETag: "XOyamG4ZEJIhX3vZRY0UvA=="
Content-Length: 15805
Content-Encoding: gzip
Content-Type: text/html; charset=utf-8
Accept-Ranges: none
Connection: close

NEW Methods

• PUT (make content available under given URL)


• DELETE (remove [content of] given URL)


• OPTIONS (display options available on URL)


• TRACE (request - loopback)


• CONNECT (set proxy behavior to 'tunnel')

http-pipelining

Client can issue several requests in a row without having to wait for each response (Speedup)

Transfer encoding

GET / HTTP/1.1
User-Agent: Wget/1.13.3 (linux-gnu)
Accept: */*
Host: homegateway.lan
Connection: Keep-Alive

HTTP/1.1 200 OK
Connection: close
Date: Thu, 15 Sep 2011 08:48:47 GMT
Transfer-Encoding: chunked
Content-Type: text/html

//Body
//CHUNK1
//CHUNK2
//CHUNK3
//CHUNK4

PARTIAL GET

Request only part of a Object (saves Bandwidth)

//Request
GET /examples/dog.jpg HTTP/1.1
Host: www.someschool.edu
Range: bytes=0-999

HTTP/1.0 206 Partial Content
Date: Mon, 05 May 2008 00:36:57 GMT
Server: Apache/2.0.52 (Red Hat)
Accept-Ranges: bytes
Content-Length: 1000
Content-Range: bytes 0-999/3980
Content-Type: image/jpeg  

binary data.....binary data...

NEW header cachecontrol

//cache-request-directives

"no-cache"
"no-store"
"max-age" "=" delta-seconds
"max-stale" [ "=" delta-seconds ]
"min-fresh" "=" delta-seconds
"no-transform"
"only-if-cached"

//cache-response-directives 

"public"
"private" [ "=" <"> 1#field-name <"> ] 
"no-cache" [ "=" <"> 1#field-name <"> ]
"no-store"
"no-transform"
"must-revalidate"
"proxy-revalidate"

Validators

HTTP/1.0 had only the field 'LastModified' to check if a cached website is still fresh

HTTP/1.1 introduces ETag as new Validator and a bunch of new Header-Fields for conditional GET

cond. GET - HeaderFields

• If-Modified-Since
• If-Unmodified-Since
• If-Match
• If-Non-Match
• If-Range

Questions / Did i miss something?

HTTP/1.1

HTTP/2.0

• Why?
• Asynchronous connection multiplexing
• Change of Message-format
• Stream
• Header Compression
• HTTP-FlowControl
• Serverside push
• Development-Status

Source: http://tools.ietf.org/html/draft-ietf-httpbis-http2-04

WHY?

Today, important features of HTTP/1.1 like persistent connections and pipelining are used only sparsely because often, not all hosts on the request/response chain support it.(e.g proxys)

HTTP/2.0 is a way to demand that all participants  support those features

It will include a broad use of security-mechanisms like encryption

It reduces the number of TCP-Connections needed  by using individual ones more efficiently (connection multiplexing)

Connection Multiplexing

This is achieved by letting independent Streams (>100) share one HTTP/2.0 connection

The streams allow a concurrent execution of different requests/responses

FlowControl is present for the DATA-frames of each stream and for the connection as well

Change of Message-format

No more Text, instead binary encoded frames

Smallest unit of transfer is a 'Frame'

• size of header 8 octets
• payload max  65 535 octets
• type of frame determined by value of field 'type'
• separate frames for header / body
• every message is assigned to a stream

CHANGE of Message-Format

Different types of frames

STREAM

Encapsulates one request-response pair

1. The client opens a stream to the server and sends HEADER-FRAMES, the last one carries a flag denoting the end of header
2. Then, the client sends the DATA-Frames and closes the Frame on his side (analogous to Pipe in OS)
3. Now, that server knows that the request is finished, he gets the required objects and sends the frames (HEADER and DATA) to the client
4. Finally, the server closes the stream also on his side

Each stream has a state (idle, reserved, open, halfclosed, closed)

Header-Compression

Today - HTTP-Headers can be relatively large and redundant, the goal is to transmit them more efficiently

• HTTP Header Fields and Values are serialized and stored in the body of HEADER-frames


• If one frame doesn't suffices , HTTP Headers can be transported with multiple HEADER-frames

FLOW-CONTROL

• Credit - based System, receivers advertise how many bytes they can receive at the moment


• There is an initial value, when the reciever processes data, it adds the new free space to the amount of bytes the Sender can send


• The reciever does so by sending WINDOW_UPDATE events to the sender


• FlowControl only applies to DATA-frames

Serverside-push

After receiving a certain request, a server can decide that it is the best to send neccessary files without being asked for it

That way, a server can speed up the transfer of Objects by sending them without request if it is inevitable to send them

1. Server gets a request and he deems it necessary to send files proactively
2. Server sends a PUSH_PROMISE-frame containing the details (streamid,..) of the stream he wants to initiate
3. Client prepares to recieve from given streamid, ...

This way, mediafiles can be present at the client before he starts to realize that he has to request them, saving precious bandwith

DEVELOPMENT STATUS

Questions / Did i miss something?

HTTP/2.0

SMTP

• Simple Mail Transfer Protocol


• PUSH - Protocol


• Uses TCP - Port 25


• between useragent and mailserver


• between mailserver and mailserver


• 7 bit ASCII

SMTP Example

telnet mailsrv.uni.at 25
Trying 123.204.666.43...
Connected to mailsrv.uni.at.
Escape character is '^]'.
220 mailsrv.at ESMTP Postfix
HELO foobar.example.net
250 mailsrv.uni.at
MAIL FROM: 
250 2.1.0 Ok
RCPT TO: 
250 2.1.5 Ok
DATA

354 End data with .
From: user@edu.uni.at
To: user@gmail.com
Subject: Vorzeigeeffekt

Hi, 
don't panic
sincerely you
Lg
.
250 2.0.0 Ok: queued as CE22E160A77    

Questions / Did i miss something?

SMTP

POP 3

• Post Office Protocol (Version 3)
• Mail-Access Protocol, used to retrieve Emails from user's Inbox on Mailserver
• Uses TCP Port 110
• Server doesn't maintains state between sessions

3 phases of interaction

1. authorization with username and password
2. transaction, where the UserAgent retrieves Emails and can mark them for deletion
3. update, server commits changes when UserAgent issues 'quit'

POP3 - Example

C: telnet mailServer 110
S: +OK POP3 server ready
C: user bob
C: pass swordfish
S: +OK user successfully logged on

C: list
S: 1 498
S: 2 912
S: .
C: retr 1
S: (blah blah ...
S: ..............
S: ........blah)
S: .
C: dele 2
C: quit
S: +OK POP3 server signing off

POP 3 - PRO/CON

+

• POP 3 is rather simple to understand and implement
• 'download and delete' -> Mails stuck on a single device

-

• lacks functionality like managing remote folders

Questions / Did i miss something?

POP 3

IMAP

• Mail Access Protocol
• Uses TCP Port 143
• More Features than POP 3 but much more complex
• Allows UserAgent to retrieve only parts of Emails 
• Every Mail is assigned to a folder
• Received Mail is in the Folder INBOX by default
• Users can move Mails between folders, create, rename and remove folders
• Folders can be searched for Mails
• Server maintains state between sessions (e.g. folder-structure)

Questions / Did i miss something?

IMAP

DNS

• Domain Name System


• Provides a core-functionality of the internet


• Hierachically structured distributed database


• Provides Translation between Hostname and IP-Address


• Uses UDP - Port 53


• Also used for Host-Aliasing, Mail-Aliasing, Load-Distribution

DNS - Involved Servers

Root DNS Server: Map the top level domain - part of the hostname to Ip-Adresses of TLD-servers

Top-level domain servers (TLD): Map the domainName - part of the hostname to the IP-Address of the responsible Authoritative Servers

Authoritative DNS server: Knows the translation of Hostname -> IP-Adress of host in his domain

Local DNS server:  Acts like a proxy for DNS-Requests, first asked, caches responses

dns message-format

RESOURCERECORD - TYPES

(Name, Value, Type, TTL)

• Type=A:  (relay1.bar.foo.com, 145.37.93.126, A, 54321)


• Type=NS:  (foo.com, dns.foo.com, NS, 54321)


• Type=CNAME: (foo.com, relay1.bar.foo.com, CNAME, 54321)


• Type=MX: (foo.com, mail.bar.foo.com, MX, 54321)

Questions / Did i miss something?

DNS