Application Layer Flashcards
what are the two possible structures of network applications
peer to peer
client server
does the server ip address change
no
does the server ip address change
no
how can servers be scaled
farming out to data centres
can client ip addresses change
yes
what do clients always communicate with in client server architecture
servers
is there an always on server in peer to peer architecture
no
in peer to peer architecture, do the clients have to go through the server always
no
how do peers use each other in peer to peer architecture
request service from each other and provide service in return
as new peers are added to the peer to peer architechture, how does this affect the system
increases load
new peers bring new capacity to give service to each other but also more demands for service
do the clients always keep the same ip address in p2p archicecture
no
why is there being no centralised system a good feature of p2p architecture
no point of attack
harder to bring down
what is the client process
process that initiates communication
what is the server process
process that waits to be contacted
which processes are performed in p2p arch
clients can perform both client and server processes
examples of apps that could tolerate some loss
audio
examples of apps that need 100% reliable data transfer
file transfer, web applications
examples of apps that require low time delay to be effective
multi player games
examples of apps that require low time delay to be effective
multi player games
example of apps where timing delays are not important
emails
what is throughput
how much data you can put through the system
examples of apps that require minimum amount of throughput to be effective
multimedia
what does it mean if an app is elastic
make use of whatever throughput they get
what does URL stand for
uniform resource locator
what part of the url is the host name
www.someschool.edu/someDept/pic.gif
www.someschool.edu
what part of the url is the path name
www.someschool.edu/someDept/pic.gif
/someDept.pic.gif
does http use udp or tcp
tcp
how does HTTP use TCP
- client initiates connection to server
- server accepts tcp connection from client
- HTTPmessages exchanged
- tcp connection closed
what does it mean for a server to be a threading server
can have multiple clients connecting to it at once
what does it mean to say HTTP is stateless
server maintains no information about past clients requests
why is the HTTP stateless
state slows things down
adds to delays and requires lots of storage
what is non persistent HTTP
at most 1 object can be send over TCP connection
connection then closed
so 100 objects requires 100 connections and 100 closes
what is persistent http
multiple objects can be sent over a single TCP connection
what is the advantage of persitant HTTP over non persistent HTTP
only needs one connection set up so quicker
what is the default mode of HTTP
persistent connections with pipelining
non persistent HTTP process outline
- url entered
- TCP syn request from client
- server accepts syn request
- client sends request as piggyback on third part of handshake
- server responds
- close TCP connection
- finish connection
- repeat for each object
persistent HTTP process outline
- one tcp connection initiation
- leave this connection open so when further requests come, send them over the same open connection
what is the time taken to transfer in persistent HTTP
syn = 1 rtt
request = 1 rtt
file transmission time
2rtt + file transmission time
what are the two types of HTTP messages
request and response
why are http requests written in ascii
human readable
what does status code 200 mean
ok
request succeeded
object later in msg
what does status code 301 mean
permanently moved
request object moved, new location later in msg
what does status code 400 mean
bad request
msg not understood by server
what does status code 404 mean
not found
what does status code 505 mean
HTTP version not supported
uses of cookies
authorisation
shopping carts
recommendations
user session state
where are cookies stored
machine cookie cache
next time you visit a site you accepted cookies for, what happens
cookie sent along with the request
how can businesses use cookies to make money
demographic of users for advertising so they can still make money off you even if youre just browsing
using your info and giving it to third parties
what is the goal of a proxy serve
to satisfy client requests without involving origin server
how does a proxy server work
- user sets browser to access the web via proxy
- browser sends all requests to proxy server
- proxy server checks cache, if the object requested for is in the cache, it returns it, otherwise it goes to the origin server to request
- then this is stored in the cache
how does the cache act as both a client and server
server for original requesting client
client to origin server
why is web caching used
reduce response time for requests
reduce traffic on main server
why is a cache a better solution than making a fatter access link
cheaper
what does it mean if the cache hit rate is 40%
40% of the requests are satisfied at cache, the other 60% have to use the access link to go to the origin servers
what is the conditional GET
will get from cache if the modified last date is before a specified date
what does DNS stand for
domain name system
examples of top level domains
.co.uk
.com
.org
.ie
what type of hierarchy does DNS have
tree
what protocol is DNS ran over
UDP
what are canonical names
complicated names are required for servers but sometimes we want a simple naming scheme
these are canonical names
what are the DNS services
- host name to IP address translation
- host aliasing
- mail server aliasing
- load distibution
how does load distribution in DNS work
goes to a geographically closer server when redirected
what does the root name server do
returns list of IP addresses for responsible TLD servers
what does TLD stand for
top level domain
are the root servers often contacted
no
what is one level below top level domain servers
authoritative DNS servers
examples of organisations that would control their own dns servers (authorative dns servers)
google
amazon
trinity
example of college local DNS
scsss.tcd.ie
what are local DNS also called
default name server
path from root to local dns
root => tld => auth => local
what is an iterated query
if the contacted server does not have the answer, it replies with the name of the server to contact instead
what is a recursive query
if the contacted server does not have the answer, it contacts the next server
what does TTL stand for
time to live
what is resource record format
(Name, TTL, Type, Data)
what are the different types of resource records
A = ipv4
AAAA = ipv6
NS = name server eg domain (url)
CNAME = canonical name
MX = email exchange
what are some examples of the flags in a DNS message
query or repsonse
recursion desired
what are the two types of DNS attacks
DDoS
Redirect
what is a DDoS attack
bombard root servers with traffic
why have DDoS attacks not been successful to date
local DNS servers cache IPs of TLD servers allowing root server bypass
how do redirect attacks work
intercept queries
why is peer to peer file distribution quicker than client server model file distribution
peers can both upload and download so can propogate among the peers
server must sequentially send N copies of the file until all the nodes in the network have the file
bottleneck is minimum download rate of one of clients, so as more clients added, time gets slower
more clients in p2p brings more capacity to the service
what are files divided into in BitTorrent
chunks
what is a torrent
group of peers exchanging chunks of a file
what happens when a peer first joins the torrent
contact the tracker
gets list of peers and connects to a subset of them
peers who give more data become preferred peers
no chunks to begin with but will accumulate them over time from other peers
what is churn in torrent
when peers come and go in the system
what chunks do peers aim to get
the rarest ones in the system in order to help the other peerrs
how is the preferred peers system kept fair
re evaluated every 10 seconds
every 30 seconds randomly select another peer to start sending to in the hopes that it will send back
what happens in synchronous web applications when user wants to load new page
must wait for page to load
while this is happening, the user cannot interact
how does asynchronous web applications work
loads data in the background, so the page can update without a wait time
example of ajax trigger
mouse scroll to end of page
how does a typical ajax request work
user clicks event handler
XHR object generated
XHR object requests page from server in the background
when the data gets back the XHR object fires an event and displays the data
how does the websocket protocol differ from AJAX
the server does not have to wait for the client to send a request
web socket, the server can also talk to the client and they can speak quickly to each other via messages
both parties can start sending messages at any time
examples of uses of web sockets
smart plug
viewing ring doorbell footage from phone
how do client and server establish connection in web socket protocol
web socket handshake
HTTP request with upgrade header
what does duplex mean in web socket protocol
either client or server can start sending messages at any time
what efficiencies are there in web socket protocol
much less control data and overhead as messages are short
requests can start in either direction
