Distributed Systems

1
Distributed Systems

• Naming Services

2
Introduction to Naming Services

• Names used to refer to a wide variety of
  resources
• Computers, services, ports, objects and users
• Fundamental in distributed system design
• Needed to
• Communicate
• Share resources in a distributed system
• Users cannot communicate with one another via a
  distributed system unless they can name one
  another

3
..Intro to Naming Services

• Name services in a distributed system
• Provides clients with data about named objects.
• Describes approaches to be taken in the design
  and implementation of name services
• Names used in a Distributed system are specific
  to some particular service
• Names are needed to refer to entities that are
  beyond scope of any single service.

4
Names

• Names are used in computer systems to allow a
  process to access resources such as entities,
  locations on a network, other processes, etc.
• A naming system is required to implement a naming
  system. Names can be critically important.
  Names that have meaning to humans can help to
  make a system self-documenting and easier to
  maintain.
• Distributed systems need a naming system that
  works across multiple systems. Many named
  entities may have aliases, making unique
  identification difficult

5
Naming Entities

• A name is usually a string of characters that is
  used to refer to an entity (which can be many
  things_
• an attribute, a host, a file, a printer, a
  process or function, an object.
• The identifier can also be a binary number.
• In order to use any entity, we need an access
  point, usually an address.
• An entity can change access points over time,
  such as a process that moves to a different host,
  or a mobile computer that logs on from a
  different location.

6
Global and Local Names

• Names are always organized in a name space.
• Thus, all names are defined relative to a
  directory node.
• An absolute name can be a confusing term, as it
  is defined within the name space.
• A global name is a name that always refers to the
  same entity anywhere in the system,
• A local name is a name whose interpretation
  depends on where it is used.
• There might be another entity with the same name
  elsewhere in the system.

7
Addresses

• It may seem logical to refer to an entity by an
  address, but it is seldom done.
• If the address changes, you lose contact with the
  entity.
• If an entity has more than one access point,
  which address do you use?
• Therefore, addresses are treated as a special
  type of name, usually with references to connect
  it to identifiers.
• A name that is separate from its addresses is
  said to be location independent.

8
Identifiers

• An identifier is a special form of name that
• refers to at most one entity
• each entity has at most one identifier
• an identifier is never reused, thus always refers
  to the same entity.
• Identifiers make it much easier to refer to an
  entity without ambiguity

9
Human Friendly Names

• Names that are intended to be used by people are
  called human friendly names, in contrast to
  identifiers intended for machine use.
• Sometimes there are different identifiers.
• A network host might have a human friendly
  hostname like gwcregmore.gmit.ie but an IP
  address of 127.168.7.3 expressed as a binary
  number.
• Human friendly names can be important for users
  who need to use and maintain the system.

10
Name Spaces

• Names in a distributed system are usually
  assigned to a name space.
• This can be represented as a labeled, directed
  graph with two types of nodes.
• A leaf node represents an entity and has no
  outgoing edges.
• A directory node has a number of outgoing edges
  each labeled with a name.
• Usually a name space has a single root node from
  which all of the names can be found.

11
Path Names

• If you follow the graph of a name space from a
  directory node to the leaf node that represents
  the entity you wish to access, the series of
  names along the graph form a path name.
• If the first node in the path is the root, it is
  called an absolute path name.
• Otherwise it is a relative path name.

12
Name Resolution

• Given a path name, you should be able to locate
  information stored in the node referred to by
  that name.
• The process of looking up a name is called name
  resolution.
• A search of the name space finds the node with
  the name that is sought and returns the path
  name.

13
Closure Mechanism

• We need to know how and where to start when
  seeking a node by name.
• Finding the initial node in a name space from
  which to begin name resolution is called a
  closure mechanism.
• For example, in a Unix file system, the inode of
  the root node is the first inode in the logical
  disk representing the file system.

14
Aliases

• An alias is another name for the same entity. An
  environmental variable is an example of an alias.
• Hard links allow multiple absolute path names to
  the same entity.
• Symbolic links use a resolution process to find
  an absolute path name for an entity referred to
  by an alias.

15
Garbage Collection

• Removing unreferenced entities from a name space
  is an important task that can be critical to
  system performance.
• If a system has to search through dead links to
  find live ones, a great deal of CPU time can be
  wasted.
• Also, especially in a dynamic system where
  entities are being added and removed often, dead
  links can require a great deal of memory and disk
  space.
• Many system lockups in Microsoft Windows have
  occurred as a result of failure to release
  resources no longer in use.

16
Unreachable Nodes

• In the graph on the next slide, derived from
  figure 4-28 in Tanenbaum, nodes in red cannot
  be reached from the set of root nodes. Since the
  closure mechanism may require searches to begin
  from the root set, these nodes may consume
  computer resources such as memory and disk space
  without being able to be used or even located so
  that the resources can be freed

17
Unreferenced Nodes
Key
Root Node
Reachable Node
Unreachable Node
18
Internet Address Protocols

• Network addressing schemes assign machine
  readable numeric addresses to network cards in
  nodes on a network.
• The most common forms are the Internet protocol
  addresses (IPv4 and IPv6) and Ethernet addresses.
• The pathname system for file spaces has been
  adapted to a similar form to provide human
  readable identifiers for locations on a network,
  including the Web.
• These are Uniform Resource Identifiers (URI),
  extended to Uniform Resource Locators (URL) and
  Uniform Resource Names (URN).

19
Uniform Resource Identifiers

• A uniform resource identifier is a protocol for
  identifying a resource. It is normally in a form
  similar to this
• protocolsubprotocolresource
• There are variations on URIs for different
  purposes,
• such as identifying a database driver,
• a file on a network, or another type of resource.
  
• Extending the model above, we might find forms
  like these
• protocol//hostport/path/subpath/resource
• jdbcoracle9ithinaccounts-payable

20
Uniform Resource Locators

• A URL is a URI that is used to locate a
  particular resource. A URL has three parts
• Protocol
• ftp//
• http//
• file/
• Host
• java.sun.com
• java.sun.com80
• Path
• /products/JDK/1.0.2

21
Uniform Resource Names

• A URN is a URI that is used as a pure resource
  name instead of as an identifier.
• It might identify many different resources, such
  as an email message ESMTP id 0J3N00GVA78VIH70_at_mstr
  5.srv.hcvlny.cv.net
• There is a special category of URI beginning with
  urn reserved for URNs, although, as the example
  above shows, not all URNs begin with urn
• An example using the ISBN number for the
  Coulouris text book urnISBN0-321-26354-5.

22
Name Services

• A name service, such as the Domain Name System,
  manages identifiers
• May be required to translate between naming
  schemes,
• such as Ethernet, IPv4 and URL protocols,
• shown on the next slide.

23
Naming domains to access a resource from a URL
24
What is the Domain Name System?

• Converts human readable names to IP Address
  (similar to a phonebook)
• DNS is client/server oriented,
• We have a name server containing the IP Addresses
  and names which serves information to the
  clients.
• Hierarchical System
• Primarily Used to resolve names to IP Addresses
  (Forward Lookup)
• Can also be used to map IP Addresses to names
  (Reverse Lookup)

25
Why do we need DNS?

• Introduced when the number of systems on the
  internet became high enough that it became
  difficult to track every systems address manually
• If you are using TCP/IP then the translation of
  IP addresses to name is done using the /etc/hosts
  file.
• This has certain disadvantages
• Every time a machine is added or removed from the
  network you have to change the /etc/hosts file.
• /etc/hosts file has to be maintained on all the
  machines.

26
Why do we need DNS?

• DNS is an automated system to do that job
• Instead of updating every machine in the network,
  the DNS server maintains a database and provides
  the client machines with information about both
  addresses and names.

27
World Wide Web

• Every time you access a website by name
• such as www.gmit.ie,
• DNS references a host record
• to resolve that name to an IP address.
• Once the name is resolved the
• web browser retrieves the content from the web
  server using the address

28
Email

• Email uses DNS for mail routing.
• Mail Exchanger (MX) records tell a mail server
  where each message should be routed based on the
  domain.
• Everything before the _at_ symbol in an email
  address is called the user portion and everything
  after the _at_ symbol is called the host portion
  (user_at_domain)
• Uses DNS to resolve the Mail Exchanger record for
  this domain to an IP Address.
• Uses SMTP to send the mail to the receiving
  servers IP address resolved by the DNS.

29
Microsoft Active Directory

• DNS is an integral part of Active Directory
• Active Directory uses service (SRV) records to
  advertise all network services to clients.
• Client can use resources without knowing
  anything about the network layout

30
Other uses of DNS

• DNS is used anytime you reference a host by its
  DNS name connect to a domain name regardless of
  the service you are using.
• Example of services used by DNS are
• telnet and ssh
• for system access to UNIX servers, groupware
  clients and backup utilities.
• Host (A) records are typically used.

31
DNS Namespace

• Hierarchical layout of DNS names
• At the top of the namespace is the Root defined
  by a null character.
• Domain names read from right to left i.e. the
  highest level of the namespace on the right.
• Root is not explicitly specified in user
  applications.
• It is specified in the DNS configuration file and
  is denoted by a trailing period.
• A zone is a grouping of machines that may or may
  not be in the same domain.
• This is a set of machines over which a particular
  name server has authority and maintains data.

32
Top level Domains

• Below the root are the top level domains (TLDs)
  such as .com .org .edu
• The TLDs are maintained by the Internet
  Corporation for Assigned Names and Numbers
  (ICANN)
• You can register domains beneath the TLDs

33
An Example of the Namespace
34
DNS Request Process

• Client sends a request for a DNS name resolution
  to a local DNS server
• The local DNS server first checks to see if it is
  authoritative for the domain and checks for
  cached copy of the requested information.
• If found it will return the response to the
  client and the process ends.

35
DNS name servers
36
Recursion

• If request cannot be fulfilled locally,
• DNS server retrieves information for client from
  other DNS servers.
• This is called recursion.
• Recursion is looking for each portion of the name
  starting from the top of the hierarchy. (ex.
  www.njit.edu)
• First step is to contact one of the Internets
  root name servers with a request for the first
  part of the DNS name (.edu). DNS servers maintain
  a list of these servers.
• The root server returns a list of authoritative
  DNS servers.

37
Iterative navigation
38
Non-recursive and recursive navigation
39
Recursion contd...

• DNS server now contacts one of the DNS servers
  returned by the root server
• It requests the next portion of the DNS name
  being queried in this case gmit.ie
• The queried server replies with the authoritative
  server for gmit.ie
• The DNS server doing the recursion finally has
  the authoritative DNS server for the domain
  being requested it can now query that server for
  www.gmit.ie

40
Hierarchical Layout of DNS

• The hierarchical layout ensures that no DNS
  server needs to hold the entire DNS database
• Instead the root DNS servers hold a list of
  servers authoritative for each TLD
• The servers for the TLD hold a list of servers
  for domains under that TLD and those servers hold
  the actual record
• This reduces the load on the servers and ensures
  that the DNS namespace is maintainable across the
  root servers and the TLD servers.

41
Caching

• Caching is another feature of DNS which helps
  alleviate the load on the servers.
• Once a server caches this information it does not
  need to perform the entire recursion process
  again.

42
Configuring DNS clients in UNIX

• Login as root
• vi /etc/resolve.conf (configuration file)
• Three entries
• Domain - This will be appended to any DNS queries
  that fail.
• Search - You can specify a list of up to 6
  domains to append in case of failed queries (use
  either domain or search but not both)
• Nameserver IP Address of the DNS Server

43
Nslookup Help
44
Some DNS servers in Unix

• BIND
• Berkeley Internet name domain
• Maintained by the Internet Software Consortium
  (ISC)
• Most common DNS sever in use today
• DNS servers are included with many Unix operating
  systems such as
• Sun Solaris, HP-UX, IBM AIX and others.

45
The X.500 Directory Service

• The X.500 Directory Service is a naming service
  adopted by ITU and ISO as a standard for
  providing a network directory service
• similar to the yellow pages provided by
  telephone companies
• to allow individuals and organizations to provide
  information about themselves to others.
• X.500 is the basis for the Lightweight Directory
  Access Protocol (LDAP) and is also used in the
  DCE directory service.

46
X.500 Architecture

• X.500 servers organise data in a tree structure
• with named nodes like other naming services,
• X.500 allows a wide range of attributes to store
  information at each node.
• Nodes can not only be searched by name, but the
  attributes can be searched as well.
• The X.500 name tree is called
• the Directory Information Tree (DIT)
• the entire structure with all the attribute
  information is called
• the Directory Information Base (DIB).

47
X.500 Architecture

• X.500 servers are Directory Service Agents (DSA)
  and their clients are Directory User Agents
  (DUA). Figure 9.10 shows the service
  architecture and one of the possible navigation
  models. Each DUA client interacts with a single
  DSA, which uses other DSAs as necessary to
  satisfy requests for information in the DIB

48
X.500 service architecture
49
X.500 Example

• The following two slides illustrate an X.500
  system.
• The first slide shows part of the DIT that
  includes the University of Gormenghast in Great
  Britain.
• The second shows one of the associated DIB
  entries with information on a member of staff.

50
Part of an X.500 Directory Information Tree
51
An X.500 DIB Entry
52
References

• http//linux-tutorial.info
• www.rpmfind.net
• George Coularis, Jean Dollimore and Tim Kindberg,
  Distributed Systems, Concepts and Design, Addison
  Wesley, Fourth Edition, 2005
• Andrew Tanenbaum and Martin van Steen,
  Distributed Systems, Principles and Paradigms,
  Prentice Hall, 2002 ISBN 0-13-088893-1