What is Clustering?

Clustering is a technology in which one or more computers work together as a single instance by sharing
their resources. Clustering is used mainly for redundancy of the services hosted on it. In other words, in a
cluster all the hosts are work together, if one of the host fails all the services hosted on it comes online
immediately on to another host with out any downtime. Apart from the redundancy a cluster is used for
load balancing also, in this case what ever the load comes to the cluster is shared by all the hosts in it.
What are the types of Clusters and How do we configure them ?
Clustering is two types, Active/Active and Active/Passive.
Active/Active : In this scenario all the hosts configured in the cluster takes all the load comes to that
cluster, and if any one of the host fails the remaining hosts share the reuests to make it online all the
time. If you have same copies of application configured on multiple hosts this type of clustering is
applicable.
Active/Passive : In this scenario load balancing is not works, as at any point of time only one host
receives all the reuests, and if any of the host fails the remaining hosts takes those services and make
them available online. !hat means, only one host is active at any point of time and the other one is
passive.
What are the requirements to configure a cluster and How it works?
!here are so many reuirements to configure the cluster, but all those reuirements are inter dependable
on each other. And these reuirements "we call them as resources in clustering#, are shared between all
the hosts, that means these should be available to all the hosts in the cluster. for e$ample, %A& storage
resource, the storage which configured in network and is accessible by all hosts "Click here for SAN
Information#. In the same way all the resources are accessible by all the hosts in a cluster. If one of the
host fails, all these resources are moved to another host. In detail, at a particular time one host is
accessing these resources, at that time those are locked by that particular host, whenever this host fails,
the lock was released and those resources are locked by another host. 'ut in active/active clusters these
resources have multiple copies as multiple instances on all hosts.
Microsoft Cluster Interiew !uestions and Answers
"What is Clustering# $riefly define % e&plain it ?
Clustering is a technology, which is used to provide (igh Availability for mission critical applications. )e
can configure cluster by installing *C% "*icrosoft cluster service# component from Add remove programs,
which can only available in +nterprise +dition and ,ata center edition.
"'ypes of Clusters ?
In )indows we can configure two types of clusters
(# N)$ "network load balancing# cluster for balancing load between servers. !his cluster will not provide
any high availability. -sually preferable at edge servers like web or pro$y.
*# Serer Cluster+ !his provides (igh availability by configuring active.active or active.passive cluster. In
/ node active.passive cluster one node will be active and one node will be stand by. )hen active server
fails the application will 0AI123+4 to stand by server automatically. )hen the original server backs we
need to 0AI1'AC5 the application
" What is !uorum ? A shared storage need to provide for all servers which keeps information about
clustered application and session state and is useful in 0AI123+4 situation. !his is very important if
6uorum disk fails entire cluster will fails.
"Why !uorum is necessary ?
)hen network problems occur, they can interfere with communication between cluster nodes. A small set
of nodes might be able to communicate together across a functioning part of a network, but might not be
able to communicate with a different set of nodes in another part of the network. !his can cause serious
issues. In this 7split8 situation, at least one of the sets of nodes must stop running as a cluster.
!o prevent the issues that are caused by a split in the cluster, the cluster software reuires that any set of
nodes running as a cluster must use a voting algorithm to determine whether, at a given time, that set has
uorum. 'ecause a given cluster has a specific set of nodes and a specific uorum configuration, the
cluster will know how many 7votes8 constitutes a ma9ority "that is, a uorum#. If the number drops below
the ma9ority, the cluster stops running. &odes will still listen for the presence of other nodes, in case
another node appears again on the network, but the nodes will not begin to function as a cluster until the
uorum e$ists again.
0or e$ample, in a five node cluster that is using a node ma9ority, consider what happens if nodes :, /, and
; can communicate with each other but not with nodes < and =. &odes :, /, and ; constitute a ma9ority,
and they continue running as a cluster. &odes < and = are a minority and stop running as a cluster, which
prevents the problems of a 7split8 situation. If node ; loses communication with other nodes, all nodes
stop running as a cluster. (owever, all functioning nodes will continue to listen for communication, so that
when the network begins working again, the cluster can form and begin to run.
> ,ifferent types of !uorum in Windows serer *--. ?
:.&ode *a9ority . -sed when 2dd number of nodes are in cluster.
/.&ode and ,isk *a9ority . +ven number of nodes"but not a multi.site cluster#
;.&ode and 0ile %hare *a9ority . +ven number of nodes, multi.site cluster
<.&ode and 0ile %hare *a9ority . +ven number of nodes, no shared storage
" ,ifferent types of !uorum in Windows serer *--/ ?
Standard !uorum + As mentioned above, a uorum is simply a configuration database for *%C%, and is
stored in the uorum log file. A standard uorum uses a uorum log file that is located on a disk hosted on
a shared storage interconnect that is accessible by all members of the cluster.
%tandard uorums are available in )indows &! <.? +nterprise +dition, )indows /??? Advanced %erver,
)indows /??? ,atacenter %erver, )indows %erver /??; +nterprise +dition and )indows %erver /??;
,atacenter +dition.
Ma0ority Node Set !uorums + A ma9ority node set "*&%# uorum is a single uorum resource from a
server cluster perspective. (owever, the data is actually stored by default on the system disk of each
member of the cluster. !he *&% resource takes care to ensure that the cluster configuration data stored
on the *&% is kept consistent across the different disks.
*a9ority node set uorums are available in )indows %erver /??; +nterprise +dition, and )indows %erver
/??; ,atacenter +dition.
"1&plain a2out each !uorum type ?
&ode *a9ority: +ach node that is available and in communication can vote. !he cluster functions only with
a ma9ority of the votes, that is, more than half.
&ode and ,isk *a9ority: +ach node plus a designated disk in the cluster storage "the 7disk witness8# can
vote, whenever they are available and in communication. !he cluster functions only with a ma9ority of the
votes, that is, more than half.
&ode and 0ile %hare *a9ority: +ach node plus a designated file share created by the administrator "the
7file share witness8# can vote, whenever they are available and in communication. !he cluster functions
only with a ma9ority of the votes, that is, more than half.
&o *a9ority: ,isk 2nly: !he cluster has uorum if one node is available and in communication with a
specific disk in the cluster storage.
" How is the quorum information located on the system disk of each node kept in synch?
!he server cluster infrastructure ensures that all changes are replicated and updated on all members in a
cluster.
" Can this method 2e used to replicate application data as well?
&o, that is not possible in this version of clustering. 2nly 6uorum information is replicated and maintained
in a synchroni@ed state by the clustering infrastructure.
" Can I conert a standard cluster to an MNS cluster?
Aes. Aou can use Cluster Administrator to create a new *a9ority &ode %et resource and then, on the
cluster properties sheet !uorum tab, change the uorum to that *a9ority &ode %et resource.
" What is the difference 2etween a geographically dispersed cluster and an MNS cluster?
A geographic cluster refers to a cluster that has nodes in multiple locations, while an *&%.based cluster
refers to the type of uorum resources in use. A geographic cluster can use either a shared disk or *&%
uorum resource, while an *&%.based cluster can be located in a single site, or span multiple sites.
" What is the ma&imum num2er of nodes in an MNS cluster?
)indows %erver /??; supports B.node clusters for both +nterprise +dition and ,atacenter +dition.
" ,o I need special hardware to use an MNS cluster?
!here is nothing inherent in the *&% architecture that reuires any special hardware, other than what is
reuired for a standard cluster "for e$ample, there must be on the *icrosoft Cluster (C1#. (owever, some
situations that use an *&% cluster may have uniue reuirements "such as geographic clusters#, where
data must be replicated in real time between sites.
" ,oes a cluster aware application need to 2e rewritten to support MNS?
&o, using an *&% uorum reuires no change to the application. (owever, some cluster aware
applications e$pect a shared disk "for e$ample %61 %erver /???#, so while you do not need shared disks
for the uorum, you do need shared disks for the application.
" ,oes MNS get rid of the need for shared disks?
It depends on the application. 0or e$ample, clustered %61 %erver /??? reuires shared disk for data.
4emember, *&% only removes the need for a shared disk uorum.
" What does a failoer cluster do in Windows Serer *--. ?
A failover cluster is a group of independent computers that work together to increase the availability of
applications and services. !he clustered servers "called nodes# are connected by physical cables and by
software. If one of the cluster nodes fails, another node begins to provide service "a process known as
failover#. -sers e$perience a minimum of disruptions in service.
" What new functionality does failoer clustering proide in Windows Serer *--. ?
&ew validation feature. )ith this feature, you can check that your system, storage, and network
configuration is suitable for a cluster.
%upport for C-I, partition table "CP!# disks in cluster storage. CP! disks can have partitions larger than
two terabytes and have built.in redundancy in the way partition information is stored, unlike master boot
record "*'4# disks.
" What happens to a running Cluster if the quorum disk fails in Windows Serer *--/ Cluster ?
In )indows %erver /??;, the 6uorum disk resource is reuired for the Cluster
to function. In your e$ample, if the 6uorum disk suddenly became unavailable
to the cluster then both nodes would immediately fail and not be able to
restart the clussvc.
In that light, the 6uorum disk was a single point of failure in a *icrosoft
Cluster implementation. (owever, it was usually a fairly uick workaround to
get the cluster back up and operational. !here are generally two solutions
to that type of problem.
:. ,etemrine why the 6uorum disk failed and repair.
/. 4eprovision a new 1-&, present it to the cluster, assign it a drive
letter and format. !hen start one node with the /06 switch and through
cluadmin designate the new disk resource as the 6uorum. !hen stop and
restart the clussvc normally and then bring online the second node.
" What happens to a running Cluster if the quorum disk fails in Windows Serer *--. Cluster ?
Cluster continue to work but failover will not happen in case of any other failure in the active node.