Sei sulla pagina 1di 5

c 


 
NLB is a distributed algorithm used to load balance network traffic across a number of hosts.

c  
A cluster is a group of independent computers that work together to run a common set of
applications and provide the image of a single system to the client and application.

c      


    
Server Cluster is a collection of servers that together provide a single, highly available
platform for hosting applications.

NLB Clusters dynamically distribute the flow of incoming TCP and UDP traffic among the
cluster nodes according to a set of traffic-handling rules. NLB usually functions at the
network level and have little to do with the actual application. There is no shared disk
requirement.

c            


No. NLB and Server Clusters CANNOT be used on the same set of servers.

c 
     
A single NLB cluster supports up to 32 hosts.

c !   "       



Yes. NLB performance begins to decrease from the linear as the cluster grows beyond 20-25
nodes.

c 
   #$%   
NLB can be used to scale beyond 32 machines by using Round Robin DNS between multiple
NLB Clusters.

For example, if virtual IP 1 (VIP1) has 32-nodes behind it, and VIP2 has another 32 nodes
behind it, you can load balance between VIP1 and VIP2 using Round Robin DNS, hence
having 64 nodes in NLB. The same analogy can be scaled to very large number of nodes.
(Although I don¶t see a reason why you would need so many nodes operating in tandem).

c 
& &    ' 
NLB Cluster host emits ³heartbeats´ to the other hosts in the cluster. If a host fails and stops
emitting heartbeats, then after a default time period of 5 seconds, the remaining hosts in the
cluster undergo a process called     to remove the failed host from the cluster and
have new client connection requests mapped to remaining hosts in the cluster.

c 
               
5 seconds are required to detect a failed host it is default time.
2 to 3 seconds to remove the failed host and redistribute its load to the alive hosts.

c &     "     



No. Heartbeat packets are emitted every second by each host and consume less than 1,500
bytes
c    "  
Yes. NLB has a Windows kernel component called WLBS.SYS. (WLBS = Windows L oad
Balancing Services)

c        " ( ( &    


)((&*
In Windows NLB, automatic recovery occurs within 5 seconds
The load balancing is more even in the case of Windows NLB, when compared to Round
Robin DNS.

c 
&       
Convergence involves computing a new cluster membership list and recalculating the
statistical mapping of client requests to the cluster hosts. There are two instances in which
cluster traffic has to be remapped due to a change in cluster membership:
1. when a host Leaves the Cluster, and
2. when a host Joins the Cluster.

c     +,-   


No. NLB does not respond to changes in the server load such as CPU usage or MEMORY
utilization or the HEALTH of an APPLICATION. NLB has nothing to do with the application
itself. It merely balances evenly the network traffic among a number of nodes based on some
port rules. If your NLB nodes are of different hardware configurations, you may face
problems when requests are sent to the slower node, because NLB has no way of finding out
which node is slow, it just evenly distributes traffic. period.

c             


 " . 
If the cluster is configured in No Affinity mode, NLB will use both the Source IP Address and
the Source Port to achieve the load balancing, and so load will be distributed amongst all of
the hosts.

c     
--    "  
= 
There is no inter-host communication possible between the hosts configured in Unicast mode
with 1 NIC.
 
Allows inter-host communication between the hosts configured in Multicast mode with 1
NIC.

c 
 (  
'    
  
Hosts can be homed to their own LAN or Virtual LAN. It will work for both Unicast or
Multicast modes.

c &  / 


 
 "  
No.

c 
    $
  



Make sure that the switch does not associate the cluster MAC address with a particular switch
port!
c 
&    #
  



Layer 3 switches need to be specially configured to work with NLB. A VLAN must be
established for the hosts in the cluster, and this VLAN must be configured to operate in Layer
2 mode.

c 
& (    
 + % %' 
Create a subnet that spans two switches and connect half of the NLB cluster nodes to each
switch. In this case, if one switch fails, you only lose half of your nodes from participating in
the NLB. Alternatively, you can have other failover arrangements on your core switch.

c  
 
!"       
   
       % %  
" 
Simply set the metric on the cluster NIC to a higher value than the non-cluster NIC.

c  +    0"  --   0 


-  
No. The entire cluster MUST be in one operational mode.

c & "" -" 1 +!  


Yes. NLB supports multiple, virtual IP addresses.

c "   "   "      +  )1+*
     
Windows Server 2003 supports specifying different port rules for different virtual IPs.
However, this is not supported on Windows 2000 NLB.

c "   .


23 45
$444

  $44#   
Yes. Mixing is supported.

c "    "   


Yes. This is supported in Windows Server 2003 only.

c  
     
Yes. In a switched environment.

c    " 5
        
 
  %  
Yes. As long as the hosts are part of the same subnet.

c 
 "   -  
Configure Network Load Balancing Manager to log each event.

c        676


Yes, but this is generally not recommended.

c 
  
&    )&0*    
NLB utilizes the TCP/IP Denial of Service attack protection.
c 
&     
NLB assumes that the
LAN to which it is homed is trusted. There are no security features to configure on the NLB
itself. Administrators should secure the network itself using firewalls, intrusion prevention
systems, etc.

c 
          %  
To configure a host to handle more or less than an equal share of the load, edit the port rule to
clear the ³Equal´ load weight check box and enter a load weight number between 1 and 100.

c 
&  ! -  & '  ! -  0  
    - !"" 
ë
   
NLB load balances traffic based only on the Source IP Address of the incoming connection.
Single Affinity mode ensures that all TCP connections originating from the same client (IP
Address) are sent to the same host in the cluster.

    


NLB load balances traffic is based on Source IP Address and Source Port of the incoming
connection request. In No Affinity mode, multiple connections from the same client may be
handled by different hosts in the cluster as long as these connections have different source
ports.

c         


  5   
   
  
Yes, for efficiency reasons configure the port rule in Single Affinity mode.

c  " 
     
    
   
 .               " 
NLB has nothing to do with your application/web site. It merely distributes requests evenly
between the nodes. The synchronization/replication of content/data evenly between the two
nodes has to be done manually or by using another solution.

c         


No, for database servers like Microsoft SQL Server, use Server Clusters/Microsoft Cluster
Service instead. You can, however, use NLB on your front end application web servers and
have them connect to the Server Cluster on your database servers. See image below.
c     2    
Yes.

c   2    5


  
    
  %          
WLBS/NLB relies on the client¶s IP address to determine which Terminal Server services a
client. If you configure WLBS/NLB to use Affinity, the IP address used by the client is
serviced by the same Terminal Server as long as you do not change the Terminal Server
cluster.

If you need disconnected clients to connect to the same Terminal Server to recover from a
µdisconnected¶ session, the client computers need to use static IP addresses and WLBS/NLB
must be configured to use Single Affinity. Note that IP addresses obtained from DHCP
servers on the LAN or through your ISP may change, as well as roaming users¶ IP addresses.
SeeKB243523.

c & "" (   


No. WINS names should not be automatically registered for the IP addresses configured on
the NLB interface. The IP can be mapped statically in WINS.

c  $2+,+   


Yes, in Windows Server 2003 NLB supports both PPTP and L2TP VPN sessions.

c  8  
!""  %  
Yes.

c  
    
Yes.

c  %  02 


Yes, it is possible, though not recommended for File and Print Services.

Potrebbero piacerti anche