70-412: Chapter 2: Configuring Netowork Load Balancing

Front 1

What is NLB?

Back 1

-is a free inbox technology that we can use in Windows server to scale up and scale out a web infrastructure.

-To scale up means to increase the capacity of a single box.

-To scale out an infrastructure, it typically means “redundancy”-setting up multiple instances of the same type of server for the purposes of load balancing or spreading the user load around and failover-which relates to hypothetically losing one box, we can still ensure high availability for our users.

Front 2

How many web servers can you aggregate through NLB?

Back 2

2-32

Front 3

These are going to be your IIS Web Servers that present the application to the users themselves

Back 3

Front-End

Front 4

This tier is where our application servers exist, where the main input, output, and business logic happens

Back 4

Middle-Tier

Front 5

is your Sequel Server, where it’s fundamentally storing the data that’s used by the applications on the web servers in one or more databases

Back 5

Back-End

Front 6

Why is NLB synonymous with web traffic (HTTP, with FTP, and basically IP traffic?)

Back 6

-NLB in windows is optimized for “stateless traffic”
-HTTP is a stateless protocol
-The idea of statelessness, means that under normal circumstances, it shouldn't matter which front-end web host the user’s connections go to.

Front 7

applications that require a persistent one-to-one between the client and the server, even the web server, such as “File Services”, and “Database services

Back 7

Stateful

Front 8

NLB has some “INTELLIGENCE”; uses algorithms, mathematical formulae to determine?:

Back 8

-“#1” which nodes are part of the cluster
-“#2” which nodes are alive and which are not,”
-“#3” Priority-when incoming connections come in;

Front 9

is a key point of any cluster solution, is an internal communication method that our cluster nodes use to query each other to make sure they’re alive

Back 9

Heartbeat

“Every 1 minute” these cluster nodes exchange heartbeat messages to determine who’s still in the cluster. “If a cluster node misses 5 consecutive heartbeats, that node is automatically removed from the cluster and then traffic is routed by the NLB stack to the remaining alive nodes

Front 10

we can specify port numbers, TCP and/or UDP ports that will be load-balanced

Back 10

Port Rules

Front 11

is a technology enabled in windows DNS by default that will return a client’s query with a list of matching “A” records.

Back 11

DNS Round Robin

has no intelligence, priority or fault tolerance, and no way for the 3 web servers to detect who’s online and have no knowledge of the other’s existence., not a Manageable solution

Front 12

How does Microsoft suggest we combine Round Robin with NLB?

Back 12

1. We use NLB clustering to balance the load on our front end servers; *WE would create a NLB cluster in each of our points of presence, with all of them mapping to the same name.

2. Now we could enable “DNS Round Robin” for each cluster; so we will have multiple cluster IP addresses and we would hand these out with Round Robin

Front 13

• Network Load Balancing Requirements?

Back 13

1.All cluster nodes need to be on the same subnet(IPv4 or IPv6) with a least 2 NICs. Depending upon the operational mode

2. Microsoft recommends “less than 250 ms latency”between the hosts-So that we don’t miss heartbeats.

3. You can combine different versions of WS2K12 in a single NLB cluster; But they need to be identical or near identical load outs on those boxes, especially in regards to Server Roles, features, line of business software

Front 14

Operational Modes:
NLB functions in three modes: ?

Back 14

Unicast, Multicast, and IMGP Multicast

Front 15

Unicast?

The Problem with Unicast?

Back 15

one-to-one IP communications; All nodes in the cluster use the same virtual unicast MAC address nodes; So NLB will actually overwrite in the registry, the MAC address of each of your nodes with a shared cluster MAC address.

Each of the nodes appear to your switches as the same box, because their all using the virtual mac, therefore the nodes will not be able to communicate with each other in the absence of an additional second adapter (can’t communicate with each other (needs a 2nd adapter).

Front 16

Multicast

Back 16

One-to-many IP packet exchanges; Each nodes maintains its own MAC address, so the nodes(servers) can communicate with each other because of course fundamentally, in Ethernet networking, each NIC has to have a globally unique media access control or mac address

Front 17

. IGMP(Internet Group Messaging Protocol)

Back 17

One-to-many IP packet exchanges..etc; If you are going to do multicast, this will prevent the switch from “port flooding” that could occur with other operational modes; In this mode: The connection traffic is forwarded only through the switch ports that participate in the NLB cluster, no other switch ports are involved

Front 18

Determines how many witness must fail before the cluster is in a failed state.

Back 18

Cluster quorum mode

Nodes are computers that participate in a cluster. Witness can be shared on a shared storage or even on file shares

Front 19

There are four cluster quorum nodes.

Back 19

1. Node Majority
2. Node and Disk Majority
3. Node and File Share Majority
4. No Majority: Disk Only

Front 20

this cluster quorum mode can be used with clusters that have odd number of nodes, is recommended for only testing environments because the disk hosting the witness functions as a single point of failure

Back 20

No Majority: Disk Only

Front 21

In this cluster quorum mode, the quorum is stored on a network share rather than a shared storage disk; Is not reliable because the file shares do not generally have redundancy features

Back 21

Node and file share majority

Front 22

This cluster quorum mode is recommended when the cluster has an even number of nodes; A disk witness, hosted on a shared storage disk that is accessible to cluster nodes has a vote when determining as do the cluster nodes; The cluster retains quorum as long as the majority of the voting entities remain online

Back 22

Node and disk majority

Front 23

this cluster quorum mode is recommended for clusters of odd numbers; The cluster retains quorum when the number of available nodes exceeds the number of failed nodes

Back 23

Node majority

Front 24

allow multiple multiple cluster nodes to have concurrent access to a single LUN

Back 24

Cluster Shared Volumes

Front 25

is a feature that enables you to automate the process of applying software updates to a failover cluster; Integrates Windows Update, WSUS, SCCM, and other software update apps.

Back 25

Cluster-Aware Updating

Front 26

blocks new connections to the cluster node but doesn't terminate the existing connections. Use this prior to planned maintenance to gracefully evacuate the cluster of connections

Back 26

Drainstop

Front 27

Stops the cluster node; All connections to the cluster node from the clients are stopped; Use this after you use Drainstop, so that you can then perform maintenance task such as applying updates

Back 27

Stop

Front 28

Starts a cluster node that is in a stopped state

Back 28

start

Front 29

Pauses the cluster node until you issue the resume command; Using this does not shut down the cluster server, but it terminates the current connections as well as blocks new connections

Back 29

Suspend

Front 30

Resumes a suspended cluster

Back 30

node

Front 31

POWERSHELL


Function: Configures and manages a cluster node

Back 31

NlbClusterNode

Add-, Get-, Remove-, Resume-, Set-, Start-, Stop-, and Suspend-

Front 32

POWERSHELL
Configures the cluster node's dedicated management IP address

Back 32

NlbCusterNodeDip
Add-,Get-,Remove-,Set-

Front 33

POWERSHELL
Create and manage port rules

Back 33

NlbClusterPortRule

-Add,-Disable,-Enable,-Get,-Remove,-Set

Front 34

POWERSHELL
Configures the cluster's virtual IP address

Back 34

NlbClusterViP

Front 35

POWERSHELL
Configure and manage the cluster

Back 35

NlbCluster

Front 36

Provides information about the cluster driver

Back 36

Get-NlbClusterDriverinfo

Front 37

Retrieves information about interfaces, including information about the NLB driver, on a host.

Back 37

Get-NlbClusterNodeNetworkInterface

Front 38

Generates IPv6 addresses to create cluster virtual IP addresses or node dedicated IP addresses

Back 38

New-NlbClusterIpv6Address

Front 39

Sets the host priority of a port rule for a specific NLB node.

Back 39

Set-NlbClusterPortRuleNodeHandlingPriority

Front 40

Sets the load weight of a port rule for a specific NLB node.

Back 40

Set-NlbClusterPortRuleNodeWeight