2 cissp ® Official Study Guide Eighth Edition

Protecting Servers
Fault tolerance can be added for critical servers with failover clusters. A failover cluster 
includes two or more servers, and if one of the servers fails, another server in the cluster 
can take over its load in an automatic process called 
failover
. Failover clusters can include 
multiple servers (not just two), and they can also provide fault tolerance for multiple services 
or applications.
As an example of a failover cluster, consider Figure 18.2. It shows multiple components 
put together to provide reliable web access for a heavily accessed website that uses a data-
base. DB1 and DB2 are two database servers configured in a failover cluster. At any given 
time, only one server will function as the active database server, and the second server will 
be inactive. For example, if DB1 is the active server it will perform all the database services 
for the website. DB2 monitors DB1 to ensure it is operational, and if DB2 senses a failure in 
DB1, it will cause the cluster to automatically fail over to DB2.
F I g u R e 18 . 2
Failover cluster with network load balancing
Network
load balancing
Internet
access
Load
balancer
Web 1
DB1
RAID
array
DB2
Failover cluster
for database servers
Database used 
by database
servers
Web 2
Web 3
In Figure 18.2, you can see that both DB1 and DB2 have access to the data in the
database. This data is stored on a RAID array providing fault tolerance for the disks.
Additionally, the three web servers are configured in a network load-balancing cluster. 
The load balancer can be hardware or software based, and it balances the client load across 
the three servers. It makes it easy to add additional web servers to handle increased load 
while also balancing the load among all the servers. If any of the servers fail, the load bal-
ancer can sense the failure and stop sending traffic to that server. Although network load 
balancing is primarily used to increase the scalability of a system so that it can handle more 
traffic, it also provides a measure of fault tolerance.
If you’re running your servers in the cloud, you may be able to take advantage of fault 
tolerance services offered by your cloud provider. For example, many IaaS providers offer 

---

Understand System Resilience and Fault Tolerance 
815
load balancing services that automatically scale resources on an as-needed basis. These 
services also incorporate health checking that can automatically restart servers that are not 
functioning properly. 
Similarly, when designing cloud environments, be sure to consider the availability of 
data centers in different regions of the world. If you are already load balancing multiple 
servers, you may be able to place those servers in different geographic regions and avail-
ability zones within those regions to add resiliency in addition to scalability. 
Failover clusters are not the only method of fault tolerance for servers. 
Some systems provide automatic fault tolerance for servers, allowing a 
server to fail without losing access to the provided service. For example, 
in a Microsoft domain with two or more domain controllers, each domain 
controller will regularly replicate data with the others so that all the domain 
controllers have the same data. If one fails, computers within the domain 
can still find the other domain controller(s) and the network can continue 
to operate. Similarly, many database server products include methods to 
replicate database content with other servers so that all servers have the 
same content. Three of these methods—electronic vaulting, remote journ-
aling, and remote mirroring—are discussed later in this chapter.