Cloud Delivery and Hosting Models - Deploying and Managing a Cloud Infrastructure - page 174

■ Understand network-related traffic trends in order to properly charge network-
related services.
■ Establish proper service-level agreements (SLAs) and meet quality of service (QoS)
promises and goals.
Network Security Isolation This type of isolation can be built on top of network traffic iso-
lation to ensure that database traffic is authenticated and secure for trusted clients. This can
be implemented using encryption such as SSL/TLS and HTTPS and some other authentica-
tion protocol, such as allowing or denying database service access through validation rules.
Functionality and Performance Validation
Reliable performance will always be a requirement for a cloud environment. But aside from
the specific performance metrics that were discussed in the previous chapter, such as disk and
file system performance, performance of the system as a whole should also be validated. This
means that all aspects must be taken into consideration, including end-to-end performance.
On-Premises Performance
Private cloud and on-premises solutions have a responsibility to internal customers. All
performance is controlled internally, so there will be no problem regarding validation
because performance will depend on the IT department and its competence. Still, bench-
marking and testing have to be done in order to objectively conclude that a system is
performing as expected according to specifications.
The methods used for validating and testing cloud performance do not stray too far
from traditional software and system performance testing methods. The tools might differ
because of different parameters, but the general process remains pretty close. The test team
may get creative and test for other parameters, such the ability of the system to detect DoS
attacks from legitimate massive traffic inflow.
Off-Premises Performance
Off-premises and public cloud solutions vary from their counterparts in one aspect, location.
The cloud infrastructure, including hardware and the virtualized environment, may remain
similar for on-premises and off-premises infrastructures, but a lot of other variables get
introduced that affect the end user experience drastically. One is the public network between
the cloud environment and the user. The Internet is a collection of networks owned and
operated by different organizations, which means that there could be thousands of different
configurations that may affect end-to-end performance. Even if the cloud environment is
performing at peak conditions, the system between it and the user may be experiencing dif-
ficulties, and there may be a lot of black holes along the way.
But for these external systems, the responsibility of performance validation already
falls to the service provider, so the users simply have to access the service provider's data