Planning for Cloud Integration: Pitfalls and Advantages - Deploying and Managing a Cloud Infrastructure - page 379

Refactor Code to Address Necessary Changes in the
Execution Environment
Public cloud vendors charge for compute, storage, and network resources and also for read
and write disk operations. Applications can require disk accessibility for storing data, disk
caching, and swapping data in and out of memory. Therefore, it can be fruitful to gather
reads and writes and bunch them into single operations whenever possible.
In fact, this serves two more purposes apart from saving costs. First, application developers
can think in terms of atomicity of operations and bunch together those operations that reflect
the atomic nature of events. Atomicity means that in a series of outstanding uncommitted
operations, either all occur or nothing occurs. Second, an I/O operation such as reading from
disk or writing to disk is the most expensive operation in terms of time consumed and latency
incurred. Bundling such operations can cause a considerable drop in I/O time and in the pro-
cess make the application more responsive.
Moreover, other optimization techniques can be used that utilize the concepts of multi-
threaded programming, concurrency, and in-memory processing to cut down not only on
response times and latency but also on the usage of compute and storage resources.
Which Service Model Is Best for You?
The cloud covers a wide spectrum of services and delivery models. Therefore, it is highly
relevant to understand the variety of cloud services and choose the service model best suited
to your organization's needs. Here, we outline the three foundational cloud service models
and their best use cases.
Infrastructure as a Service (IaaS)
IaaS provides basic computing resources and allows the customer to transfer an existing
workload to the cloud with minimal changes. Here are two use cases for IaaS:
Extra Capacity for Short-Term Peak Workloads In an e-commerce or retail business, work-
loads can be expected to peak at certain times of the year, such as Christmas, Thanksgiving,
Easter, Black Friday, and so on. These shopping peaks could also coincide with aggressive
advertising campaigns. Quickly provisioning capacity during these times is important to
keep up with the buying pace. IaaS cloud service models can use the elastic ability of the
cloud to add virtual resources to boost the capacity. The linchpin here is traffic monitoring
because of the variance in user traffic that occurs.
Storage System for Storing Media Some domains consume huge amounts of storage
capacity, including for audio and video files. Typical use cases involve a film production
house, a music production house, or even a video-enabled security surveillance system or
a security company offering such a service. In addition to the files being stored, the service
could allow storing metadata about the videos and make them accessible on demand.