238
F. Firouzi and B. Farahani
requires that new machines be added to the system to share in handling the
workload. When it comes to databases, horizontal scaling often means partitioning
data (i.e., a single node only holds part of the data). Horizontal scaling is generally
easier to achieve through the addition of new machines to the existing server pool.
Horizontal scaling examples include MongoDB, Cassandra, and Google Cloud
Spanner. Clustering is usually accomplished by using a load balancer at the front of
the cluster to receive incoming requests and send them to nodes within the cluster.
The primary advantages of clustering include:
• Improved Performance – Adding nodes as required and balancing the load across
them enables quick and accurate responses to client requests.
• Improved Reliability – Clusters eliminate single points of failure because if a
node fails the load balancer sends requests to other notes until that particular
node is functioning properly again.
• Reduced Cost – Clustering is a cost-efficient method for achieving optimal
performance and scale because it requires commodity hardware only.
• Easy Maintenance – Cluster nodes can be taken down for maintenance purposes
or upgraded as required during business operating hours without interrupting
performance because additional nodes within the cluster are able to handle
incoming requests.
4.12 A Paradigm Shift from Cloud to Fog Computing
As the number of devices, data, and interactions continues to rise, Cloud architecture alone is unable to handle the deluge of information. The Cloud is able to
provide computing access, storage, and easy, cost-effective connectivity; however,
centralized resources can also generate delays and create performance issues if
data or devices are distant from a public Cloud or data center. In this context,
fog/edge computing has been created to tackle the aforementioned problems.
Cisco’s definitions for edge and fog computing are as below [33]:
• Edge Computing (Known as “Edge”) – Moves the processing closer to the data
source and does not require that data be sent to a remote Cloud or centralized
system for processing. Because it removes distance and reduces time usually
required to transfer data to a centralized system, the speed and performance
quality of devices, applications, and data transport are improved.
• Fog Computing – Defines how edge computing works and supports operations
for storage, computing, and network services among endpoint IoT devices and
Cloud computing centers. In other words, while fog is the defining standard,
edge is the foundational concept. Fog provides the required structure in edge
computing, enabling enterprises to move computing from Clouds or centralized
systems to edge, resulting in improved, scalable performance.
F. Firouzi and B. Farahani
requires that new machines be added to the system to share in handling the
workload. When it comes to databases, horizontal scaling often means partitioning
data (i.e., a single node only holds part of the data). Horizontal scaling is generally
easier to achieve through the addition of new machines to the existing server pool.
Horizontal scaling examples include MongoDB, Cassandra, and Google Cloud
Spanner. Clustering is usually accomplished by using a load balancer at the front of
the cluster to receive incoming requests and send them to nodes within the cluster.
The primary advantages of clustering include:
• Improved Performance – Adding nodes as required and balancing the load across
them enables quick and accurate responses to client requests.
• Improved Reliability – Clusters eliminate single points of failure because if a
node fails the load balancer sends requests to other notes until that particular
node is functioning properly again.
• Reduced Cost – Clustering is a cost-efficient method for achieving optimal
performance and scale because it requires commodity hardware only.
• Easy Maintenance – Cluster nodes can be taken down for maintenance purposes
or upgraded as required during business operating hours without interrupting
performance because additional nodes within the cluster are able to handle
incoming requests.
4.12 A Paradigm Shift from Cloud to Fog Computing
As the number of devices, data, and interactions continues to rise, Cloud architecture alone is unable to handle the deluge of information. The Cloud is able to
provide computing access, storage, and easy, cost-effective connectivity; however,
centralized resources can also generate delays and create performance issues if
data or devices are distant from a public Cloud or data center. In this context,
fog/edge computing has been created to tackle the aforementioned problems.
Cisco’s definitions for edge and fog computing are as below [33]:
• Edge Computing (Known as “Edge”) – Moves the processing closer to the data
source and does not require that data be sent to a remote Cloud or centralized
system for processing. Because it removes distance and reduces time usually
required to transfer data to a centralized system, the speed and performance
quality of devices, applications, and data transport are improved.
• Fog Computing – Defines how edge computing works and supports operations
for storage, computing, and network services among endpoint IoT devices and
Cloud computing centers. In other words, while fog is the defining standard,
edge is the foundational concept. Fog provides the required structure in edge
computing, enabling enterprises to move computing from Clouds or centralized
systems to edge, resulting in improved, scalable performance.