16
F. Firouzi et al.
Data storage, besides the data processing, is another challenging issue in the era
of big data. As more and more devices are connected to the Internet of Things,
the amount of data they generate will drastically increase. They will be sending
messages with their status, sensor outputs, metadata, and other messages. Despite
the large amounts of data, it still must be stored. Two common methods are listed
here, NoSQL databases and time series databases. Traditional techniques (SQL
databases) are usually not feasible because of the amount of data being stored, with
its varied and often unstructured nature. NoSQL databases offer high throughput and
low latency of storage and retrieval. Since there is no schema, dynamic new data
types are allowed. Couch Base, Apache Cassandra, Apache Couched, MongoDB,
and Apache HBase (Hadoop) are examples of frameworks that use NoSQL. There
is also a NoSQL in the cloud solution offered by IBM’s Cloudant (a distributed
database) and AWS’ DynamoDB. A time series database can also be a NoSQL
database or even a relational database. The indexing and queries are all based on
timestamps in the data. Some frameworks using time series databases are InfluxDB,
Prometheus, and Graphite.
1.1.6 IoT and Cloud Computing
The two worlds of IoT and Cloud experienced swift and independent progress.
However, the complementary features of IoT and big data generated many new
opportunities and advantages. Cloud computing is the solution to the increased
demand for storage and processing. The cloud is defined as a group of servers
and computers connected over the Internet in a large, distributed infrastructure. The
concept is to deliver on-demand services over the Internet. The model is typically
based on pay for the usage consumed (metered service), with the ability to scale
up and down as needed (elastic resources). Amazon, Microsoft, and Google are
dominating this Infrastructure as a Service (IaaS). They also provide Platform as
a Service (PaaS) and Software as a Service (SaaS). The advantage to consumers
is a lowered computation cost versus purchasing the hardware and then paying to
operate and support it in-house. In summary, the main drivers for integration of IoT
and Cloud are listed below [10]:
• Device lifecycle management: As the Internet of Things grows in size, the
number of devices that need to be registered, managed, and updated while
maintaining security requirements also grows and must be accommodated. It is
possible for tools to configure and update firmware and software over the air
(FOTA). The cloud platforms enable device lifecycle management, so devices
can be connected, registered, on-boarded, updated remotely, and even remotely
diagnosed should something need to be fixed. This reduces the operation and
support cost of the devices. That means the enterprise Internet of Things is
remotely managed, with minimal time and a reduced cost of ownership. In other
words, a 360-degree view of the IoT devices is possible via the cloud.
F. Firouzi et al.
Data storage, besides the data processing, is another challenging issue in the era
of big data. As more and more devices are connected to the Internet of Things,
the amount of data they generate will drastically increase. They will be sending
messages with their status, sensor outputs, metadata, and other messages. Despite
the large amounts of data, it still must be stored. Two common methods are listed
here, NoSQL databases and time series databases. Traditional techniques (SQL
databases) are usually not feasible because of the amount of data being stored, with
its varied and often unstructured nature. NoSQL databases offer high throughput and
low latency of storage and retrieval. Since there is no schema, dynamic new data
types are allowed. Couch Base, Apache Cassandra, Apache Couched, MongoDB,
and Apache HBase (Hadoop) are examples of frameworks that use NoSQL. There
is also a NoSQL in the cloud solution offered by IBM’s Cloudant (a distributed
database) and AWS’ DynamoDB. A time series database can also be a NoSQL
database or even a relational database. The indexing and queries are all based on
timestamps in the data. Some frameworks using time series databases are InfluxDB,
Prometheus, and Graphite.
1.1.6 IoT and Cloud Computing
The two worlds of IoT and Cloud experienced swift and independent progress.
However, the complementary features of IoT and big data generated many new
opportunities and advantages. Cloud computing is the solution to the increased
demand for storage and processing. The cloud is defined as a group of servers
and computers connected over the Internet in a large, distributed infrastructure. The
concept is to deliver on-demand services over the Internet. The model is typically
based on pay for the usage consumed (metered service), with the ability to scale
up and down as needed (elastic resources). Amazon, Microsoft, and Google are
dominating this Infrastructure as a Service (IaaS). They also provide Platform as
a Service (PaaS) and Software as a Service (SaaS). The advantage to consumers
is a lowered computation cost versus purchasing the hardware and then paying to
operate and support it in-house. In summary, the main drivers for integration of IoT
and Cloud are listed below [10]:
• Device lifecycle management: As the Internet of Things grows in size, the
number of devices that need to be registered, managed, and updated while
maintaining security requirements also grows and must be accommodated. It is
possible for tools to configure and update firmware and software over the air
(FOTA). The cloud platforms enable device lifecycle management, so devices
can be connected, registered, on-boarded, updated remotely, and even remotely
diagnosed should something need to be fixed. This reduces the operation and
support cost of the devices. That means the enterprise Internet of Things is
remotely managed, with minimal time and a reduced cost of ownership. In other
words, a 360-degree view of the IoT devices is possible via the cloud.