AWS focuses on a scalable communication platform

As many manufacturers strive to present a complete IoT platform - from the terminal to the analytics solution, AWS is taking a different path. The service provider sees his core competence in his message broker and the associated rule machine.

However, ASW's in-terminal, intelligent analytics applications and integrated analytics applications as an integral part of the IoT platform are in vain, but AWS offers analytics capabilities. Only have to be booked and of course also be paid extra. "It's easy to integrate," says Constantin Gonzalez, Principal Solutions Architect's technical contact for major corporate clients at AWS Germany.

For this AWS accesses the terminal interface, for the programming of which the service provider has a software development kit in the program. Gonzalez has worked for AWS since 2012, allowing him to follow the origins of the AWS IoT offering from the very beginning. " IoTis all about connecting billions of devices to the web, ensuring sufficient security - that is device and user identity, data authenticity and control, privacy and privacy, and a reliable infrastructure."

AWS has been offering its IoT platform since November 2015. Previously, in March 2015, the company acquired the start-up 2lemetry and carried out preliminary work, for example, developed basic technologies and designed a suitable service design. 2lemetry's IoT services evolved following the buyout, the price of which AWS does not communicate.
Scope of AWS's IoT offering

Today, the AWS offer initially includes the aforementioned software development kit, which can be used to develop software for embedded devices. "We support many popular platforms, such as Arduino, Raspberry or Broadcom," says Gonzalez. For unsupported hardware, AWS provides customizable embedded code in C, as well as an interface to the MQTT messaging protocol. In addition, a RESTful interface is maintained, because RESTful is, according to Gonzalez less resource-consuming than HTTP and therefore better suited for working with solutions that incorporate mobile phones.

By default, MQTT also includes a messaging model. External end-user, analytic, or control-edge applications connect through open software interfaces that AWS customers have access to. According to Gonzalez, there are no technical limitations regarding the possibilities.

The core know-how lies in the Device Gateway, which acts as a message broker, the rule engine and a shadow mechanism. The rules engine contains the rules that the system uses to filter, organize, and respond to the MQTT messages received from the devices, linking the connected systems on the receiver and transmitter sides, for example, with stored applications.

The number of devices and messages is basically unlimited. The messages are arranged according to so-called topics, whose relation to one another roughly follows the logic of a file tree.

The communication is protected by TLS 1.2, certificates, if desired, also user-own, and as usual with AWS by firewall technologies. A device registry ensures that each connected device gets a unique identity, under which it is known throughout the infrastructure.

Security for devices and gateways is provided by AWS itself. In addition, users can bring their own certificates. The transmission of all data from and to the devices is encrypted using TLS 1.2. If you want a two-factor authentication when communicating with end devices, you can build a trusted platform module into them, but AWS is not responsible for this topic any more than for the development of the edge software.

And an on-premise alternative will be sought in vain at AWS. Who runs AWS IoT relies on the cloud. Of course, various providers, such as Salesforce, are also doing this as part of their own platforms.
Rule engine of the AWS IoT offering

The rules engine is deliberately kept simple and works with SQL statements that can refer to all messages. However, for example, images or sounds from this engine can not be processed directly, but possibly only on the basis of their again alphanumeric representable metadata. "For the analysis of such contents, users can connect special analytics modules," notes Gonzalez.

However, it is quite conceivable that the rule engine generates, for example, based on the metadata of visual or acoustic signal data, for example, an alert, the user points to the need to start at a certain point. "The purpose of designing our platform was not to give a complete picture of each case, but to empower our customers to make their domain knowledge as easy as possible," explains Gonzalez.

The Shadow feature also helps to manage the artifacts of devices that are temporarily out of network or shut down. Once they reconnect to the network, they synchronize via a protocol using the shadow service.

Price Structure of AWS's IoT Offering

The pricing structure of AWS IoT varies by geography. No matter where AWS IoT is used, however, once 250,000 published (ie sent to devices) or sent to the system messages per month within the first twelve months are free. Afterwards, AWS charges $ 5 per million in Europe.

However, it pays to pay attention to the details: "For the purpose of the calculation, a message size step corresponds to a 512-byte data block processed by AWS IoT - either published by the service or made available. You can send messages up to 128 KB in one block. Units are charged with 512 bytes each. A payload of 900 bytes is therefore calculated as two messages, "it says in the AWS online accessible price information.

This is followed by numerous details on how individual MQTT activities are assessed in terms of price calculation. The price calculation is therefore by no means as banal as it seems at first glance. Another crucial factor is the length of the messages, which should definitely be taken into account.

For the transmission of data into various other AWS services (Amazon S3, Amazon DynamoDB, AWS Lambda, Amazon Kinesis, Amazon SNS and Amazon SQS) the service provider does not demand anything. After that, however, there are the usual charges, for example, for the use of memory space, for data queries, etc. on.

With regard to the quality of service, customers can be transferred between the service qualities defined in MQTT: zero (message is transmitted according to best-effort criteria) and 1 (message is delivered at least once). Gonzalez: "This can be critical if a message has to be delivered exactly once. But here, customers can decide for themselves if they want to add the appropriate software routines to ensure that. "The customers knew what they would get and could adjust to it. Offering more QoS mechanisms would make the whole service slower for all customers.

As a special benefit of the AWS offering, Gonzalez highlights its long experience in providing extremely scalable platforms, immediate usability for usage-based billing, customer data sovereignty, and great flexibility in software used on both the end-user and application side.

The IoT business is important to AWS in the future, and as far as demand is concerned, Gonzalez is very satisfied, even though AWS is closed to concrete figures. In fact, AWS has already won several customers in Germany with Philips, MyTaxi, Siemens and SKF. Besides, one works closely together with partner Intel.

The offer is constantly evolving. As usual, listen to the wishes of the customers. On their wish list high above is the AWS neglected edge intelligence, but Gonzalez does not want to commit to when it could be implemented. "For us, IoT is still the first day," says the manager.