Device Servers To The Rescue.

The Device Server bridges the gap from the device to the Internet, enabling network connectivity.

So you've joined the masses by ordering your groceries online and have scheduled delivery to your home as yet another attempt to free up time in your hectic life. You're dreaming of the day that you can telepathically send grocerystore.com your shopping list so you don't "waste time" logging on and placing an order. While the promise of intelligent refrigerators and microwaves with bar-code scanners and Internet connections will bring us closer to shifting the paradigm on reordering our staple grocery products, the greater benefit of these technological changes are already being realized.

Though the network-enabled refrigerator may be a novelty, it does showcase an extremely interesting shift--the transition of appliances and tools as islands of information to integrated devices or nodes on a network. By giving the appliance a unique address on a network or sub-network, it is enabled to communicate rather than simply serve as a standalone product. Internet Protocol (IP) standards provide the global addressing scheme that has enabled the dramatic growth in e-commerce and is poised to change the way everything in the world works. Proliferation of IP addressable devices essentially flattens the communications hierarchy. If you combine this form of communication with intelligent appliances, you unlock vast amounts of information. The question is does this type of solution exist?

Device Servers To The Rescue

The Device Server is the enabler of network connectivity. It provides the encapsulation of serial data into TCP or UDP/IP packets for transmission across the ubiquitous Ethernet. It is through this innovative connectivity that devices supporting serial communications can be remotely accessed and managed. Anything that can be done through the local serial port can now be done remotely from anywhere in the world.

The world is loaded with intelligent, computer-controlled devices acting as islands of information. They do their job well most of the time, but when they require maintenance, there must be a local connection either through a technician with a laptop or a serial connection to a server that is on the network. By attaching the device's serial port to an IP-addressable Device Server, the device can now be diagnosed, configured, and managed over the Internet, remotely and without the use of a server as a gateway. The Device Server bridges the gap from the device to the Internet, enabling network connectivity.

An example, far from the world of the fridge and the microwave, is that of a debugging tool. Engineers designing new products often find themselves running systems designed to analyze code running on hardware. A debugging system typically operates on a PC and interfaces to the newly designed product through a serial port. If a programmer wants to test new code, he must be physically near the product to apply it. Rather than use a RS232 cable, a programmer can use a Device Server attached to the serial port. Now, the product becomes a node on the network and development reprogramming and testing can be done remotely. The debug tools still reside on the PC, but the application data is transmitted through the network card instead of the serial port. The data is sent to the Device Server's IP address and the encapsulated serial information is presented to the serial port. The Device Server simplifies the development environment by allowing access by engineers anywhere on the network (Fig 1).

How Does It Really Work?

The Device Server consists of a serial input, processor, memory, RTOS and IP stack, and the Ethernet physical layer interface. The serial input can be RS232, RS-485 (two wire or four wire), or TTL for board-to-board connectivity. The RTOS, IP stack, and the processor work together encapsulating the serial data into IP packets for transmission across the network. Memory holds network and serial buffers, the non-volatile OS and IP stack, and custom applications.

Encapsulation involves taking the raw serial data and "wrapping" it with the information needed to send it to another destination on the network. The data is generally wrapped with headers, including a Preamble indicating that a new and valid frame is about to start, the Destination and Source addresses, a Length field identifying the size of the encapsulated data, and a Frame Check Sequence field for use in verifying the integrity of the IP packet. Demultiplexing reverses this process once the IP packet has reached its destination: essentially handing the destination its original serial data stream (Fig 2).

The Device Server's Versatility

Capable of more than just simple encapsulation of serial data for transmission in IP packets; the Device Server can send SNMP or email messages to network managers and even serve up Java applets and web pages. SNMP messaging allows the network-enabled serial devices to send messages to the Device Server that trigger certain calls in a Management Information Base or MIB. These SNMP messages are sent to remote Network Management tools that understand them and graphically display the information and status. Through the use of a Device Server, an appliance capable of self-diagnosis can send a standard formatted message to a network manager through a management tool such as HP OpenView or other enterprise network management tools.

The Device Server is capable of serving Java applets from its memory to any browser. For instance, the applet can be a menu program for configuring the product, serve accumulated diagnostic data, and provide real-time updates and static product data such as serial numbers and revision control information. In addition, Web pages can be served from the Device Server containing such static information and links to information located on other network servers. This way, a user could get to the product manual by simply clicking on the link in the embedded web page. These are powerful ways of providing the appliance's users with access to important and relevant information.

Simple scripts can be developed to send emails when certain serial strings are seen. This way, a device could send an error code that the Device Server would understand as a trigger to send email.

Eliminating The PC As A Gateway

PCs are robust devices with capabilities far beyond what most people use them for. However, using Device Servers instead of PCs offers four main advantages:

1. Reduced purchase cost - While there are a number of low cost PCs available now, the cost of a Device Server is significantly less. Device Servers have no need for a hard drive, monitor, keyboard, or mouse and are priced accordingly.

2. Smaller form factor - A standalone Device Server is the size of a deck of cards and, therefore, can take up much less space than a PC. In some cases, the Device Server is small enough to be embedded into the product that it is supporting.

3. Lower Total Cost of Ownership (TCO) - A Device Server does not have the same OS requirements as a standard desktop so the administration of the device is substantially reduced (if not eliminated). Lack of a hard drive, fan, and other peripherals with moving parts further decreases maintenance support costs.

4. Decreased management needs - Device Servers are standards-based, using IP as the communications means. As a standard Ethernet connection, no protocol gateways are required--further eliminating management and equipment costs. The Device Server is, then, managed as any Ethernet port with an IP address and can be accessed as such.

Considerations For The Future

As intelligent devices and Internet appliances become more pervasive in our daily lives, a number of issues must be confronted in order to maintain network Quality of Service and basic peace of mind.

* Security is important - Encryption, password protection, and Address resolution allow users to secure their products' link to the network. Only users with priority access are capable of getting to the network-enabled devices through a Device Server. The Device Servers can be set to only communicate with certain IP addresses and, in some cases, are more secure than a typical PC connected to an Ethernet network.

* Network design counts - Some devices connected to an IP-based network may be transmitting time sensitive data. If this is the case, network design incorporating careful layout and traffic flow will reduce collisions and create a more deterministic environment for time critical messaging. While this is considerably more important when using TCP/IP and Device Server technology for control systems, even monitoring and configuring devices can become difficult on an overly congested network.

* IP addressing is evolving - The network is dynamic and rapidly expanding and will, at some point, outgrow its current 32-bit addressing scheme. IP Version 6 will be upon us, offering higher QoS and security features. 128-bit addressing will support a theoretical max of 340 trillion trillion trillion (3.4 x 1038) addresses, making it possible for just about everything on planet Earth to have its own IP address. Device Server technology can make this a reality.

The mass adoption of Ethernet and IP protocol as the communications standard has opened the door for change. Device Servers are poised to enable network connectivity of existing systems and will establish themselves as the communications medium for future product designs due to versatility and ease of implementation. Device Servers give users the option of adding open standards-based networking to their existing serial applications. In industries like retail POS, automation, medical, security access and control, telecommunication, and commercial IT, system integrators have used Device Servers to gain real time remote access to critical information, minimize down time, manage inventory, and increase productivity.

Kevin Burger is the OEM business development manager and Richard Geasey is the strategic marketing manager at Lantronix (Irvine, CA).