Web Services: Wisdom And Management

By Craig Thomas, CTO, GroundWork Open Source

We have all heard the old chestnut, "the wonderful thing about standards is that there are so many to choose from." Standards make for a pretty dry topic, but open standards implemented with open source bring powerful benefits to mid-market systems and network management customers. I'll get off this standards bandwagon soon, but just wanted to shed some light on a couple more for now.

In the past couple of Technology Insights columns, the subject of Service Modeling Language (SML) has been a common theme. From the SML website (http://www.serviceml.org):

The SML specification is used to model complex IT services and systems, including their structure, constraints, policies, and best practices. [...] SML will allow for the creation of best practices and policies that automate the services' validation, development, operations, updates and end-of-life - the full lifecycle.

SML, then, allows for the modeling of services and systems and policies and practices. It enables software to automate the full lifecycle.

But how does the software that understands SML make the systems and services conform to the model? How do the managed systems and services describe themselves so that the planned state versus actual state can be compared?

Enter a pair of standards: WSDM, pronounced "wisdom", and WS-Management, pronounced "WS-Man".

Figure 1, WSDM and WS-Man side by side

WSDM was the result of work by HP, IBM, Grid community and partners. It was submitted to the Organization for the Advancement of Structured Information Standards, with version 1.0 ratified in March, 2005, and version 1.1 ratified in August, 2006.

WS-Management was the result of work by Intel, Microsoft and partners. WS-Management is part of the Web-Based Enterprise Management set of standards. WS-Management is at version 1.0, in a preliminary state as of April, 2006.

Figure 1 is a look at the two standards, side by side. Funny thing is that they cover pretty much the same territory. Fortunately, the two groups are working on a "unification" of their standards. They have identified what to choose from which existing standards, and plan an architecture that converges on a single set of standards. Figure 2 shows the target set. Names may change as work continues, but the goal has been called out.

Figure 2, WS-UnifiedManagement

Participants in this unified set of standards expect to have covered the territory and proposed standards by the end of June, 2007. That is a pretty fast track for standards.

So what is the benefit to GroundWork's customers?

Here is a future-tense scenario, a user-story, that illustrates how software that implements all these standards could be used.

1. A system administrator walks into a heterogeneous environment with state of the industry hardware and software from HP, Microsoft, Dell, RedHat, Intel, Cisco, BEA, VMWare. The previous system administrator kept all knowledge about the network on his laptop, and mysteriously the disk drive has been wiped as clean as can be, DBAN clean.
2. The system administrator installs GroundWork Monitor Professional (remember this is a futuristic tale), clicks on the big green "discover" button, and a map of the systems, networks, services, and interdependencies starts to appear.
3. A subnet across the world starts to take shape. Hey, there's an HP Open View implementation there in that division. Another subnet, another division, this time with a CA implementation. There are existing GroundWork installations, and a couple of Nagios®* systems, too. Here comes a Microsoft network, complete with Active Directory, SharePoint servers, and a gaggle of Windows laptops. There's a Microsoft System Console in this subnet. All of these management systems are interoperable, since they are standards-aware.
4. As the network's nodes materialize, the management endpoints are detected. Web services on all of the machines start to report details of the configuration of each system: software versions, patch levels, backup policies, security policies. Hypervisors are discovered, as well, and begin to report details about the virtual machines and services. A mesh network of environmental sensors is discovered in each computer room, temperature, humidity, noise level, and vibrations are reported.
5. Great! everything is looking good so far. Just two teensy weensy problems. First, there don't seem to be any credit card gateways on the e-commerce servers, and the backup policy is missing on the IT box where GroundWork is installed.
6. The system administrator commits the discovered system model for use as a baseline.
7. Next, she drills in on the information in the e-commerce subnet, and sees that there is a router that is unresponsive. A little drag-and-drop, onto the help desk's web service icon, and a trouble ticket is dispatched.
8. A right-mouse-click on the IT box with the GroundWork software installed shows a menu of operations policies, including backup. A daily backup with seven days of recovery sounds okay here. Click to commit, and the policy is installed on the backup server.
9. The trouble ticket on the e-commerce routers has been dispatched. The techs find and fix the problem, and the routers start working. Since they are web service enabled, they advertise their status, and the nearby monitoring systems are updated. Notification continues, and an update is made in the GroundWork Monitor system as well. A comparison with the earlier baseline version of the system model shows that the redundant routers are working.

So how big a fantasy is all that? Well, not such a big deal, provided you don't have devices, systems, and software from multiple vendors. But nearly every business makes use of heterogeneous systems. Ours does, doesn't yours?

Web services are becoming ubiquitous. Intel's AMT puts web services onto vPro systems such that they are accessible when the computer is off, as long as the power cord and network cable are connected. Archrock's mesh networks offer management endpoints via web services; every mote in the mesh is addressable. Application servers are innately web service enabled.

WS-*, pronounced "ws-star", standards are what will be used to assure interoperability among the web services implemented in products from different vendors. Service Modeling Language will allow software systems to automate the management of components that offer web services.

GroundWork is actively engaged in open source communities that are implementing these open standards in open source. Within Eclipse COSMOS, the big 4 is represented, particularly with CA and IBM. These companies are just as interested in interoperability as GroundWork. Their interest is driven by the same thing as our interest: our customers need it. Customers are also demanding that WSDM and WS-Management get unified.

Let's get one less choice in the standards arena.

* Nagios is a registered trademark of Nagios Enterprises