Reducing the Cost of Systems Operations

Management Tools is also known as : Advanced Management Tools, Agile Management Tools, Agile Project Management Tools, Based Management Tools, Best Project Management Tools, Best Time Management Tools, Build Management Tools, Business Management Tools, Change Management Tools, Client Management Tools, Code Management Tools, Computer Management System Tools, Configuration Management Tools, Construction Management Tools, Contract Management Tools, Control Management Tools, Cost Management Tools, Customer Management Tools, Development Management Tools, Document Management Tools, Email Management Tools, Engineering Management Tools, Exchange Management Tools, File Management Tools, Free Management Tools, Free Project Management Tools, Free Time Management Tools, Knowledge Management Tools, Life Cycle Management Tools, Life Management Tools, List Management Tools, Management Analysis Tools, Management Planning Tools.

Today, many organizations are paying high maintenance fees for expensive proprietary management tools that are functionally inferior to the native management tools they get for free.

Management Summary

 

Increasingly, major IT vendors -such as Microsoft, Oracle and IBM- are providing systems management tools as part of their base products in order to drive down the application running costs for their customers. Centauri provides Abbey National with a cost-effective way of bringing these different management tools under a single umbrella, this way we can manage the service of these applications collectively rather than separate, discrete components. This gives us a big picture of our systems' service quality, more akin to the way our customers and users experience it, and allows us to bring IT 'out of the backroom and into the boardroom' in terms of the information we provide.

Jon Ratcliffe, Head of Enterprise Support at Abbey National Bank (UK), June 2003

The current economic downturn is forcing organizations to implement cost reduction programs across their enterprise. IT systems management infrastructure is a prime candidate for cost cutting since significant savings can be made by removing expensive intermediary software products, such as BMC Software PATROL and CA Unicenter. Once a vital part of systems monitoring, these tools are now largely redundant because:

1. The trend among IT vendors - such as Microsoft, Oracle and SAP - to embed systems management subsystems in their base products.
2. Advances in network capacity that make centralized monitoring a more cost-effective alternative to remote server-based monitoring.
3. The proliferation of systems management standards, such as web based enterprise management (WBEM) that makes management tool integration more straightforward.

The opportunity for cost reduction has existed for several years, but its potential has not been exploited since the cost of integrating these disparate tools into a common management umbrella outweighed the cost of using a one-size-fits-all proprietary product. The availability of high-function/low-cost event management tools - such as Centauri Business Service Manager - has changed this situation. But perhaps the most significant impetus for change is the economic recession, forcing organizations to question the adage that "if it ain't broke, don't fix it".

Today, many organizations are paying high maintenance fees for expensive proprietary performance management tools that are actually functionally inferior to the native management tools they get for free. Worse still, according to Gartner, 50 percent of serverbased network and system management (NSM) software purchases are never fully deployed across the intended device population. This white paper looks at this waste. It sets out a strategy for reducing annual maintenance costs by removing redundant intermediary management tools. The strategy is called disintermediation.

Proxima Technology's Centauri Business Service Manager provides a low-cost service level management solution that allows organizations to bring native management tools under a single management umbrella. Where native or standards-based agents are not available, Centauri can be configured for agentless monitoring, retrieving performance information directly from the application. Centauri dramatically decreases annual maintenance costs (validated by our customers) by replacing expensive proprietary management products with those supplied with the application. Furthermore, the load on the managed servers is significantly reduced because the proprietary agents are no longer duplicating the tasks of the native agents. Centauri is not a "rip and replace" solution and can be layered over an existing management infrastructure, allowing the proprietary agents to be replaced by native agents (or other agentless monitors established) over time to avoid a risky big-bang approach. For best results, disintermediation is applied as part of a wider IT/business alignment exercise, which ensures cost savings are not made at the expense of business service.

In addition to enabling disintermediation, Centauri provides a digital dashboard that shows - in real-time - how IT is supporting the business, highlighting the cost of poor quality to the business caused by service degradation. Centauri also supports the Six Sigma quality management process and uses this to improve the quality of service.

Trends in Distributed Systems Management

 

The Late 80's

In order to understand where we are as an industry and where we are going, it is useful to review from where we have come. During the late 80's and early 90's, distributed systems management (excluding mainframes) was largely restricted to network management. Tools like HP OpenView, Sun NetManager, and IBM NetView were the standard. In addition to network monitoring, these tools also offered some capability in the management of the UNIX operating systems; checking resource utilization and other basic tasks. During this period, low cost mini computers running UNIX began to proliferate throughout organizations, largely to feed the insatiable appetite of the IT Department, who was churning out applications using 4th Generation Languages (4GL) such as Oracle SLQ*Forms and Ingres ABF. Suppliers of these application development tools - RDBMS vendors such as Oracle, Ingres, and Informix - were not concerned about the management of these technologies: they were simply too busy adding features to their base products to make them more competitive in an increasingly cutthroat market.

Consequently, in the early part of the 1990's, demand started to build for systems management tools that could manage these distributed environments. Patrol Software was founded to address this demand in 1992 by Martin Picard, the designer of Oracle's SQL*Net product. The first version of PATROL was released soon after. PATROL was initially built to monitor Oracle databases with rudimentary support for UNIX monitoring added later. PATROL consisted of a console/agent architecture: agents ran remotely on the servers to perform monitoring operations and send status information back to the central console for display. Two decisions taken by the PATROL architects early on were to prove pivotal in PATROL reaching its market leading position. One was to detach the management expertise from the agents, the other was to make the agents autonomous from the console.

Detaching the knowledge about the systems to be managed from the agents that performed the monitoring allowed PATROL to potentially manage any type of application. Libraries of expertise - called knowledge modules - could be defined and loaded into the agent to make it aware of a new application to be managed. These knowledge modules contained all the information the agent needed to discover and monitor the application. The idea at the time was to encourage everyone and anyone to build knowledge modules for PATROL and, in doing so, make PATROL a defacto standard in the industry. However, with the acquisition of Patrol Software by BMC Software in 1995, this potential was never fully realized and while there are a significant number of knowledge modules available in the market, PATROL itself is not defacto.

The second major decision affecting PATROL's architecture was to make the agents autonomous from the console. In contrast to its competitors - EcoTOOLS and DB-Vision - PATROL's agents were able to function without there being a persistent connection to a console. At a time when network capacity was at a premium, this decision was to prove a real winner since network traffic flowing between the console and agent was kept at a minimum. Ultimately, this meant that PATROL could scale to manage the sort of large-scale environments that BMC Software's customers were running. The lack of scalability of EcoTOOLS and DB-Vision proved to be their downfall.

IT Vendors Wise-Up

Although PATROL's potential to be a defacto standard for application management was never fulfilled, its future was very much called into question when IT vendors, such as Oracle, Sun, and Microsoft, got wise to systems management. In the late 90's, leading IT vendors realized that the inability to manage their applications was a barrier to their adoption. For example, Microsoft's customers would never move to the latest version of Windows if they were spending all their time keeping the current version running. As a result, all leading IT vendors have initiated programs to make systems management native to their applications. Today, despite a few false-starts, all leading vendors have their management tools in place: Oracle has Enterprise Manager, SAP has CCMS, Microsoft has MOM, Sun has MC, Compaq has Insight Manager, Dell has OpenManage, and so on. In most cases, these are not only supplied free of charge, but it is actually not possible to run the application without them.

With a new vice president of its Management Business Group [Kirill Tatarinov, formally R&D VP for BMC Software and Patrol Software], Microsoft has the potential to become a network and systems management (NSM) industry leader.

Gartner FirstTake, 19th July 2002

Apart from being free, these native tools have other advantages over proprietary management tools like PATROL and HP OpenView: they exhibit best practice in managing the application. There are no upgrade issues (e.g. out of sync releases) and resource consumption is minimal. So the question is: if these native systems management tools are so good, how come management vendors like BMC Software and CA continue to turn in such healthy profits? To date, the answer has been simple:

1. The cost of integrating these many systems management tools has outweighed the cost of using a single proprietary agent.
2. Internal processes in the IT Department discourage cost reduction. For example, as new servers are rolled out, a standard set of management tools are installed regardless of whether the server participates in a mission critical transaction.

 

Disintermediation in Systems Monitoring

Ultimately, the goal of any systems management function is to provide IT service; one that maximizes the operational efficiency of the business, but at the lowest cost possible. Increasingly, attention of the systems management community has focused on digital dashboards - the front end for business service management. The digital dashboard is a realtime display that shows how IT is supporting critical business activities and at what cost. Because businesses depend increasingly on IT, the dashboard provides insight into the operational status of the business itself. For example, a digital dashboard allows the manager of Customer Call Center to take appropriate actions to mitigate the effect of IT disruption. A decent dashboard should highlight the cost of poor quality and areas of over-capitalization as being candidates for quality improvement initiatives.

The data collectors that exist between the digital dashboard and the application components that enable the business are intermediaries - a means to an end. They collect information, analyze it, and pass it on to a digital dashboard for display. Disintermediation is the process of removing unnecessary intermediaries: in systems management terms, replacing expensive proprietary monitoring tools with either native tools embedded in an application or removing them altogether and retrieving metrics directly from the dashboard platform itself. Disintermediation is made possible because of the availability of competitively priced event management tools, such Centauri Business Service Manager from Proxima Technology. Centauri is able to retrieve event messages and other performance data from a plethora of management tools, effectively bringing all the native management tools under a single management umbrella, and displaying the service provided on a digital dashboard.

Today, the proliferation of native systems management tools has meant that most organizations are running at least two management tools to do the same job: one being native, such as Microsoft Windows WMI, the other being a proprietary management agent, such as CA Unicenter. Apart from being an additional burden on resources, the true opportunity cost to the organization is that you are paying more money for a proprietary management tool that provides less value than the free native product! In fact, if that isn't bad enough, if you look carefully under the covers, you will see that these high-cost proprietary management agents actually retrieve their data from the native tool. Eliminating this redundancy and all the costs associated with it is the driver for disintermediation.

Agentless Monitoring

During the initial stages of PATROL development, one trap the developers found themselves in was adding meaningless and irrelevant monitoring parameters simply to differentiate the product. At that time, many customers were conducting superficial evaluations and would often be more influenced by the quantity of monitoring rather than the quality. Consequently, to keep up with EcoTOOLS and DB-Vision, the PATROL developers found themselves adding parameters that served little practical value other than to get a check-in-the-box in a paper product evaluation. Today, most of the monitoring parameters are switched off. As a result, many organizations are seriously questioning whether these highly priced monitoring tools can be justified and many organizations are looking for the most cost appropriate alternatives.

We must also recognize that the world is a very different place from the early 90's when PATROL and other proprietary agent-based products were created. One factor that has changed significantly is network capability. An original design imperative for PATROL was that it make minimal use of the network. However, here we are 10 years later - where people are routinely distributing multi-megabyte files across a corporate network - spending time and money to prevent a few bytes of management data flowing across a network. Indeed, it is not uncommon for organizations today to waste more money trying to minimize network traffic through expensive proprietary agent license costs, software configuration, and maintenance than they would if they implemented a centralized management solution. Factor in the reality that most of the monitoring being carried out by these remote monitoring tools is actually switched off and you begin to see the significant opportunity for cost reduction in systems monitoring.

Agentless monitoring is a management configuration that does not require an agent to be installed on the managed server. With agentless monitoring, requests for information can be executed from the central console and the appropriate management actions carried out depending on the returned information. This approach leads to significant cost savings since no management software need be licensed for the remote server. Furthermore, implementation and configuration is significantly easier than the agent based solution. The downside is that the quality of monitoring is reduced.

Although the notion of centralized monitoring is not new, it is a more feasible option now because network bandwidth is not the issue it was in the 80's and 90's when many agentbased solutions were designed. However, agentless monitoring is not a panacea and is really only appropriate to non-mission critical situations where the requirements for monitoring are not great.

Current Day

As the "Big 4" management vendors-BMC, CA, HP and IBM-put in place plans to celebrate the 10 year anniversary of their agent based management tools, it is worth stepping back to consider how much enterprise-computing has changed in this period. Given these significant changes-the Internet, consolidation of UNIX, growth of network capacity, proliferation of Windows, commoditization of RDBMS, wide scale adoption of package applications-the question is: are these tools even appropriate anymore?

Increasingly, you are not alone in asking these questions. The Big 4 vendors, who stand to lose out from disintermediation, will tell you that their tools are as appropriate today as they were a decade ago. Yet, their aggressive revenue growth plans do not factor in native management tools, management standards and agentless monitoring scenarios. While the price point of their component products may be falling, disintermediation is certainly not on their agenda. It's not that they are inept; they are just out of touch. Buyer beware!

Disintermediation with Centauri

Proxima Technology believes that replacing proprietary systems monitoring tools with native and/or agentless monitoring capabilities will result in significant cost savings, a belief shared by Microsoft, SAP, Oracle, Sun Microsystems, IBM, and other companies looking to lower total cost of ownership by embedding systems management tools in their products. By integrating these native tools into the Centauri management umbrella, you have the same seamlessly integrated management infrastructure that you have with your proprietary management tools - such as Unicenter and PATROL - but at a fraction of the acquisition and maintenance costs. However, disintermediation is not a panacea and organizations embarking on such a strategy need to be careful not to throw the baby out with the bath water. Proprietary management tools continue to have their place, particularly in more mission-critical applications where large volumes of metrics are relevant. Additionally, proprietary management tools often provide facilities useful to administer systems monitoring functions in larger sites, although this is offset by the fact that there is more to administer with proprietary tools, particularly when compared to agentless monitoring configurations. Probably the greatest downside of disintermediation is that just about every management scenario needs to be treated as a special case and you will be faced with different options for the monitoring approach you take.

Although the savings from disintermediation are significant, it is foolish to ignore the implementation effort required. This effort is greatly reduced when it is part of a broader IT/business alignment exercise since data collection is addressed as a matter of course. In this case, disintermediation is one of the many benefits of IT/business alignment with Centauri. Many organizations have been able to justify their business service management investment in terms of the cost savings from disintermediation. Disintermediation, therefore, consists of two steps: plan it and do it. This may sound facetious, but the planning stage is critical and is the IT/business alignment step. A Service Catalog must be established that provides a clear sense of priority. This step can be avoided in very tactical situations where only a few data collectors are being replaced. In all situations, do not be driven by a desire to replace all proprietary monitoring tools since there exists a trade-off between cost reduction and increased risk to the business. However, this risk is significantly reduced by:

• The data collector itself; e.g., Microsoft WMI offers robust and comprehensive monitoring and is near risk free, while an agentless database monitor is less robust.
• The criticality of the application; clearly, mission critical applications pose a greater risk than non-mission critical applications.
• Taking a top-down approach that ensures you understand what services are critical to your business before you attempt to replace any proprietary data collector.

Additionally, one significant advantage of the approach presented here is that the two management environments can run in parallel and only when a steady state is established in the new environment can the transition commence. Furthermore, this can be so for any level of granularity - e.g., for a specific application, database, server, for all servers of a type - and can be carried out over any period of time.

The approach to disintermediation described in this white paper is more fully documented as part of the Proxima Implementation Method (PIM). This describes a top down approach to service process analysis and uses Risk Priority Number (RPN), borrowed from Six Sigma, to justify the inclusion of a base metric within the service management model.

Step One: Planning for Disintermediation

Before you can replace any proprietary management tools, you need to understand what IT services are critical to your organization's business venture. With this understanding, you have a reasonable basis for making a cost/risk decision since you know which components (and consequently which monitoring tools) are really important. In effect, you need to establish a Service Catalog. This may already be available in an existing service level management tool, it may be written down somewhere, or you can create this definition as you establish a common event management umbrella for all system monitoring tools, including the yet-to-beintroduced native tools. Organizations who have committed to Centauri will find this to be the most sensible approach.

Centauri provides a cost effective event management capability that provides support for your current proprietary tools, your new native tools, and can also be configured for agentless monitoring. Event messages and performance data are retrieved by Centauri and a service hierarchy established. The service definitions consist of the objects, threshold values, and conditions that determine the OK state for the service. The services are cross-referenced with business processes to establish business context and show criticality. Correlation rules are also defined for the service to show the inter-relationships that exist and provide a basis for root cause analysis while making a necessary distinction between the cause of a problem and its effects. With such a model, Centauri is able to alert appropriate administrators of the cause of a problem and also dispatch advisory "take no action" warnings to the administrators who may be affected by it. This ensures that no one wastes time on unnecessary problem resolution activities.

Establishing a Service Catalog and Centauri event management umbrella is not a trivial task, but it is necessary to minimize the risk of disintermediation - and is ultimately the key to successful cost reduction. If you are looking for early payback, you can approach this business unit by business unit, gradually extending the coverage provided by your management umbrella as you go.

The next step in the process is to start replacing proprietary agents. The steps presented try to balance the goal of securing early cost savings against introducing unnecessary risk to the business. Table 1 shows the key applications and alternative methods for data collection available and provides a snapshot of implementation activities.

Step Two: Putting in Place Your Disintermediation Strategy

 

Monitoring Oracle, SQL Server, and other JDBC Databases

Database monitoring tools are a good starting point for disintermediation. Indeed, customers often state that they can justify the cost of Centauri alone on its ability to replace these tools. These tools are low hanging fruit because they are expensive, provide much more information than is required, and are discrete enough for a 1 for 1 replacement - making the replacement more straightforward. You can take two different approaches, depending on the criticality of the database (as shown in your Service Catalog). Data collectors of low to medium priority Oracle and SQL Server databases can be replaced by Centauri's agentless configuration. This free adaptor provides sufficient information for most situations, reporting on common failure modes such as space (e.g., table extents and freespace) and performance (e.g., processes and user sessions).

Oracle's own embedded management tool - Enterprise Manager - is the obvious choice for your mission-critical Oracle databases, although (at the time of writing) you would be required to manually configure Centauri for Enterprise Manager. Similarly, SQL Server is well covered by Microsoft WMI and the Centauri Adaptor for WMI.

You will likely find support for this strategy among your DBAs. They generally prefer the native management tools over the proprietary management tools that they were often forced to use as part of an overall strategy to implement a consistent management tool or were implemented when embedded management tools simply didn't exist. Indeed, your DBAs may prefer implementation of the native management solution across the board - including nonmission critical databases as well critical ones.

Monitoring Microsoft Windows and Windows Applications

Disintermediation enthusiasts find themselves in heaven in with Microsoft Windows Management Instrumentation (WMI), Microsoft's native management agent. WMI offers the highest degree of monitoring for Microsoft Windows and applications and is a clear sign that Microsoft is serious about systems monitoring. WMI has the potential to replace all data collectors running on a Windows server and it is for this reason that it does not make a good starting point for disintermediation. That is, you may risk biting off more than you can chew in the early stages of implementation. If you follow the steps suggested here, by the time you are ready for WMI, you have had had plenty of practice with database data collectors - agentless, native, or even WMI itself (with SQL Server).

WMI is supplied as a component of Windows 2000 to provide management information and control in enterprise environments. It is also available for Windows 95/98, Windows Millennium Edition (Me), Microsoft Windows NT, Windows XP, and the Windows .NET Server 2003 family. Microsoft's dominant position in the market clearly makes WMI important, but WMI's sophisticated capabilities make this even more significant. WMI is the Microsoft implementation of Web-Based Enterprise Management (WBEM), an industry initiative to develop a standard technology for accessing management information in an enterprise environment. WMI uses the Common Information Model (CIM) industry standard to represent systems, applications, networks, devices, and other managed components in an enterprise environment, although WMI includes additional capabilities over and above CIM.

Although Microsoft requires applications certified for the Windows platform to support WMI, vendors committed to the Windows platform will utilize WMI as a matter of routine. Dell Computer Corporation is a good example. Their OMCI instrumentation package enables Dell computers to be managed remotely. Any management tool that supports WMI natively does not require custom integration to manage Dell systems instrumented with OMCI. Where such instrumentation has not been carried out, WMI includes tools necessary to instrument any application running on the Windows platform.

WMI is an extremely capable native agent, offering far more information than is available through proprietary agents. WMI also has significant cost of ownership benefits: not only is it free from licensing and maintenance costs, as an embedded component of Windows, it is very efficient on resource utilization. The Centauri Adaptor for WMI uses WMI to query and set information on desktop systems, applications, networks, and other Windows components. Developers can use WMI to create event-monitoring applications that alert users when important incidents occur.

The WMI management infrastructure includes the CIM Object Manager (CIMOM), which is the interface and manipulation point for CIM objects and information. It acts as a facilitator in gathering information and manipulating object properties. CIM software components (called "providers") act as Component Object Manager/Distributed Component Object Manager (COM/DCOM) servers that handle requests from the CIMOM when needed. The CIMOM stores data in an area called the WMI repository. Management objects stored in the WMI repository can be either physical or logical entities. All CIM objects that the CIMOM has knowledge of reside in the repository. Users may contribute to the repository by compiling files in the Managed Object Format (MOF).

WMI is designed to provide effective management information in an enterprise environment. To further this goal, Microsoft has designed WMI to interact with other Microsoft management tools. Specifically, WMI can integrate with Active Directory and Active Directory Services Interfaces (ADSI) to increase the ability of WMI to manage objects over a network. A directory service is a component of an operating system that stores information on managed objects and provides that information to a user. Active Directory is the directory service included in Windows 2000. Using Active Directory, a user can manage files and other objects across an enterprise network.

The interface between Centauri and WMI uses Microsoft's query language called WQL. This SQL-like language is used to request data on types of events available through WMI. At the time of writing, Centauri provides the three default queries that notify you of the following events:

1. A new Windows event log entry has been created.
2. A service has changed its state.
3. The disk space on any of the drives has changed.

WQL queries are straightforward and typically just a single line of code. Proxima is continually adding queries for typical monitoring scenarios; it is worthwhile to checkon Proxima's TechZone website for the latest complete list.

Application Management: mySAP

Throughout disintermediation, you will frequently be surprised at the vendors who are doing an excellent job of embedding systems management systems into their applications (and equally surprised at those that are clueless about this subject!). One example of this excellence is SAP. As the dominant supplier of Enterprise Resource Planning (ERP) software, we can assume SAP set the agenda in the packaged software market and so provide a useful point of reference for the industry as a whole. We can assume their lead in embedded systems management tools will encourage their competitors to follow suit.

To facilitate the management of their mySAP.com application(s), SAP have created Computer Center Management System (CCMS). CCMS offers an integrated suite of tools for monitoring and managing multiple mySAP.com solutions from a single console. You can also use the tool to manage databases or to automate operations such as resource distribution. Key features of CCMS include:

• Instance and profile management tools for defining application servers and managing system profiles.
• Automatic load-balancing for user logons, updating, output, and background processing.
• Tools for routine tasks such as sending system messages, displaying the system log, and managing critical system functions.

mySAP Technology also provides functionality for managing databases. Features include database backup, restore, and recovery; database system check and optimization; space management, monitoring, and analysis; and data archiving.

Although CCMS is comprehensive, SAP also supports partner solutions through their Complimentary Software Program (CSP). Under this program, third party management vendors are encouraged to integrate mySAP management capabilities into their own solutions. This could be to manage resources not covered by CCMS, but required by mySAP nonetheless, or to incorporate mySAP status data into a broader management perspective.

Proxima Technology's Centauri Adaptor for SAP allows you to integrate mySAP Technology into a common management umbrella to provide, for example, a corporate event management or digital dashboard solution.

Monitoring Basic Response Times

Centauri adaptors are available to monitor and measure the availability and response times of business critical services, such as HTTP Web server, directory service (LDAP, DNS, NIS), Telnet service, FTP service, and mail services (SMTP, IMAP4, POP3). These adaptors operate in an agentless configuration. Because an agentless monitoring configuration does not require an agent to be installed on the managed server, these monitors can be set up quickly and easily.

Centauri supports agentless monitoring in a number ways. One method is to connect to a server via the standard Telnet (or Secure Telnet) interface. Centauri's Telnet based screenscraper allows performance data to be retrieved from any system that supports Telnet - including all UNIX variants such as SCO UNIX. Because Telnet is available on many proprietary platforms that often fall outside of a distributed systems management architectures-such as VMS, Tandem and OS/390 -it can be managed in the same way. Because of the vast amounts of data that can potentially be collected in this manner, simple scripts need to be created to specify exactly what information will be collected.

SNMP-Your Monitoring Catch All

Simple Network Management Protocol (SNMP) is one of the most widely used and flexible systems management standards. Created in the late 80's by the Internet Engineering Task Force (IETF), it is a standard frequently touted by the Analysts as being in decline and yet keeps on going. As its name suggests, SNMP is a network management standard used in TCP/IP networks but it has been adopted widely to instrument a variety of applications. For example, statistics about an ORACLE database are made available in the form of an SNMP Management Information Base (MIB). This allows any SNMP MIB browser to poll the MIB and retrieve accurate statistics about ORACLE with minimal overhead to the ORACLE database.

SNMP's strengths are its flexibility, ubiquity, and efficient use of a network, although several years ago when bandwidth was at a premium, SNMP's polling nature was considered a problem. As network capability has increased, this is not so relevant. SNMP can be considered as the default management API. Where no other management standard or capability exists, it can generally be assumed that an SNMP capability is available. Obviously, SNMP will be the mechanism of choice when looking to retrieve data about network devices.

Initiatives to Keep in Mind: SOAP, JMX, and WBEM

Three initiatives to keep in mind, all of which will further push the frontiers of disintermediation are SOAP, JMX, and WBEM. Mentioning these now will give this white paper an extended "use by" date.

SOAP
Simple Object Access Protocol (SOAP) is a standard that allows applications to share data and invoke remote procedure calls via XML and the Internet. SOAP is not a systems management standard, per se. It satisfies a more general growing need for business partners to exchange structured data over the Web independently of each other's underlying application platform. It is designed to let organizations easily publish data and services over the Web. As such, it functions as a wire protocol to connect multiple sites - called endpoints - each of which might use an information server, object broker, or other facilities to integrate and process the information. However, the adoption of SOAP by the vendor and end-user community makes it a very relevant alternative to proprietary agents. An application that supports SOAP and publishes systems management statistics - such as Hewlett Packard Total-e-Server - can be integrated with any management tool that supports SOAP directly without the need to use proprietary data collectors.

"NSM integration will occur via XML [SOAP] for NSM application-to-application communications, thus, creating the opportunity for enterprises to build loose federations of integrated NSM services. [SOAP] will be used because it is the glue needed for NSM; it is already used to integrate applications (and NSM applications are just another type of application); and it is standards-based and Internet-oriented. By 2004, a combination of XML [SOAP] and CIM will be used as the primary method for NSM application-to-application integration (0.8 probability)."

Gartner, "NSM Frameworks: Death of Black Holes", R. Paquet, August 2000

JMX
Java Management Extensions (JMX) provides the capability to manage Java applications and Java-based application servers and integrate with existing management tools. There has been tremendous growth in the adoption of Java application servers to host mission-critical applications during the past two years. This growth has created a need to better manage this environment to improve its availability and performance, which are intrinsically linked to the availability and performance of the business transactions it enables. JMX provides the best standard instrumentation opportunity to achieve the manageability goals, especially for Javabased platforms such as WebLogic and WebSphere.

JMX is a management specification, defined through the Java Community Process, and is an evolution from previous Java management efforts, such as JMAPI and JDMK. It consists of a management architecture, a set of APIs, and a set of management services to manage at object level. JMX specifications define extensions that allow any Java-based resource to be inherently manageable.

In their 2001 report on JMX, Gartner expected that by 2003, 50 percent of new productiondeployed Java applications based on Java application server platform will use JMX for management, up from less than 10 percent in 2001. As it turns out, Gartner's expectation has not quite been borne out in reality; JMX is one of several standards and is still yet to secure a position. Consequently, at the time of writing this paper, JMX is not quite as ubiquitous as one would hope; nonetheless, it is an important standard and worth keeping in mind.

WBEM
Although Microsoft has been quick to come to market with their WMI implementation of WBEM, they are not alone in supporting this standard. Sun Microsystems also supplies a Java implementation of WBEM for their Solaris platform, called Solaris WBEM Services. This implementation has also been submitted to WBEMsource to be made more widely available as a standard.

At the time of writing this white paper, Centauri support for WBEMsource is still in development. Proxima, however, regards WBEM as significant.

Grande Finale: Now Switch Off Your Agents!

Your end game for disintermediation is to de-install your proprietary agents. Depending on the architecture of the proprietary management tool, this may done as part of disabling a monitoring tool. We advocate a stepwise approach here: you can never be too cautious. Run your two monitoring solutions side by side until you are comfortable with your replacement.

Throughout this program, remember that disintermediation also has its critics. Many systems management vendors are threatened by this program and will throw many obstacles and objections in your path. Disintermediation is not a one-size-fits-all approach and there will seldom be a 1:1 match between proprietary and native agent solutions in terms of scope of coverage.

One last thing: don't forget to cancel your maintenance agreements!

Summary

Over the next few years, the face of network and systems management (NSM) will change significantly and will have a major impact on the role IT plays in an organization. Breakthroughs in data analysis will allow organizations to better understand the business value of IT. This value will be further enhanced by the reduction in total cost of ownership (TCO) that disintermediation brings.

Disintermediation is an important initiative. It has been created not by a series of "fly-bynight" vendors, but by the leading IT vendors that drive our entire industry. Companies like Microsoft, Sun, SAP, IBM, and Oracle are embedding systems management tools in their products as a means of driving down the TCO burden for their users. Breakthroughs in network capability have opened up other avenues for cost reduction, including agentless monitoring configurations.

Unlike Microsoft, Oracle, and SAP, Proxima Technology is not a driver for this trend, but it is a catalyst. Through our Centauri technology we are enabling disintermediation - allowing organizations to operate native and agentless solutions under a single integrated management umbrella and further enhancing this with a service level reporting dashboards.

Today's driver for disintermediation is fueled by a necessity to cut costs in an economic recession, but because systems management is prominent in lowering TCO, the question isn't so much why disintermediation, the question is: when?

About Proxima Technology

Proxima Technology, Inc. provides software and services to improve business service and accountability through service-level measurement, reporting, and problem notification in distributed computing environments.

www.proxima-tech.com Telephone
Australia 02 9458 1700
Germany 040 32005-405
United Kingdom 0870 870 0732
United States 720 946 7200

©1998-2003 Proxima Technology, Inc., Centauri, and Centauri Business Service Manager are trademarks of Proxima Technology. Specifications are subject to change without notice. All other brand or product names are trademarks of their respective owners.

Contents

 

• Management Summary
• Trends in Distributed Systems Management
  • The Late 80s
  • IT Vendors Wise-Up
  • Disintermediation in Systems Monitoring
  • Agentless Monitoring
  • Current Day
• Disintermediation with Centauri
  • Step One: Planning for Disintermediation
  • Step Two: Putting in Place Your Disintermediation Strategy
    • Monitoring Oracle, SQL Server and other JDBC Databases
    • Monitoring Microsoft Windows and Windows Applications
    • Application Management: mySAP
    • Monitoring Basic Response Times
    • SNMP-Your Monitoring Catch All
    • Initiatives to Keep in Mind: SOAP, JMX and WBEM
    • Grande Finale: Now Switch Off Your Agents!
• Summary