A (barely) passing grade
Web-based tools help net management vendors boost rating in annual survey, but age-old problems persist.

By Neal Weinberg
Network World, 5/11/98

The move to Web-based management tools is on and users are reaping at least some of the promised benefits, but the overarching, age-old problems of automation, reliability and integration of network and systems management tools remain largely unsolved.

In a nutshell, those are the most telling findings of the 1998 Network World Network Management Survey. Sponsored by Deloitte & Touche Consulting Group, the survey is based on telephone interviews with 202 Network World readers who use network and systems management tools or supervise people who do.

If responses to this year's survey can be boiled down to a letter grade, vendors scored only a "solid C'' on their ability to deliver features readers are looking for, says David Lee, senior manager with Deloitte & Touche. That's slightly better than last year's C-, but it's a long way from the honor roll.

Overall satisfaction with the level of improvement of network and systems management products was 3.72 on a scale of 1 to 5, with 5 being "extremely satisfied." That's higher than last year's score of 3.44, which was the low watermark since Network World began conducting the survey in 1994. However, the 1998 score falls short of the all-time high score of 3.85 in 1996.

The most striking difference between this year's results and those of past surveys is the influence of the World Wide Web. Nearly one-third of respondents said they are using Web-based management tools, as compared to 18% in 1997. Users were particularly happy with the ability of Web-based tools to enable them to manage their nets from anywhere with an Internet connection (3.90); their ease of use (3.81); the ability to share management information with more people (3.76); and the ability to give end users information they can use to help themselves (3.73).

Web-based tools fared somewhat worse in their ability to aggregate information from multiple sources (3.45) and correlate management information (3.54). John McConnell, president of McConnell Consulting, Inc. in Boulder, Colo., says that's because many vendors simply grafted a browser interface on existing management products tackle those chores rather than develop Web-based tools from the ground up.

Lee points out that some end users expect to be able to check the status of a trouble ticket with a Web browser no matter where they are, no matter what platform they are on and no matter what platform the information is on. But vendors simply haven't invested enough time in this area, he says.

At United Parcel Service (UPS) of America, Inc., for example, systems architect Richard Weiss is looking for tools that will allow him to send alerts from the central console to a remote IS person's e-mail account. He also wants to post statistical reports on a Web page so groups designing network applications can have easy access to the information. And he wants end users to be able to track the progress of fixes to their particular system.

McConnell predicts the next generation of Web tools, which are just beginning to hit the market, will provide "true Web-enabled management,'' with features such as real-time updates, pop-out windows and the ability to organize information into easily searchable categories.

Wrestling with the Web

McConnell adds that there's a shift underway in the network and systems management world from just maintaining boxes to maintaining service levels, which are agreed-upon levels of application performance. Tools that can monitor service-level agreements (SLA) are "still relatively immature,'' but within six months to a year, they should be ready for enterprise use, he says. An example would be a tool that measures the response time of a Web-based transaction.

Lee agrees. Network managers have traditionally focused on just keeping the network up, but now they are increasingly concerned with managing the performance of a particular application an end user is running.

Unfortunately, that's where the current tool kits receive the lowest scores. Readers said they were especially dissatisfied with their ability to optimize application performance (3.19) and troubleshoot application problems (3.30).

"We want to know [about a problem] before the users know,'' says Scott Herron, senior systems engineer at Pharmacy Research Associates in Charlottesville, Va. Herron is satisfied with the network management capabilities of Hewlett-Packard Co.'s OpenView, but he wants "something better than the default tools that come with Windows NT'' to manage his distributed NT-based system.

So he is looking into ManageX, a systems management product that can monitor SLAs, issue warnings of potential application problems, and take automated corrective action. It can also is-sue Web-based reports. ManageX was developed by start-up NuView, Inc.; HP bought the product in November.

Other products that McConnell says will help address service-level and application monitoring issues include:

• Visual UpTime, from Visual Networks in Rockville, Md., which monitors WAN service levels.
• WiseWan from Net Reality, Inc. in Sunnyvale, Calif., which manages Web bandwidth congestion.
• VitalAnalysis and VitalHelp from Vital Signs, Inc. in Santa Clara, Calif., which monitor application performance from the end user's standpoint.
• S3, or Street Savvy Software, from Next Point Networks, Inc. in Westford, Mass., a Web-based product that offers real-time service-level information and application monitoring.

Whither frameworks?

Integration of network and systems management information is the issue that management frameworks such as Computer Associates International, Inc.'s Unicenter TNG and the IBM's Tivoli TME were intended to tackle. But only 25% of respondents said they are considering one of the frameworks. The good news for framework vendors, however, is that that number is up from 19.5% last year and from 22% in 1996.

By contrast, 55% of respondents said their strategic direction was to use an SNMP-based plat-form as their network and systems management foundation. Jeff Hersh, a manager at Deloitte & Touche, says the issue comes down to complexity. "The year 2000 is coming up rapidly, and companies don't have 12 to 18 months to implement complex solutions,'' he says.

McConnell agrees. "Certainly, right now, a lot of the bloom is off the frameworks,'' which are seen as expensive and difficult to implement.

Clark Stockdale, MIS director at Filmet, a Pittsburgh-based printing and photography company, is looking for an enterprise-wide management system so he can create a single help desk for his six locations. He looked at Unicenter TNG but decided the integrated solution was "overkill'' for a company of 350 employees. "It seems like more than what I'm looking for,'' he says.

UPS came to a different conclusion. The company, based in Mahwah, N.J., is moving from a combination of home-grown tools and individual products to the Tivoli TME framework. UPS has spent the past two years consolidating six independent IT groups under the Tivoli umbrella.

If that seems like a long time, consider the scale of the undertaking. UPS' Weiss says the combined system will include 12 mainframes, 90,000 workstations, 1,500 Windows NT servers, 2,300 NetWare servers, 3,000 OS/2 servers and hundreds of AS/400s. He adds that one factor that weighed heavily in Tivoli TME's favor during the yearlong vendor selection process was its scalability.

"Before, everything was discrete. We had six different groups doing six different things. We're using Tivoli to bring all the areas together into one central facility to see how a problem propagates itself across the system,'' Weiss says. "We're in a reactive mode right now; it's hard to diagnose where a problem is by looking at symptoms.''

For example, if a database administrator sees a slowdown in performance, it is difficult to deter-mine whether the problem is hardware, software or network related because "each area is being addressed by different groups.''

With the new system, Weiss hopes to save money and improve service by "not having two groups chasing the same problem down'' and by being able to identify potential problems before they occur through a sophisticated set of alerts.

The past two years have been spent in development work, and Weiss is now in the testing phase. He expects to be up and running this summer.

Seeking to centralize

But the goal of centralization remains an elusive one, with only 28% of respondents saying they have been able to pull it off. A full 50% are still managing components separately using element managers, and another 21% are managing networks as one entity and systems as another.

For example, Sealmaster Bearings in Moorhead, Ky., has 14 sites, each handling its own net-work, according to network administrator Paul DeBord. He is currently looking at enterprise-wide products from CA and HP that will monitor his LANs and WANs, notify him of potential problems and track usage across the network.

Rusty Sweet is one of those skeptics who doesn't believe such a product exists. "If I could get one product to manage all my equipment, I'd be a very happy man,'' says Sweet, network man-ager for GTE's telecommunications division in Indianapolis. He is sticking with point products to manage 4,700 routers, switches and bridges. One reason behind his decision is that he already has specific tools that go with specific vendor products, such as CiscoWorks for his Cisco Systems, Inc., routers.

Sweet's biggest complaint is with the reliability of HP OpenView. He concedes that "it's a far cry better than it used to be,'' but he still gets annoyed when OpenView traps or identifies a problem, but fails to display or pass along the alarm, or when it misses a problem altogether.

In fact, reliability topped the list when readers were asked what items would increase their level of satisfaction with network and systems management tools. Other features that scored high on reader wish lists were automating processes that give quick responses to end users, reducing complexity, cutting costs, automating repetitive tasks and boosting scalability (see Figure 1).

For their part, vendors claim to be making strides toward improving their products and responding to customer concerns. Olivier Helleboid, general manager of HP's OpenView division, says HP has improved OpenView's ability to pinpoint the root cause of an event and to automate a response.

The company also added another layer of management capabilities to OpenView that allows a network manager to keep tabs on the performance of an enterprisewide application from SAP AG, for example.

Mike O'Rourke, vice president of enterprise management solutions for Tivoli, says the company has made strides integrating point products into the TME framework.

Tivoli has also focused on helping network managers prioritize alerts based on their importance to the business. In the past, when two alerts showed up on a network manager's console at the same time, it was up to the user to figure out which was the higher priority. Now Tivoli's software is able to automatically determine that one alert is from a remote laptop and the other is from a server running a critical SAP application, O'Rourke says.

Tivoli has made a number of moves to beef up its Web presence. For example, TME 10 Information/Management Version 1.1, released last November, came with a new Web interface. In March, Tivoli announced Cross-Site, a suite of management products for electronic commerce. Also, you can now post reports from the latest version of TME 10's Output Manager on the Web for all appropriate parties to see.

Vendors may indeed be making improvements, but it takes time for users to become fully aware of them. Try as they might to keep up, some users say they are simply swamped.

A year ago, Larry Connors, network specialist at Seattle-based Alaska Air Groups, Inc., was struggling to come up a cohesive strategy to manage the airline's systems, which include main-frames and terminals, Windows and Macintosh PCs, SNA, TCP/IP and IPX traffic, routers and hubs, kiosks and Web sites.

"I haven't done anything on that, I've just been too busy,'' Connors says. Over the past year, he has been replacing terminals with PCs, phasing out the Macs, and adding more T-1 lines. "We're continually adding to the network,'' he says.

"It's pretty hard to pick a direction [for network and systems management] until things settle down,'' Connors says. He is looking at a product called Network Health from Concord Communications, Inc., which is supposed to detect network bottlenecks before they happen. "We don't do anything proactive at this time,'' he says.

Other respondents are making do with what they've got until something better comes along. Art DeLeon, manager of IT at Kvaerner Metals in San Ramon, Calif., currently uses Microsoft Corp.'s System Management Server (SMS) to keep tabs on seven servers (four NetWare, two NT, one Unix) and 450 desktops. He says he's content to stick with that while waiting for new management features Microsoft will deliver with Windows NT 5.0, which is expected to ship late this year or early next. NT 5.0 will feature the Microsoft Management Console, code-named Slate, which is intended to support management applications from third-party vendors. It is being developed in conjunction with a new version of SMS that is currently in beta.

And finally, there are readers who are genuinely happy with what they've got. D. Scott Starkey, network support analyst at Brinks Home Security in Irving, Texas, began installing Novell, Inc.,'s ManageWise last October and got it "fully tweaked out'' in January. The product is per-forming as advertised, keeping Starkey posted on his 20-server, 500-node network.

His next move is to install Novell's ZENworks product when it comes out later this quarter. Starkey is looking to ZenWorks for features such as the creation of a "digital persona,'' so a call center worker can sit down and log on at any workstation and be recognized by the system. "I'm very happy with how things are going,'' Starkey says.

Too bad he is the exception, not the rule.