Exclusive: Vieo takes AIM and scores a hit

Managing today's complex datacenter, it’s often difficult to see the forest for the trees. Sure, you can meticulously monitor servers and applications, but given the fragile nature of the environment, application performance can degrade even when individual hosts and databases seem normal.

Free IT resource TechNet: More ways to know it, share it, and keep it running. Sponsored by Microsoft Free IT resource Attend the SOA Executive Forum: Breaking SOA Bottlenecks SOAExecForum.com/may2007 Sponsored by InfoWorld Vieo 1000 v1.1 Vieo, vieo.com Excellent  8.7 criteria score weight Ease-of-use 9 25% Integration 9 25% Manageability 9 20% Setup 8 20% Value 8 10% Cost: Starts at $150,000 for 24 managed nodes; expands to a total of 144 ports in 12 port increments Platforms: Vieo monitoring agents are available for HP-UX, Linux, Solaris, and Microsoft Windows Server 2000 (IIS5) in addition to SQL Server 2000 Bottom Line: The Vieo 1000 represents a fundamental shift in delivering application QoS. Unlike traditional systems management tools, it directly connects to each host and network component and continuously learns about their behavior. The Vieo 1000 then reacts to datacenter changes by dynamically provisioning existing resources. As a result, your most critical functions are maintained at top performance. About our Reviews and Scoring Methodology

To address this dilemma, a new IT discipline has emerged called AIM (adaptive infrastructure management). Perhaps today’s best example of AIM is the Vieo 1000 AAIM (Adaptive Application Infrastructure Management) appliance. This new breed of management solution not only understands the whole IT environment — servers, applications, and networks — it also interconnects all these resources so it can reallocate them in near-real time to, for instance, meet unforeseen Web traffic.

The Linux-driven, custom-designed 2U rack-mounted Vieo 1000 appliance includes a controller blade, plus up to two management blades that contain fast Ethernet and Gigabit Ethernet ports. In this standard configuration, you can manage up to 24 hosts or network components; an expansion unit holds four management blades that can scale this solution to 144 management ports.

From the outset, Vieo 1000’s simplicity was apparent. Architecturally, it appears much like a layer-2 switch. As such, anyone with basic networking experience should be able to handle setup. Wiring the mulitplatform test environment and adding the Vieo 1000 to the network took just a morning. You should plan on another few hours to install Vieo Agents (which monitor host, application, and system resources) on application servers. After I installed the agents, the Vieo 1000 successfully discovered all the systems in the test environment and was ready for productive work.

Internally, the Vieo 1000 hardware capitalizes on a high-speed InfiniBand I/O architecture and asynchronous communications. So there wasn’t any noticeable difference in network response after installing the appliance. Moreover, agents required a scant 45MB of memory; on a single-processor Pentium III system, the software consumed less than 2 percent of CPU resources.

Vieo 1000’s application console, a well-designed Java dashboard, gave me a clear-cut view into everything happening within my test datacenter — and a simple way to manage it — from one spot. I clicked on any object (for instance, a Web server) and instantly saw highlighted the associated application server, database, and networking services. This visibility is often missing from standard management tools that monitor individual components.

The Vieo 1000 achieves its highest level of effectiveness by first learning about the datacenter environment, an automated process that runs from a few days to two weeks.

During this time, the appliance adjusts knobs and records gauge settings. The virtual gauges measure the console’s surroundings (such as Web server response time) and the knobs control managed elements (such as the number of application servers). This process yielded an accurate datacenter model, including the role of each application server and the transactions that typified my Web site.

Next I created software templates that the Vieo 1000 applied to provision servers — a simple and quick step. I merely verified that my storefront application was working properly and then specified which directories and files on the reference server would be required to imitate the host.

After reviewing training results, I felt the Vieo 1000 did an excellent job of determining the best overall configuration for my test setup. But sometimes you may want to set different AQoS (application quality of service) policies, such as giving priority to a revenue-producing application during specific hours or to a better Web response time. And here’s yet another unique part of this product: Even though you can drill down to see and manage individual servers and parts, such as network cards, the Vieo 1000 rolls up specific resources.

As a result, I was able to simply click on Store Front in the navigation pane of the application console to receive top-line graphs of actual and target response times for the entire tier.

With this display, I was able to specify a slightly lower response time to be achieved; it was then up to the appliance to use its acquired knowledge to meet my policy setting. That’s just was AQoS should be: relief from sweating the details.

The pinnacle of my Vieo 1000 test was turning on Adapt mode and watching the appliance go into action. There are many possible scenarios, from alerting you to probable causes of AQoS failures to automatically applying recommendations. In the first case, when I increased Web traffic and response time faltered, the system displayed likely reasons, including high memory utilization on an application server. Having this information in such precise form is an invaluable troubleshooting aid for IT staff.

For the second test, I requested the Vieo 1000 to automatically apply an aggressive AQoS recommendation. In this condition, based on learning, the appliance determined that an additional WebLogic application server would be needed to handle the extra load. It then selected a server in the idle resource pool, picked and applied the appropriate host software template, and added the server to the application server tier — all in about 10 minutes.

At first, I was skeptical of entrusting the Vieo 1000 with such a sweeping approach to managing systems at the heart of an organization. However, my month-long evaluation prooved such fears to be unfounded. This ingenious device never faltered in preserving AQoS when confronted with a variety of network and system disturbances. There are many fine monitoring tools, and other AIM solutions will likely surface in the next year. But I’d venture that Vieo 1000’s simplicity, strong analytic tools, and agility will be hard to match.