Virtualization Management Twist: EMC Bridges Virtual and Physical to Help Spot Problems

EMC Corp. announced today it is rolling out a new version of its network management system designed to fix one of the features that make virtual servers easy to use but difficult to maintain.

EMC’s Smarts Server Manager is an update of the company’s existing line of Smarts network and systems-management applications, which it acquired when it bought the privately held Systems Management ARTs, Inc. in 2004.

The new edition comes with the ability to bridge the gap between physical and virtual networks to help identify the root cause of an error that might otherwise be difficult to identify.

Virtual server infrastructures do more than eliminate the need for an application to know much about the hardware on which it’s running and make it possible for more than one application to share a server, according to Bob Laliberte, analyst at Enterprise Strategy Group.

They also sever the one-to-one connection between application and server—which traditionally helped data-center managers identify the machine that was experiencing the problem when an application reported having problems.

“Now you’ve got four or five applications running on a single physical server and they can move,” Laliberte says. “They could be on this physical server today and that one tomorrow.”

Add in virtual storage technology that eliminates direct physical connections between servers and storage devices, and maybe cloud-computing services or interfaces, and identifying the root cause of a problem becomes a disaster in itself.

“It’s hard to manage an environment when your applications are playing hide and seek and your storage is lying to you,” Laliberte says.

Minor hardware problems such as the failure of the only interface card between one server and the network don’t appear in network-management application consoles as clear indications of a specific problem, according to Jim Frey, analyst at Enterprise Management Associates.

They appear as an “alarm storm” made up of failure notices from applications trying to access that server, help-desk complaints from users who can’t get to their data or applications, and network alarms identifying bottlenecks and congestion in the network around the failed part, he says.

“When you’ve got a virtualized infrastructure it’s much harder to know what’s causing this,” Frey says. “Each of these servers is a mini ecosystem, with multiple VMs running on the same server with a virtual switch running between them, and if the virtual switch fails it can cause the same problems as a physical switch.

“In that kind of environment, it goes beyond valuable to be able to apply root-cause analysis to identify a problem,” Frey says. “It’s necessary.”

Smarts Server Manager—a set of applications EMC bought along with Smarts in 2004 and extended to cover its own storage product line as well as the networking products the systems then covered—is designed to identify both physical and virtual servers and identify all the connections and hardware or software dependencies of each, according to EMC.

The system uses a database called the Smarts Codebook Correlation engine to identify the behavior of specific products, and the Smarts Root Cause Analysis tool to track back through a pattern of alarms to let an operations-center staff know that they can fix problems.

The virtual-server components of the system are designed to cover VMware ESX servers and virtual machines as well as machines running cluster services software from either Microsoft or Veritas, as well as Microsoft’s Virtual Center last-generation virtual-management server.

An update later this year will be able to manage Microsoft Hyper-V-based VMs, an EMC spokesman says. While the VMware-enabled version of the product is new, the software itself is not; EMC has been using it for almost two years in its services division and decided to make it a separate product only late last year. It’s also not the only product EMC’s first shot at managing VMs or cloud-computing environments.

“Smarts has been around for a long time, though,” Laliberte says. “Its root-cause analysis is pretty bulletproof.”

The system can also exchange both information and control functions to integrate tightly with VMware’s own virtual-server management applications as well as third-party management software from BMC, HP, IBM and other competitors.

“The EMC Smarts people have done something here that hasn’t been done well yet in large scale, which is mapping the virtual to physical servers and clusters,” Frey says. “This is not something you’d buy standalone to manage your virtual infrastructure, but for the Smarts users, this is really valuable.”

It is not cheap, however. Prices for adding virtual-server management to Smarts systems start at $30,000 for current Smarts users and go up based on the base price of Smarts itself. Smarts is priced according to the number of devices being managed.

Still, the ability to map and troubleshoot virtual environments is becoming increasingly important, Laliberte says.

“In our most recent research, of the companies we survey 74 percent were using virtualization in production environments and 39 percent were using it in tier-1 critical applications—SQL databases, Exchange, Sharepoint, other things they identified as being critical to their operations,” Laliberte says. “Once you’ve made the switch, from using VMs in test and dev environments into using it in production, then into tier-1 applications, there’s no fooling around anymore. The ability to troubleshoot problems becomes really critical.”