Traditionally, network and system management staffs administered their separate domains using tools focused specifically
on those domains. Now, server and storage virtualization have blurred the boundaries and requires departments to join together and utilize new tools to maintain an increasingly complex processing environment.
Server virtualization erases the boundaries between servers to create a single, virtual application processing resource, while storage virtualization creates what appears to be a single storage device from multiple physical devices. By doing so, network links are effectively transformed into internal links within what must be viewed as a single, very large application processing system consisting of processors and storage devices spread across a network.
In the past, system management and network management were separate disciplines. System management provided a view of the applications executing within a single system and its attached storage devices. Resources such as CPU priority, memory utilization and disk space were managed within the confines of that system.
Network management addressed the connections between systems. Remote file access and client server architectures have long required administrators to view both network and internal system indicators in order to address failures.
But with those types of application it was clear where application processing resided, where data was stored, and over which network links data flowed. It was clear which problems belonged to the systems support staff and which belonged to the network staff.
Recent advances in network-wide server and storage virtualization blur these boundaries and require an integrated view of applications and the links between them.
In a virtual system, operating system software creates multiple separate environments. Each environment appears to the software executing within it to be a complete computer system. Many such environments can be created within a single system, so many applications can execute simultaneously without interfering with one another.
Operating systems have offered multiple, independent environments for many years. Recent advances have added two more capabilities.
In the past, each application environment offered the same services as the native operating system. An environment on a Unix system offered the same services and operating system interface as the native Unix system. All applications on the Unix system had to be designed to operate under that variety of Unix.
Now, virtualization enables applications written for one operating system to execute in a virtual environment under a different operating system. A Windows application can execute on a Unix system, or a Unix application can execute on a Windows server.
The second new capability enables multiple servers, each running virtualization software, to appear to be one very large system. Since applications can execute on a server independent of the operating system, any application can run on any available system. Virtualization management software monitors the load on each system. If an application currently residing on one system requires more resources, management software can transparently move it to a system that has available resources. If a system fails, all of the applications running on it can be moved to another system.
Storage virtualization means that when a data set grows and will not fit on a physical device, management software can move it to a device where space is available.
Systems and storage devices can be located in one data center connected by a LAN or can be distributed geographically, connected via a wide-area net.
Management software must adapt to deal with virtualization. Traditional network management software has dealt with specific devices and their status. But with virtualization, it isn't very useful to find out that a particular switch port failed. Which applications will fail as a result?
It was easy to determine the impact of a failure when applications and data were confined to a specific system. But with virtualization, any application can be executing on any system at any time, and data may be located on any device. Combined system management and network management staffs must have tools that enable them to pinpoint problems and take action to deal with them.
Network management and systems management software have traditionally come from different vendors or from different organizations within a single vendor. Now, vendors such as HP, IBM and CA are bringing together products to address virtualizaton. HP's OpenView, IBM's Tivoli and CA's Unicenter products bring together both network and virtualization management capabilities under a single framework.
In addition to the evolution of existing tools, new tools are being developed to view the combined server, storage and network system as a single application processing system. For instance, BalancePoint from Akorri provides an overall view of how efficiently the system is performing. Tom Joyce, vice president of marketing at Akorri, said, "Most of storage management and network management is about availability.… What is needed is a single set of metrics that describe how well the system is meeting end user needs."
Virtualization requires network and system administrators to widen their scope and expand their set of responsibilities. It will require the evolution of existing tools and the development of new ones to deal with the added complexity that virtualization brings.
About the author:
David B. Jacobs of The Jacobs Group has more than 20 years of networking industry experience. He has managed leading-edge software development projects and consulted to Fortune 500 companies, as well as software startups.