Practical guide to virtualization in data centers

Many companies are implementing virtualization technology (most notably VMware) to achieve a variety of benefits – including reduced capital spending on hardware, reduced power and cooling demands, and more flexible allocation of processing capacity. Virtualization, however, creates many new management challenges. These challenges arise from two basic causes:
##CONTINUE##
1. Virtual servers by their nature need to be managed differently from physical servers.

2. Data centers are only partially virtualized. That is, servers may be physical machines or they may be virtual servers sharing physical machines with other virtual servers.

The co-existence of virtual and physical servers in the data center adds a new level of complexity to management operations. As a result, IT managers must re-think the way they manage their increasingly dynamic mix of virtual and physical servers. With a virtualized server environment, IT organizations can access virtual and physical servers from a single interface; access virtual servers distributed across multiple VMware VirtualCenters from a single interface; consolidate events and alerts from both physical and virtual servers; granularly control access rights to both virtual and physical servers in a common manner; maintain management assignments for virtual servers even as they are moved from one physical machine to another; and granularly audit management operations performed on both virtual and physical servers

These are all essential capabilities that virtualization vendors do not themselves provide. This article explains how virtualization will impact your management operations. By taking the advice offered here to heart, IT organizations can avoid many of the pitfalls associated with VMware ownership and maximize total business returns on their investment in virtualization technology.

WHY VIRTUALIZATION?

IT organizations are under constant pressure to do more with less. They are constantly being asked to roll out more applications, support more users, deliver higher service levels, and handle more intense processing workloads. And their budgets are rarely increased proportionally to these escalating demands.

Virtualization offers relief from these demands in several ways.

One is more efficient, “multiplexed” use of server hardware. Instead of having to devote individual servers to each application, virtualization allows multiple applications to be run on a single server – as long as those applications only utilize a relatively small percentage of the physical server’s resources. This ability to run multiple applications on a single server can substantially reduce capital spending on server hardware. Lifetime ownership costs may also be lower for a single large server than they are for multiple 1U rack units.

Virtualization also offers a respite from the physical limitations of the data center. Many IT organizations have maxed out the physical capacity of their existing data centers in terms of floor space, cooling capacity, and/or power utilization. They would therefore not be able to add more servers without expanding or moving their data centers – which would be prohibitively expensive. Virtualization lets them postpone such a painful undertaking by allowing them to load more applications onto the server infrastructure they already have in place.

IN THE REAL WORLD

As appealing as virtualization may be, IT organizations are introducing it into their environments with varying degrees of caution. Some are only starting to experiment with virtualization solutions in the lab. Others have begun to roll virtual servers out into production. But even the most enthusiastic early adopters of virtualization are only applying it to a certain percentage of their new production machines. And no one is overhauling their entire IT environment to make it entirely virtual. In fact, for many applications – especially high I/O ones like email – virtualization is neither practical nor advisable. No data center will, therefore, ever be fully “virtualized.”

Different companies will implement virtualization to different degrees, but IT organizations will always have to deal with a mix of physical and virtual server resources. The reasons for this include: (1) significant investments in physical servers; (2) IT organizations are still figuring out how and when to implement virtualization; (3) they’re running into management problems; (4) and adoption rates depend on many non-technical factors. Every IT organization must therefore understand the factors that constrain its implementation of virtualization technology.

IMPACT ON SERVER MANAGEMENT

Virtualization technology can help IT organizations significantly improve the economics of data center infrastructure. However, virtualization also introduces new complexities and challenges when it comes to managing that infrastructure. Some of these challenges arise from the nature of virtualization technology itself. Some of them arise from the fact that virtual servers must be managed in conjunction with all the servers in the data center that remain “unvirtualized.” Together, these two sets of challenges significantly impact the way IT organizations must operate.

Giving specific IT staff appropriate privileges for specific virtual machines can already be somewhat complicated. Authorized systems administrators, for example, may have to be given root access while application specialists are only given access to the software running on the operating system. When multiple virtual servers are running on individual physical machines, the administration of these rights becomes even more complex. To provide an application specialist with access to just one of the ten virtual servers running on a given physical machine, for example, an IT manager has to administer appropriate privileges on that physical machine’s ESX service console which is the management gateway for both the VMware virtualization layer and the individual virtual servers running on the physical machine.

Matters get even more complicated as IT organizations implement more virtual servers. That’s because they have to implement VMware VirtualCenter to manage multiple ESX servers. However, most IT managers don’t want to give their systems administrators and application specialists access to VirtualCenter, because such access would potentially give them inappropriate control over the data center’s virtual infrastructure. So, instead, they provide their technical staff with the IP addresses and or management URLs of the specific virtual machines for which they are responsible. This manual approach to the distribution of IP addresses and/or management URLs can create problems if and when IP addresses of virtual servers change.

At this time, VMware doesn’t enable IT staff to unify their views of multiple virtual servers residing on different physical machines, if the ESX service consoles of those different physical machines are connected to different VirtualCenters. Of course, each VirtualCenter can theoretically manage hundreds of physical servers. However, IT organizations have generally chosen to deploy multiple VirtualCenters before they reach that threshold in order to segment their virtual environments by platform or location. So, if virtual servers in two locations support a single application, the application specialists in charge of that application can’t view both servers from a “single pane of glass.”

With physical servers, there is a one-to-one relationship between component failure and application outage. When multiple applications are running on that same server, the stakes go up. Virtualization theoretically insulates applications and services from component failure by enabling IT organizations to run mirrored virtual servers and/or to quickly migrate virtual servers from one physical machine to another. But not all IT organizations have sufficient idle capacity on their other VMware-enabled servers to handle the workloads that would result from the simultaneous loss of twenty or more virtual machines due to the failure of a single physical machine.

IT organizations implementing VMware have to be particularly sensitive about the vulnerabilities created by multiplexing large numbers of virtual servers onto a single physical machine, and they must be sure they have the out-of-band access to those physical machines necessary to address BIOS- and hardware-level issues. Every IT organization must ensure that it is fully cognizant of all potential management issues it will face as it embraces virtualization in the data center.

-----------------------------
BY Ryan Sia, Vice President, Asia Pacific, Avocent
Source：ECN ASIA