Cole: What is your definition of a virtual data center?
Schulz: My definition is an environment that is highly efficient, highly optimized and highly productive, where you are able to do more with available resources, available power and a pooling capability -- but you also have the agility to adapt to changing business dynamics and new workloads.
It's more than just consolidation. That is one aspect, but there are also the aspects of using that efficiency and virtualization to support business continuity, disaster recovery, high availability, load balancing and other performance issues.
Virtualization 2.0 is not the desktop and it is not the cloud -- it is life beyond consolidation. Ask somebody why they haven't virtualized all their servers and they say they can't consolidate everything. The association is that people think virtualization is only for consolidation. But it's more about how to be more flexible, to move services and storage around to meet the needs of different workloads. It's a way to leverage resources more effectively and address those that can't be consolidated. It's using virtualization for operations, efficiency, agility and flexibility.
Cole: How will it improve operations over current architectures?
Schulz: For functions like business continuity and disaster recovery, one aspect is that you have servers and application servers that don't lend themselves to consolidation. To maintain high availability and business continuity, you put applications on a virtual machine within their own dedicated blade. The machine is there for live migration, where you move the applications from one physical server to another when you need to do things like planned maintenance and load balancing. So there are always a few physical servers standing by for the VM to fail over when the need arises and you don't end up paying for extra hardware.
If you have two clustered servers today, they may be on active/stand-by mode -- one active and the other stand-by. In storage systems, if you have a dual quad controller system, you run those in active-active mode -- both controllers are doing work, but if one fails, the other takes over. Why not take that paradigm to your servers? Make both of them active so they are both productive and you don't have one as strictly a cost center, just in case. You end up using resources more efficiently to actually get work done.
Cole: How does today's data center manager get from here to there? What steps should he take?
Schulz: The first step is realizing that virtualization means more than consolidation. It's a tool that provides abstraction, transparency, agility and flexibility. You should be looking at your different applications, servers, storage, data sets to determine where the core issues are. Then you'll want to start using virtualization to support data migration for business continuity/disaster recovery -- going to that awareness of what the technology can be used for. Where are the pain points to leverage the technology beyond consolidation?
The third step is where you need to start re-architecting, re-optimizing and looking at it as a tool for hardware as well as software savings. If you bring 10 servers down to one server, you still have 10 virtual servers, 10 operating systems, 10 licenses, 10 tool sets, applications and other things that need to be managed. Go and re-architect software-wise, as well as in your processes and procedures. You've saved on hardware, but what about software and personnel expenses? You need to start looking at overall infrastructure resource management in which you optimize the entire architecture across the server, network and storage facilities. Start re-architecting the way things are done.
Cole: Aren't some vendors, namely VMware, already moving beyond the concept of the virtual data center and into cloud architectures?
Schulz: The interesting thing about the cloud is that it means virtually anything. It's a different way of describing the virtual data center. Cloud computing is nothing more than another tier of computing. Cloud storage is another tier of storage. Cloud management is virtualization with added transparency.
Basically it is a higher level of abstraction; you're just not worrying about the details. Virtualization is a layer of abstraction too, but the cloud is the buzzword du jour. A virtual data center knows no boundaries -- it's all virtual, all distributed. The cloud can be all localized. It can be a product, a service, an architecture, a solution, a concept. So the line between what is a cloud and what is a virtual data center is so narrow as to be one and the same.
The general paradigm of the cloud has been around for 20 years. The original information-as-utility model described one big cloud that you plugged into. But virtualization is a real technology. Is the cloud an actual product? It could be some vendor selling cloud services, products, technologies. Virtualization is a tool, a technique, that can be used for enabling the cloud. Clouds leverage virtualization. They are very tightly intertwined.
Cole: Are all data centers destined to become virtual data centers, then?
Schulz: Yes, to different degrees. Will all data centers adopt virtualization and adjust themselves to be more flexible, agile and have different tiers of resources -- on-site, local, remote? Yes. Will all data centers move onto the Amazon cloud? No. Will all applications stay in traditional environments with no virtualization technology whatsoever? Maybe, maybe not.
On a going forward basis, most IT organizations will have a mix of local and remote, public and private -- different tiers of storage, network processing and production that will effectively become the virtual data center.