What are the challenges of data center virtualization?

Nothing really goes wrong, The key is take your time, don't try to rush the conversation of every computer in one night. I would also make sure your purchasing the correct hardware. We went cheap on our SAN, then 6 months later we had to upgrade it to the more powerful one to handle our SQL Database. It would have been more cost effective to purchase the correct one from the start. Sometimes you think the Sales guy is just trying to upsell you, but he may actually have a good technical reason why you need the more expensive product.

Judah,

Great comment about upselling. I definitely play hardball when people try to upsell me. I guess it's because I see it everywhere these days. Especially, with the economy being so bad. It's easier to be skeptical about upgrading, but in the end, you go through more of a hassle than if you would have just upgraded. I appreciate your feedback and I'll try not to make the same mistake. Thanks again for your advice and response.

-TG

Networking is a key feature, you need to be sure you´re ready, in my experience companies need trust in the security you provide for the servers, there are some security issues in virtualization today, mainly because of planning like Anand said.

To be sure you have the big picture for a highly virtualized enviroment (30 servers or more) I suggest to break apart your network into vlans by knowing your users needs for each service, most times users or servers do not need to see each other, and there you have bandwidth optimization oportunities and security.

A lot of people don´t stop to plan this point, it can be a big problem when you have 10 vm`s running in a host with 2 nic`s, i suggest to always use exclusive nics for the virtual switches (do not share them with the host, this will save you from a lot of headaches, espetially with Hyper-V).

High availability is a key issue in virtualization scenarios because clustering is easier than ever, there are some key benefits in the financial matters, because now you don´t need for example a SQL Cluster to be proected against a physical server problem, which means you just need the virtual OS license to prevent this, it´s not suggested for every application, but that depends on the business needs and size.

Hope this help you.

MP

VM-VLAN,PROD-VLAN,BACKUP-VLAN....divide it this way
because these 3 different traffic types needs isolation
also you need to see the storage side for these Servers hosting VM's
go for the HBA/Switches that supports NPIV(N-Port ID Virtualization) technology.
--create zones for VM's and make sure all VM;'s are in the same zones (for Vmotion sort of vm migration technologies)
--use performance monitoring tools build in windows/unix for I/O monitoring, Network utilities and storage vendor tools for I/O as well.
---snapshots would be an awsome technology while dealing with vm hosted boxes, rather then traditional filesystem scan backups
soo
Planning and Monitoring your Planning(post implementation) will lead to successfull virtualization implementation.
at the upper management level,
IT INfrastructure Management tools
Change Management tools
Configuration Management tools
Problem/Root cause analysis Tools
these also need to be given re evaluation

Ted, the contributors above are absolutely correct when they talk about planning. Start with understanding what you have today? Do you have mostly small file/app servers? Or do you have high volume transaction database servers? Or do you have high volume (network traffic) web servers?

I wouldn't recommend putting high volume (network traffic) servers on VM's nor would I suggest putting high transaction volume databases on VM's. However, you want to put as many small servers on VM's as possible.

Are your servers internet facing or internal? Do you have prod and non prod networks? Do you have a layered network? Do you want to allow mixed environments on your Virtual host servers? Do you want your VM environment clustered? Where do you want the cluster, at the same site or at a different site? Do you want the cluster for load balancing or for DR?

The biggest challenge as mentioned above is good planning and having a clear vision for what you want to accomplish with virtualization. If you don't plan for what you want to accomplish, you will find the benefits offered by virtualization will elude you.

Good luck.

I think the biggest challenges in virtualization are not technical but rather administrative and logistical. For example, at one of my clients we decided to virtualize many physical servers onto a shared ESX infrastructure. This altered how different support staff maintained their own physical servers. Previously, all staff could connect to the ILO cards for support and had full autonomy on their respective physical servers.

With the shared ESX infrastructure in place, there was a dedicated team that became owners of all virtualized servers. This caused much confusion on who was responsible for the newly migrated servers and caused turf wars because not everyone had access to the ESX management tools.

Before virtualizing, I think its important to determine how your technical support and administration model will be impacted

Ted, you have received several really key answers so far, including changes to your support structure and decisions about which services need to be consolidated.

Here are some other considerations.

1. Accountability of systems gets a little bit more complicated.
In smaller organizations, you might have a combined team that handles servers and networking. Storage is probably handled by the server team. These environments will generally be fine from an operational standpoint when virtualization and consolidation is introduced.

For larger organizations, however, there will be more complexity, as your server, storage and networking teams are usually separate, with clear boundaries between them. Who will be responsible for the virtual network of your VM environment? What about the storage? Do you keep the same boundaries, or do you let the server guys make more decisions in other areas, or do you setup collaboration teams?

2. Troubleshooting gets a bit more complicated
Who comes running when a particular application is unavailable? Most organizations have a reasonable understanding of the problems they encounter such that they know if the server team, network team or security team needs to be involved. What happens when you can't tell immediately, due to the integrated nature of a virtual environment? Have you updated your incident handing procedures to account for that? Are you making sure your team's skills are being upgraded to address the knowledge they now need to have to support the environment?

3. Planning, planning, planning
A properly functioning, scalable and secure virtual environment needs to be well planned. Throwing it together and just running with it will create significant problems that take a long time (and much expense) to resolve. You cannot make the same assumptions you make for a physical server environment, and you have to take redundancy seriously because your decisions simultaneously impact 3-4x or 8-10x more machines per physical box than they do in a non-virtualized environment.

4. Procedures and policies
Be sure to implement and enforce deployment procedures after the virtual environment is up. You don't want people building a server instance in the internal corporate network, and then giving it access to/from the internet without going through all the right channels, just because it's easy to flip a switch and move things around. All the upfront planning in the world doesn't help you if you don't ensure a smooth ongoing operational state.

Virtualization done right is an expensive but worthwhile investment. Plan well, purchase properly, monitor regularly and reap the benefits.

-ASB: http://xeesm.com/AndrewBaker

Ted,

1. We took advantage of was the VMware Capacity Planner, which provides the following - Gain insight into IT resource utilization and develop a virtualization roadmap for server containment and consolidation. VMware Capacity Planner can help you plan for capacity optimization and design an optimal solution to achieve maximum performance. Here is a link to the page that describes the service. http://goo.gl/5NJ8 Its no free, our vendor sponsored it for us. gave us good insight into our env and a proposed virtual env.

2. Microsoft has a free offering - Microsoft Assessment and Planning Toolkit same concepts.

3. Depending on your storage connectivity (fiber channel or iSCSI) you should consider segmenting that traffic from your normal traffic.

4. If you are going to P2V host machines, consider whether they have an OEM version of windows, because when you P2V a machine and bring that machine up in your virtual env, you will find that you will be prompted for a windows license key and you may find yourself reinstalling windows with your Microsoft supplied license key.

5. You may want to consider not virtualizing one DC, a DC that is paused or stopped in a virtual env for an extented period of time may start having timing issues and 2042 event id errors.

Hope this helps, you can validate these with your virtualization vendor.

Capacity Planning and Management is a huge challenge. In fact according to Forrester study it is #1.

Here is my list:
1. How to identify current performance problem?
2. How to predict future ones?
3. How to decide where to place VMs?
4 How to correctly size resources for each VM
5 How to identify and remove unused VMs, snapshots, etc