I’m doing a 1-hour open Q&A session on May 8th for the PASS Virtualization Virtual Chapter. Bring your VMware and Hyper-V questions about setup, performance, management, monitoring, or whatever, and I’ll answer ’em.
You can even get a head start here – post your questions in the comments below, and I’ll build slides to answer ’em ahead of time. That way you can make sure you get the best answer possible. (Well, from me anyway, ha ha ho ho.)
Then come join us on the webcast and hear the answers. See you there!
27 Comments. Leave new
Brent, I can set up virtual machines on my laptop using vmware workstation 9. I have no issues with that. However, what I really want is to create a cluster so I can work with Availability Groups. Setting up networks seems to be the issue for me I continue to have as I am not really good at configuring IPs, subnets etc.
How do I go about setting up the network for one domain controller and 2 SQL Servers in workstation 9 so that I can create a WSFC? What type of network connection should I use in my vm (bridged, NAT etc) and what IP address and subnet can I use so that all my vms can talk with each other and still work on the internet so that I can get windows updates etc.
Keith – great question, but unfortunately this webcast is focused on server virtualization technologies. I’m not really qualified to do VMware Workstation support.
The good news is that this is a well-covered topic – just Google for how to create a cluster on VMware Workstation, and there’s dozens of step-by-step tutorials. Start with a Windows cluster first. Building a SQL Server cluster atop an existing Windows cluster is fairly easy.
Hey Keith,
I am working on same project creating WSFC on my Vmware Workstation. I had followed some online instructions. My current configuration on the VMs are Domain Controller:NAT ; Other two nodes have two network adapter each, NAT and Host-Only. The safest way to go is assign static IP addresses.
HI Brent,
My issues all relate to configuring and monitoring storage. I have VM Ware in production and Hyper-V in the Lab.
Which configuration options (VMDK/VHD, iSCSI, …) provide better performance? (We have an HP LeftHand iSCSI SAN in both environments)
Does it make any difference (in the virtual environment) if the database files are split across multiple LUNs?
Many similar questions.
Thanks
ray
Ray – that’s a great question, and I’ll cover it in the session. Can you give me a little more details on your environment?
How are you connecting from the VMware hosts to the storage? (Type of port, and quantity on each end, like how many in the server and how many in each of the LeftHand boxes)
How many LeftHand boxes do you have, and how many drives total?
What type of RAID are they using?
Do you have one network switch dedicated to iSCSI, or two?
How many virtual machines (and physical ones) are sharing the LeftHands?
So glad you’re doing another session on virtualization. We are running SQL Server 2012/2008R2/2008 Standard 64-bit on VMWare ESXi 5.1. Each host is set up with a single resource pool to which is allocated all of the host’s available resources. Each SQL Server shares the host with 6 to 10 other servers. Memory and CPU are far from over-allocated. We host a couple dozen databases per server and the databases are small, for example one larger 2012 server has 29 GB total data + log for all databases.
Each SQL Server is given a Memory Size of 4 GB with memory usage unlimited. SQL 2012 tends to run (according to the VSphere monitor) right around 4 GB. Is this because the balloon driver is keeping the memory usage down to the Memory Size amount? Under stress, will the server ever use more memory? (I’ve seen memory usage go up by as much as 600 MB but never more than that.) I’d like to ask our server unit for more memory but I’m going to have to have something to prove the server needs more, for example performance test results at 4 GB versus 6 GB.
Nicole – I’m not sure what “Memory size of 4GB with memory usage unlimited” means. Can you go into Task Manager and click Performance tab, can you take a screenshot and email it to help@brentozar.com?
Hey SQL 2008R2 Standard Edition license question for you. Vsphere 5 allows you to add vCPU and Cores. If I have a single vCPU with 2 or 4 cores is that 1, 2, or 4 licenses used?
That’s a great question. SQL Server 2008R2 licensing is by the CPU socket, so if you edit your VMware guest configuration to have 4 cores per socket, and you give it 4 vCPUs (which are cores), it’ll show to the guest as 1 socket, 4 cores – which is 1 socket of 2008R2 licensing.
Brent – What if we are using SQL 2012 instead? Licensing on VM’s always confuses me for some reason 🙁
Sreekanth – SQL Server 2012 no longer licenses by the socket, but by the core instead.
Hey Brent,
With virtualized sql servers becoming more commonly used over the past years i’m wondering if you have some guidelines regarding the maximum size (mainly cpu based) that you would feel comfortable to support. Preferably compared to the available resources on a single host machine.
Based on my experience there are always more cpu’s assigned to the different vm’s than there are available on the host(s).
Hi, Toine. Sure, I’ll address that in the webcast. Thanks!
Hey Brent,
When setting up a server that has 4 vCPU and 16 GB of Memory should we request reserver both CPU and Memory. Could doing so reduce ready time on cluster approaching over allocation of CPU? Are there any gotchas to setting a reservation for CPU and Memory.
Sorry, that should say request a reservation for both CPU and Memory. Can’t type this morning.
Hi Brent,
My question is:
We have our servers with Dediserve, so we are guests, and I realized that some days we have a great performance and SQL SERVER respond very fast, but some days it seems like the host server is very busy and the performance decreases?
Are there actions we could take to prevent this? De we have to change some sql server parameters to run better in this type of environment?
Thanks
Hector – do you have any monitoring tools to see performance at the host level?
Hi Brent,
what tools do you recomend to mesure the host performance from guest machine?
thanks
Hector – I actually wouldn’t recommend monitoring the host from the guest. I’d monitor the hosts using virtualization-specific monitoring software.
Hi,
I have a smaller IT environment – 1 DC, 1 App Server, and SQL running on the DC (I know). What is the best way to virtualize this?
Thanks,
clm
Carol – ooo, unfortunately, running SQL on the DC isn’t supported. Fortunately, virtualization is the perfect time to fix that – you could install a virtualization host, and run separate guests for each of them. However, typically in a 3-server environment, virtualization is going to raise your costs rather than lower them.
My manangement is very keen on having our already virtualized SQL Server 2008 Enterprise Edition server clustered as well. Other than being able to keep one node up while the other is under going maintenance is there really any advantage to clustering virtualized SQL Servers? Are we gaining any redundancy or reliability not already provided by virtualization?
Hi, Tom. I got several questions around that, actually, and I’ll be covering that in the webcast. Thanks, and see you there!
Hi Brent,
I have one more question that most likely will already be covered, but what is your view of having say 2 servers with 4 CPU and 16GB RAM and 1 instance each setup to use ~13GB RAM compared to having 1 server with 8 CPU and 32GB RAM with both instances (which are set to not use more than ~13GB RAM each). Both instances are to be configured with a maxdop of up to 4.
Thanks in advance.
Toine – great question. In a virtual environment, I don’t recommend running multiple instances on the same guest. VMware and Hyper-V can’t do load balancing between the two instances, moving them around to different host hardware, for example. You’re better off keeping them separated. Hope that helps!
Hi Brent,
Thanks a lot for the previous answer. Here is another question.
Which disaster recovery (besides making backups and being able to restore them) setup do you suggest for instances running standard edition: log shipping (either standard or own scripts) or database mirroring? If you would go for database mirroring what would be the limit for latency due to only being able to run in high safety mode?
When running enterprise i guess one would always run availability groups in synchronous/asynchronous mode depending on the latency. Or would you even here sometimes consider log shipping or database mirroring over availability groups?
Thanks in advance.
PS: Sorry for not having any questions directly related to the hypervisor.
Now i came up with a generic hypervisor related question.
Is it really all about IOPS?
Short question with most likely a much longer answer.