Upgrading to a virtualised environment

Oh man! Ad-blocking software has been detected! :'(

This website is run by the community, for the community... and it needs advertisements in order to keep running. Blocking our ads means your killing our stats!
Please disable your ad-block, or become a premium member to hide all advertisements and this notice.

Well the company I work for has decided to invest in virtualisation I am currently designing the network to provide redundant setup so that if one host server (vSphere) was to fail the secondary server should take over. The setup I am proposing is to store the virtual images on dual NAS setup but a little unsure how to go about this the idea was to use DFS namespace but after looking into this im not sure. I would like it if one of the NAS's were to fail for what ever reason the virtual servers shouldnt be effected (failover solution). Does anyone else use a dual setup which hosts their VM images? How do you provide redundancy for your images? OR am I going about this the wrong way? I did consider storing the VM images locally on the host machines but it seems more common to use SAN or NAS setup.

Any comments would be appciated.

 

Oh man! Ad-blocking software has been detected! :'(

This website is run by the community, for the community... and it needs advertisements in order to keep running. Blocking our ads means your killing our stats!
Please disable your ad-block, or become a premium member to hide all advertisements and this notice.

First of all it's best to know what storage you're planning on using? What are you currently using and have you thought about whether this is going to be re-used or whether you're buying new storage to start from scratch? Are you planning on FC, iSCSI or NFS datastores?

Ideally you want your storage array to have the redundancy built in (dual controllers, dual network switches, dual power supplies (phases), multiple nics in each ESXi host with multipathing enabled and configured.

Unless you're going down the whole HA\FT route and using SRM for site recovery then you could potentially make life very complicated for yourself for no real reason.

Finally what kind of budget are you looking at spending? How many VM's? How much growth are you expecting over say the next 3, 5 and 10 years? What are you expecting your ROI to be?

So many questions to ask and not enough information to really solution this from a single post.

As a suggestion I would probably suggest having a read of the VMware vSphere Design book to get the idea on the questions you need to be asking.

 

Simon thanks for your reply we currently use a stand alone NAS for network data. For our new solution we are quite restricted due to the nature of my job. We're looking to use dual NAS configured (shared storage) with NFS data store. For the NAS we're looking to use dell power edge servers and yes I was looking to implement HA, FT and vMotion.

Currently we run 7 servers for general network purposes AD, DNS, DHCP email etc. hoping to vm each of these.

- - - Updated - - -

The Redundancy in storage is the biggest worry the network needs to be available as much as possible ideally 100% apart from scheduled downtime. Does anyone use vSphere Appliance Storage? Looking into it would provide the reliability required and also allow for HA, FT and vMotion configuration. The setup we are looking to implement is the use of 2 host machines.

 

As Simon said, massive can of worms to be opened before going down that route. However a couple of major hurdles that must be looked into before even looking into the design element are:
What OS's are you currently using because if they are OEM's your technically be breaking MS's software agreement by p2ving them if thats what you are thinking of doing.
Have you performed an IO collection phase and do if so do you know how many IOPS you need (including peek usage periods within your business). No ones going to thank you for spending money on the infrastructure if things are slower after the spend, it should be equal too or faster than the previous setup.

What RTO and RPO would you be looking for and how are you currently backing up/ protecting your environment. You can get great performance gains out of ripping out the old agent based backup/av products and using things like veeam and kaspersky which are built for virtual environments.

If cost is a major factor it might be possible to reutilise your existing physical servers as storage or there are options for using local datastores these days, In our company we champion things like EMC VNXe, Netapp 2040's or ScaleComputing S & M series clusters for small shops. For example a ScaleComputing cluster can be had for about 9K with 3-6TB usable.

But As I say before you need to measure your current infrastructure before even thinking of buying new shiny toys (as tempting as that is the boring stuff must come first).

Sounds like a lot of fun.

- - - Updated - - -

FT isnt available in the essentials plus pack you will need to look at the big money versions to get FT in. YOu wont really need a dual NAS setup just buy a SAN that is truely fault tolerant, all the ones I mentioned previously are, so you would be looking for storage device that has dual power supplies, storage processors etc etc. Also its worth asking your preferred reseller to give you contact details of a customer who has one of whatever you are interested in. Most resellers/consultancies have them and it allows you to get a real world information on how stuff works.

 

Thanks for input obviously a lot to consider here.

 

Hi Snake

Some valid points made by Dales and Simond, I do this for a day job, as Dales has mentioned you will need a clear indication of the IOP's required and also a thing peeps sometimes miss is latency to the storage, so you will either need to use shared storage or fast internal storage (this has a few gotcha's but with Hyper-V 3 and Esx 5.1 you have a migrate-all ability) To re-utilise existing servers as backend San is do-able however you would need to put some sort of storage virtualisation between, such as Falconstor, Datacore, Nexcenta, openfiler etc, with a few of these you could actually use some SSD and have this as a hot cache giving you amazing read performance, also most of these solutions allow for snapshotting, replication (async and sync) and failover if setup correctly. This will give you resilience, maybe the way to go. Otherwise you will need to look at replicating vm's to a warm standby cluster somewhere using internal or external storage, using the replication ability of the storage arrays (most mid level arrays do this) or lower level storage arrays and a software replication solution. There are many of these CA RHA. Double take etc.

Hth

Regards

SteveL