Atlantis Computing in a VMware View environment

A couple of years ago, I had the pleasure of being introduced to Atlantis Computing. Atlantis would help solving the storage IO issues customers were facing when implementing VDI. Basically their solution would cache storage IO in memory so that disks wouldn’t be the bottleneck anymore. Nowadays, Atlantis does way more than that and they call it “storage optimization”.

Atlantis ILIO, the product name, comes as a virtual appliance and runs on the VMware vSphere hypervisor. You need an appliance on every ESX host. Traditionally, that appliance sits between virtual machines and storage (local storage and/or shared storage). The appliance is using physical ESX memory for its operation. The appliance caches storage IO and also does inline deduplication. By doing that, it boosts VDI performance, makes you able to run more VM’s per storage device and also, you don’t need a high performance storage device. With ILIO Diskless VDI, you don’t even need physical storage anymore. VM’s are running in memory.

The ILIO solution gives you a couple of possibilities:

  1. When you use ESX servers connected to shared storage, for VDI, you could lower the specs of your SAN. You need less performance and also less disk capacity from your array.  Pick a more mainstream array instead of SSD based arrays. Eventually this comes down to a lower price per virtual desktop. Also, the “fear” around storage with VDI becomes less important. VDI doesn’t need to be difficult anymore.  This solution is a good fit for stateful/dedicated full clone desktops
  2. More and more customers are running stateless desktops on local ESX storage (so no need for a shared storage array for those VM’s). For storage they often chose SSD’s or FusionIO for performance. To save on capacity, you could use ILIO purely for deduplication but also think about the ILIO Diskless VDI option: no storage at all. All VM’s run from ESX memory. The ILIO appliance takes ESX memory and uses it as a datastore.

This week I also heard the following: use local storage for stateful/dedicated full clone virtual desktops. Use ILIO for boosting performance and dedup and also use VMware Mirage as a backup tool, in case an ESX host would fail and to backup local data and apps. Interesting thought, isn’t it?!

Bottom line, there are several solutions which can absolutely help with the VDI storage IO issues. They all have a different price, purpose and maybe even give you additional advantages. Take a good look at these solution and chose which one suits you best and gives you the lowest price per desktop.

Cloud based data solutions- Are they forgetting about VDI? Part 2

A little while ago I wrote an article about Cloud Data Solutions like Dropbox, Box.net, Mozy Stash and Project Octopus and that these products can’t be installed in a Floating Pool VDI environment today. The Dropbox client gets installed in a user’s profile so you can’t install it in your master/parents/golden image VM and deploy it to your users. Other products save their config information (server, account name/email address and password) in HKLM in stead of something which is roaming like User\AppData\Roaming. That means all users would have to fill in their credentials every time they logon to a new VM. That’s not the way you want it to be, right?

But, is this the only issue when you look at Cloud Data solutions and roaming desktop users? Let’s brainstorm some more. Let’s assume this client issue can be fixed. Fixing where it saves the credential information shouldn’t be too hard to fix, right?!

So, let’s set the baseline here; we have a VMware View environment (the issues probably apply to other solutions as well), Floating pools, refresh after first use, roaming profile/Persona Management and Folder Redirection, and a Cloud data agent which is working just fine in a floating environment.

What happens when a user logs on for the first time? First thing that needs to be done is to configure the agent. Fill in a username, password and the URL of the Data server/provider. Dropbox, Mozy and other will place their own folder under user\data folder so you are ready to go. Add data to that folder and it will be synced to the “Cloud” (and this Cloud can be on or off premise).

Second time a users logs on, there is no need to fill in the credentials anymore. That was fixed. Yay! The folder under user\data folder is there as well. It’s being roamed by Roaming Profiles/Persona Management and so is your data. Maybe you will have to wait a bit till everything is in sync.

Data is part of your profile (in this example). Is that what we want? Not really. First of all, you will get a huge profile with all that data. When you are using Roaming Profiles you might experience slow logon and logoff times. This is much better with Persona Management though. Secondly, and more important, aren’t you saving data twice now? One time in the Cloud and 1 time in your profile. Your company probably paid an amount of money to get that “Cloud based data storage environment”, like Dropbox/Box.net/Mozy Stash. In this case they also have to invest more money in disks for their file servers because Roaming Profiles are exploding in size and they are saved on central file servers. Folder redirection to speed up logon time when you use Roaming Profiles won’t change a thing. You still need to have central storage to redirect to.

So, this way, data will become expensive. How to solve these issues? Well, the easiest way is not to use an (offline) client. You won’t have the client issues inside a floating desktop and you won’t have to save data twice. Everything you do regarding your data is via your browser. Create a document and upload it directly to the Cloud. Realistically, this can be done but it’s difficult, especially when you edit documents a lot; downloading, edit, uploading again.

Maybe the offline clients can redirect data in a smart way; the folder does show content but it all is located in the Cloud. When you double click a file, then it will be downloaded to your machine, to a temp cache folder. Same happens to other way around. When you drag and drop a file from your VM into the data folder, it will sync to the Cloud and deleted from your VM.

I’m not a developer so I’m not sure if this is possible. What I do know is that Cloud based data solutions and VDI still need some work and thinking. I’m positive though. It will be great.

Cloud based data solutions- Are they forgetting about VDI?

If you follow VMware’s End User Computing strategy/vision, you probably have read that a computer consists of building blocks; Hardware, Operating System, Applications, Data/Settings. Traditionally these layers are tied together and it is very smart to untie them by using virtualization.

Nothing new so far. it is smart to decouple layers. You will be able to manage each piece separately. Users will be more flexible; Any time, any device, anywhere users will be able to get access to apps, settings and data.

So, let’s focus on data for a sec.

We know and understand you shouldn’t lock data in a PC. It’s a bad thing to do. You need to be behind that PC to be able to access that data. So, it’s common to redirect data to a file share inside the company. Now you can access it from any PC inside your company (Folder redirection and Roaming Profiles are mostly used together so that’s my assumption; Folder Redirection and Roaming Profiles/Persona Management are in place).

The limitation with Folder Redirection is that you can’t access it from any device, access it easily from outside your company nor share it easily. The solution for that; Dropbox, Box.net, Mozy Stash and Project Octopus, to name a few. It has to be mentioned; Mozy Stash and Project Octopus are beta/alpha products, not generally available. If you would like to know more about the differences and benefits between Stash/Octopus and Dropbox, click here.

What’s the beauty of the mentioned products? Your data is located in the “Cloud”, and “Cloud” can be on or off premise. Data is residing on a platform which is accessible. All you need to have are Clients on your devices, configured to access your account and your data. Go to the Dropbox or Mozy website and you will be able to get Clients for Mac, Windows, iPad, iPhone and Android (and maybe even other platforms). Great!! You can access your data from all of your devices. Exactly what we all want!

But…what about VDI environments…and let me stick to what I know best..VMware View environments? What about it?, you might ask. Just install, for example Dropbox for Windows and a user is good to go. I noticed a tiny little issue and I wanted to see if it is just me seeing this as an issue. I’m curious about your thoughts.

I do believe the most efficient and user centric way of computing is to use Linked Clone Floating pools with a delete/refresh after first use. All users will get a clean VM, the VM won’t grow much and will revert to its original size. Via ThinApp users will get their apps and Persona Management/Roaming Profiles will give the user’s look and feel. With traditional Folder Redirection, every time you logon to a new VM, you will be able to access your data from a share. I truly like this mechanism and encourage everyone to aim for this method.

So, no Folder Redirection but let’s use Dropbox in stead. I should install Dropbox in my golden image/my parent VM. Then deploy a pool from that parent and everyone should have Dropbox. Configure it and data should be available to that user. Your configuration should be saved in a part of your profile that roams, like AppData\Roaming. Persona Management/Roaming Profiles will save this setting and it will be there on every VM, so you only configure it 1 time. To go easy on the size of your profile, the Dropbox folder will be redirected to a share (default installation is user\My Documents\Dropbox).

Well, I really thought this would make sense but no! The case with Dropbox is that is gets installed in the user’s profile. It’s a per user installation (and to me, also very much per device, so per user-per device). Install it in your golden image and your users won’t even see Dropbox being installed.

I expect other products to have similar issues (I know some do). Either the installation is in someone’s profile or the configuration is device dependent (either saves under AppData\Local or under HKLM, result; not roaming). You don’t want to configure your Data repository every time you logon to a new VM, I assume.

I can access my data from an iPad and iPhone but I can’t access it from my main work environment, a VM. Maybe I’m missing something here and is something totally different coming, solving this issue completely or making it irrelevant. If it isn’t, then I truly hope “they” won’t forget VDI solutions.

VMware View- VMware View & Fusion-IO VDI appliance

On the 11th of February, VMware and Fusion-io in The Netherlands launched the “VMware View & Fusion-io VDI Appliance”.  Simplicity is the main word to describe this out of the box VDI infrastructure.

The appliance consists of a server with enough CPU and memory for 50 or 100 users, VMware View Bundle (vSphere included) and a Fushion-io storage card. A VMware partner could preconfigure this for end customers so it’s almost a matter of powering up the server and start using your new Virtual Desktop environment.

In my opinion, this appliance is introduced to say VDI isn’t difficult with this appliance. Many of us heard the scary stories about storage and VDI. Also, VDI is “difficult”. With this appliance customers shouldn’t be afraid anymore; Storage isn’t an issue anymore. The Fusion-io card delivers so many IO’s (and I have tested the cards and I was blown away!) that you don’t have to focus on VM’s and storage IO’s.

See attached the brochure about the VMware View & Fusion-io appliance.

VMware View & Fusion-OI 2011