It was my recent good fortune to get the loan of two Dell EqualLogic systems from Dell. They're mainly for use in my work in VMware SRM. But along side that it gives me a chance to look at the integration with VMware that Dell have been working for sometime. This follows a recent theme of mine covering the various plug-ins from the storage vendors including EMC’s VSI and NetApp VSC.
I’ve made a decision that where a storage vendor has these kind of datastore provisioning tools – is to cover them along side the standard method of creating a new volume/LUN in their fully-blown management systems. I think in short-term we might see these plug-ins evolve to allow even higher management functions such as actually configuring replication for storage with vCenter. Imagine that – you right-click a datastore in vCenter, you see a Dell/EMC/NetApp menu, and underneath that you see link that says “Configure DataStore for use with VMware SRM”
For the moment the Dell HIT-VE does allow you to setup replication of a volume once the appropriate “partner” relationships have been configured between two or more Dell EqualLogic Groups. So I will be including that in chapter on Dell in my SRM book.
I was really quite taken with how easy it was to setup the Dell EqualLogic – it took very little time, and found the management tools easy to follow. That’s always helpful when you’re taking on a new technology. Once set-up I had a chance to play with the Dell EqualLogic Host Integration Tools for VMware (HIT/VE) which includes the
- EqualLogic Datastore Manager – Allows administrators to provision new Datastores, expand existing Datastores, and monitor storage usage in the VMware Datacenter
- EqualLogic Virtual Desktop Deployment Tool – VMware View aware desktop provisioning wizard which allows the deployment of manual virtual desktop pools using array based space efficient thin clones and also allows the refresh of these pools.
- Auto-Snapshot Manager/VMware Edition 3.0 – Creates and restores Smart Copies of Hypervisor consistent Virtual Machines (VMs). Utilizes snapshots, clones, and replicas within PS Series group allowing for fast restoration or disaster recovery
Dell have a different approach to EMC/NetApp with their plug-ins. Firstly, there’s a virtual appliance that you import to your vCenter. The VA has two NICs – one to communicate to your VMware virtualization layer so it can speak to vCenter – and another NIC to speak to the EqualLogic Array(s). Once the appliance has been imported and powered on you can login using the default username/password of root/eql. This in turns executes a script that builds a menu system to carry out the post-configuration:
It’s a pretty simple affair of pressing the number on the keyboard – and providing the appropriate details. Step 2 and 3 are where you configure your IP settings on the VA, and Step 4 is where you configure the vCenter.
Step 5 is where you begin to make the VA aware of your Dell EqualLogic Array’s by supplying the “Virtual Group” IP address. In case you don’t know, the main management unit in Dell EqualLogic is the Group that contains many “members” the term Dell use for the actual array itself. This IP is the main one used when you connect remotely to configure your storage. In my case this is 18.104.22.168, you’ll also need a user account, the default being “grpadmin”, and its password – much depends how you have delegated.
As with the other storage plug-ins I’ve been recently covering, the Dell HIT-VE does support cloning of virtual desktops for use with VMware View manager – and as such to create desktop pools – you will need to provide the name of one of your Connection Servers, and a user account and password.
Notice here I’m using account I setup in View called “View-Admin”. I discovered that the View does not allow the use of the named “administrator” account with plugs-in of this type, although Dell HIT-VE will accept the “administrator” account for vCenter. At this stage there’s no mandatory requirement to run step 9. This allows you manage the “database” behind the Dell HIT-VE, and you can use this to backup, restore, and upgrade the HIT. Of course after the reboot it would be good practice to change the password for the root account.
You want to keep your eye on Step 10 if you making any major configuration changes, as you might need to reboot the VA to allow changes to take affect. For example, update the HIT Appliance to speak to a different vCenter – and I was bit confused why the icons weren’t appearing – sure enough there was a pending reboot needed of the HIT which I hadn’t noticed – and of course, I had to re-run step7 to make sure the plug-ins were registered with the new vCenter. So, you know I’m working with multiple versions of vSphere – and I’m spending some of my time flipping my lab from one version to another – and that’s what caused that reconfiguration to take place.
Once you login into vSphere Client you will find the UI will be extended with shortcuts to the Dell EqualLogic system under the “Solutions and Applications” section of the “Home” location in vCenter:
If you do click at any of the EqualLogic icons, you will be asked to input your vCenter username and password – before the pages open. This authentication dialog box opens for the very first connection, but after that it doesn’t appear again.
From any of these icons (Datastore Manager, Auto-Snapshot Manager and Virtual Desktop Utility) – you can get to a Configure HIT/VE link which allows you to change some of the settings configured on the virtual appliance – and from the Virtual Desktop Utility you can also change the settings for the View Connection Server also:
Provisioning New Datastores
As you would expect the main “Host & Clusters” and “Virtual Machine & Templates” view is updated with menu options specifically associated with Dell EqualLogic:
Selecting this option cranks up a web-page and is very similar to the New Volume wizard in EqualLogic Group Manager. It allows you to set the number of volumes, their size, and if they will be thinly provisioned – notice the ability to set the block size at the bottom of the page – notice how I’ve selected the cluster as the scope here – so every ESX host in the cluster will be granted access to the storage – you do, of course, need the iSCSI stack on the ESX hosts configured for that to work smoothly!
Next, you can set the controls on the volume – to control which ESX hosts has access to the volume. This case the Dell HIT-VE will examine the iSCSI Stack/vSwitch configuration on each of the hosts, and automatically ensure that the volume is created with “Allow simultaneous connections from initiators with different IQN names” – and populate the “Access” tab on the volume with the correct rights for the ESX hosts to connect to the volume.
After this point you're able to label the datastore, and place it in a datastore folder if you have them – in my case I keep my local datastores and infrastructure stores in separate folders – this way I can use permissions to control who can see what and where in vCenter.
Once you click next and finish – this starts off “Job” windows which breaks down the processes going on in the background. I quite like this feature – it kind of reminds me of the “Jobs” feature of Microsoft Hyper-V/SCVMM. I like to know what's going on in detail, in case anything goes haywire it gives me something to go on.
Note: You want to bring up this job window at any time – it can be brought up by clicking the “circle” icon in the top right-hand corner in any of the “Manager” icons.
If you're looking for more detail on the volumes created with the Dell HIT-VE, you can use the “Equallogic DataStore Manager” icon under “Solutions & Applications”. From here you can carry out other management tasks like doing a Smart Copy, Create a clone and Resize a datastore. I particularly like the “Resize a datastore” option, as it not only makes the iSCSI LUN larger, but automatically does a “Grow VMFS volume” task to boot!
If you need it there’s a link to the Dell EqualLogic Group Manager. Notice how other volumes originally created in the Group Manager application appear as well, so long as they have been correctly allocated to the ESX hosts in the cluster. Volumes that the ESX hosts do not have access will not appear.
Provisioning new virtual desktops for VMware View
There’s a couple of places where you can carry out a Smart Clone. You can create multiple copies of a VM from the right-click on context menu. This works for any VM whether its a virtual desktop or ordinary VM. This could be useful if for demo purposes you need to fire-up a number of VMs very quickly.
If on the other hand your purpose is to create new virtual desktops for VMware View, I think you would be better placed using the “EqualLogic Virtual Desktop Deployment Utility” in the “Solutions & Applications” area. I was quite struck by how much control Dell are offering in this plug-in. Most storage vendors plug-in handle the main job of creating a desktop pool, selecting the pool type (Dedicated or Floating) and of course the all import array-accelerated cloning process. Interestingly, the Dell HIT-VE allows you to modify the default settings on the pools it creates such as whether you allow users to reset their desktops, or whether you prefer the PCoIP Protocol over RDP. These controls reside in the “Modify Default Settings” dialog:
You crank-up the virtual desktop cloning process by clicking the “Create desktop pool” link, which will in turn bring up wizard to drive you through the steps.
Step 1: Desktop Pool Settings
Step 2: Select your Virtual Desktop Template
Note: Allow the UI says the VM image size is 32GB – it is in fact a thin-virtual disk.
Step 3: Deployment Unit Layout
The wizard will also allow you provision new storage – that Dell refer to as a “Deployment Unit” – each deployment unit represents an ISCSI LUN formatted with VMFS containing up to a maximum of 64 VMs. You have the choice of using either a full clone method or linked clone approach. I think it's a no brainer to use Dell’s “linked clones” it will be quicker to deploy large numbers of VMs and save disk space at the same time. This part of the wizard will also help do some capacity planning as well. You see virtual desktop is more than its virtual disk; it does consume disk space including the all-important VMKernel swap file. So a 4GB Windows 7 VM without reservations would consume 4GB worth of virtual disk space. The deployment wizard allows you to set a percentage based “reserve” to take into account the additional disk space required to complete the job in hand.
Of course it’s possible to have multiple deployment units. So you could create 4 each with the full allocation of 64 – that would create a total of 256 virtual desktops.
Note: By default the wizard uses “Full Clone”, and that can trigger an alarm because of the size of the data store. Don’t panic! Just switch to “linked clones”. Remember this will allow you to provision more VMs than you have storage – effectively allowing for similar over-commitment that VMware does for memory on ESX. Here I’m allowing the deployment unit the possibility to have up to 64 VMs in it, in the next part of the wizard I will be asking it just to give me 10 desktops in the first instance.
Step 4: Capacity Planning
VMware doesn’t give the storage vendors the hooks or the APIs to know when a new connection is established in order for them to spin up additional VMs. As we saw with the NetApp VSC and the EMC VSI – all the vendors are currently forced to create “manual pools”. With the Dell HIT-VE the administrator can deploy in “bricks” of say 50 virtual desktops, deploy ahead of time maximum of 500 with 10 volumes but only register 200 to begin with. The benefit to the other part is the other 800 are not taking up any extra space – and when they need to grow it registers the VMs in a few minutes instead of copying and creating more. I think this is an interesting approach to the limitations in the API’s currently, I think ideally the storage vendors would benefit from having more intelligence back from VMware View – so the storage vendors can respond to the changing load and demands of the users.
Step 5: iSCSI Access
This windows is exactly the same one we saw earlier in the deployment wizard. This wizard handles the ACL to the iSCSI LUN, populating the “Access” tab of Group Manager with the IQN’s of the ESX hosts in the cluster. So its important to make sure your VMware Software ISCSI is setup and correctly configured!
Step 6/7: Virtual Machine Folders & Resource Pool Placement
The next part of the wizard controls where the virtual desktops will be created in the VMs & Templates Folders view in the vCenter Inventory, as well as its Resource Pool location in the Host & Clusters view. You can also set how you want the desktop to be named and numbered. As with VMware View the Dell plug-in will create a VM Folder, but it won’t create a VM Resource Pool.
Step 8: Desktop Settings
Finally, you are able to select the VMware “Guest Customization” that you use for adding virtual desktops to your domain, additionally you have the option to modify the settings of the desktop that we saw earlier.
At the end of this process you're left with a pool in View Manager, all you now have to do is handle the entitlements (the permissions) that allow a user or, more normally, a group to gain access.
Using Smart Copy on Dell EqualLogic iSCSI Volume
The third and final big feature of the Dell HIT-VE is the Smart Copy feature. In case you don’t know Smart Copy is a snapshot feature that allows to take “point-in-time” copies of either a VM or a datastore – and then revert back the VM or datastore – if something really horrible and unpleasant happens to you. If you have been following my series on these storage vendor plug-ins it’s along the same lines as the NetApp VSC “Backup and Recovery” feature.
As with the other functionality I’ve been exploring – the plug-in allows for a right-click context sensitive link to the feature, or you can access it from the EqualLogic Auto-Snapshot Manager. There’s a scheduling process where you can ask for a snapshot to be taken at regular intervals as well.
Once you select the “Smart Copy” option a one-page wizard will open up allowing to the administrator to trigger the snapshot process:
The “Include PS Series volumes” option is used to also snapshot any EqualLogic iSCSI volumes that are mounted directly to the guest OS of the virtual machine in format of a VMware RDM. The “Perform VM memory dump” will create a VMware snapshot of the VM to also save the memory state – this enables the ability at restore time to bring the VM back to the state it was running in when the snapshot was taken. The MS VSS or VMware sync drive flush is always performed as the VMs are always put in VMware snapshot mode prior to Dell performing a snap of the volume on the array. This is not an option you want to use too frequently, as if there are many VMs with a decent amount of RAM it can take quite a while to write all that to disk. The “Set created PS Series snapshots online” does exactly as it says on the tin, once the snapshot is created it is put on line and attached to the VMware environment, this does provide a means of quickly cloning a datastore of VM, you’d still need to manually register the VMs, but that would be quicker than the normal cloning process, and this might be useful in some test or dev use cases.
The Error handling section - are related to also having non-EqualLogic storage in your environment, and how you’d like the tool to behave should you attempt to create a smart copy which consists of a mix of EQL and non-EQL storage, do you stop, or do you just ignore the non-EQL storage. For me this is quite an important option because my VMs are scattered all over different vendors arrays. One thing I’ve noticed with these plug-ins is the various vendors mainly assume your are JUST using there storage. As consequence if they come across another vendors storage you sometimes get strange errors. It’s nice to see Dell making some moves to reduce this – after all a VM stored by any array vendor – could be on local storage…
This then will show the “Job” process where by the Dell HIT-VE engages a VMware Snapshot, and then creates a smart copy. In my case I just had one VM on the volume and this took very little time indeed:
This quick Smart Copy is then listed in the Auto-Snapshot Manager together with the date and time that it was taken.
That’s great of course – but what’s really needed is a schedule from which this happens at frequency that allows me to roll back the volume to a previous known good state, or alternatively present the snapshot to the ESX hosts – so it can be mounted. That way I could simply copy back a VM that might have become deleted or better still mount the virtual disks of the snapshot to the original VM – and use standard Windows or Linux tools to copy back lost data.
Again this can be set-up directly from the vSphere Client on the right-click of any datastore managed by the EqualLogic array.
The first part of the wizard gets you to set a schedule name, and basic frequency (once, daily, weekly or re-use an existing schedule), and the second part of the wizard then adapts to allow more granular control. The final part of the wizard allows you to set a retention policy for how long these Smart Copies are allowed to exist for. Clearly, if I’m snapping every 5mins – but only retain 10 snaps – I’m only going to be revert back to the last 50mins worth of changes – but remember this is just a demo – you will need to adjust schedule according to your needs.
In my case I set a very frequent schedule because I didn’t want to wait around too long for the snapshots to be taken. Remember these snapshots are just the “delta” changes before the snapshot was taken. Depending on the number of updates, the volume and size of these deltas will vary, and care must be taken that you have enough storage to hold them.
These schedules can be modified and managed – from the Auto-Snapshot manager window – so if you need to modify their frequency, temporarily disable them or delete them – you can do.
1. Modify snapshot properties to make to “Allow simultaneous connections with from initiators with different IQN Names” – This allows more than one ESX hosts to access the volume, and the VMFS partition within it
2. Set Access Type to make the iSCSI volume read/write – by default the snapshot is read-only.
3. Set snapshot online – so it is accessible to the hosts
4. Rescan an ESX host
5. Run the resigniture wizard on the volume
6. Rescan all the ESX hosts
7. Copy the VM from the snapshot to the original sources.
That’s quite a lot of work isn’t it – fortunately the Dell HIT-VE automates all of that process for you – in a way delivering a self-service restore process directly from the array, initiated directly from the vSphere Client.
So for example I had a lost or damaged VM and I wanted to restore in it complete. To do that I would go to the Auto-Snapshot manager icon in “Solutions & Applications”, and by selecting the Snapshot, I could see the VMs contained within it, and then I could right click and select “Restore from Smart Copy”
This would crank up a wizard that would allow me to restore a VM like so:
In my case the focus of my Smart Copy was the datastore – but it can be a folder of VMs. The cute thing about this – is as new VMs are added to the folder, they are automatically enrolled into the Smart Copy schedule without any additional work.
This is my last article in the series of looking at the storage vendors plug-in for vSphere. Things have certainly come along greatly in the last 18 months or so. I think they are a really great way of managing your storage as it relates to your virtualization layer. whilst these tools are excellent – in a corporate space where those roles represent different fiefdoms within Corporate IT – will the Storage Admins ever allow you to play with these plug-ins? Food for thought indeed.