VSI 7.0 – the VMAX release

Last week we got ESA 4.1 with some nice VMAX additions, today VSI 7.0 with even more – hence “the VMAX release”. Alright it’s not really named that but that’s what I’m calling it. Considering VSI 7.0 gets us to about 99% of the features the VSI thick client had for VMAX, this is the release we’ve been waiting for. For many VSI releases now I’ve attempted to bring forward the remaining thick client features into the web client but haven’t had much success. VSI has expanded, to be sure, both in terms of platform support and VMAX features, but there are only so many development resources to cover so many platforms. And in the noble effort to make things as homogenous as possible, invariably, in my opinion, we sometimes miss the mark. Granted I am biased being mostly a VMAX guy, but I’ve also been with the product since inception, too many years ago now to count, and have always had a vested interest in delivering the best for my customer. It took some convincing to offer features that are solely unique to VMAX but they agreed. Alright enough about sausage making, the release is here so on with it already.

Let’s start with a “me too” feature concerning NMP (PP/VE readers feel free to move to the next feature). Currently XtremIO offers the ability to change both the preferred path policy to Round Robin and the IOPS parameter (how often a path switches) to 1. On the VMAX side by default we use Round Robin so that wasn’t a concern for us, but we have been recommending IOPS to 1 for a long time. The IOPS parameter can be changed in a couple ways.  First, you can do it on a per-device basis with a simple command – presuming you know the NAA of the device which you can pull from the vSphere client. The second way involves adding a claim rule and then rebooting the box so that the existing presented devices pick up the new setting. We offer both these now through the VSI interface. It allows you to change current device IOPS and then set the rule for future devices.

Because we share the interface with XtremIO, you will need to select the correct radio button.  I’ll list the steps and follow it with a screenshot.

  1. Right-click on the ESXi host and select All EMC VSI Plugin Actions –> ESXi Host Settings
  2. Choose the VMAX3/VMAX All Flash radio button. If you toggle between the buttons you’ll notice the host options change. Note the special instructions at the top concerning ESXi hosts which share connections to both an XtremIO and VMAX3/AFA array. Because the architectures of these two arrays are distinct (yes even for VMAX All Flash), there are different settings required.
  3. Check the boxes for the settings you wish to execute.  Either one will require root access.
  4. Provide the root credentials.
  5. Complete the wizard


Click to enlarge – use browser back button to return to post

VSI will now go out and execute the changes and if there are no issues you will simply return to the vSphere Web Client.  Come to think of it, if there are problems you’ll also return to the client. Yeah so that’s not good but it is the truth. There is no error handling here if something goes wrong. Fear not, you can’t break anything. Nothing bad will come of VSI attempting and failing at this task but you won’t get the changes you need. OK so how do I confirm there was no issue and what could go wrong? Only two things can fail – you supply the wrong password and nothing gets done, or adding a claim rule fails because of a duplicate as ESXi will not let you add a rule if one already exists. The place to look for failures is in the VSI log. Here is an example of a credential failure. I boxed the task in blue, and then the failure in red.


Click to enlarge – use browser back button to return to post

I know, not so ideal. I’ve asked them to push the error to the user or at the very least make a test connection to the host after entering the credentials in the wizard. It wouldn’t help the case of a duplicate rule but that is far less common than incorrect root credentials.

Another long awaited feature that is in VSI 7.0 is support for multiple vCenters. This has been a particular annoyance for customer running Embedded Link Mode in vSphere 6 where you share an SSO (PSC) and thus can see any vCenter no matter which one you connect to. Just follow the rules for setup in the Product Guide – i.e. register each vCenter – and you are good.

Now moving on to the main event, the biggest feature lacking in the VSI Web Client has been the ability to see detailed information about all the datastores in a single screen or all devices/LUNs. In the thick client there was a “Datastores” view which included all datastores presented to the host. You could select any datastore and a bottom panel would give you important information about the VMAX device underlying the datastore – storage group, device ID, code version, etc. Similarly there was a LUN view which provided all the information in a single panel, and included a column telling you if the device was used as a datastore, RDM (which VM owned it), or wasn’t used. This was by far the feature customers most liked about the thick client. VSI 7.0 now offers the same capability.

There are two new views available in the monitor tab at the host OR cluster level (cluster is new to the web client):


Click to enlarge – use browser back button to return to post

These new views provide almost the same information as the thick client did (e.g. no thin pool information as no thin pools). We’ve also pulled all VMAX devices from the previous view – EMC Datastore Viewer – so as to avoid confusion. If you have no VMAX devices, the new tabs will be empty.  Similarly if you have only VMAX devices, the EMC Datastore Viewer will be empty. For this feature I went with a demo which is below. One useful feature I point out is the ability to figure out which datastores would benefit from UNMAP because we tell you how much data is used on the array versus what VMware thinks is used.

Use 720p if possible and full screen

Some caveats with the viewers:

  • Technically there is no support for SRDF/Metro devices. Now I did not have any trouble when I tested but just in case you do, it isn’t officially supported. Again, there is nothing to break here rather it would mean perhaps no device detail would show.
  • And the second technicality – VSI 7.0 does not support the latest 5977 2016 Q3 SR. We are currently validating it and hopefully will have support before VSI 7.1, but in my setup I used SMI-S 8.2 with the newer coded arrays and did not have issues. Again that is not a support statement, just my personal lab experience.
  • We cache information for 20 minutes for performance. So changes to a device (e.g. capacity utilization) will not show up immediately.
  • The cluster view will show duplicates since it lists all host devices. There is no filtering.
  • The columns in the viewers cannot be resized. We wrap the information so nothing gets cutoff, but I filed an enhancement to enable this as it should behave like an excel spreadsheet.
  • If you have VMAX devices presented, but you did not register the array, you will only get some basic information in the viewers (I showed an example in the demo). You can click on the picture below to see what it looks like. To resolve, add the array in VSI.


Click to enlarge – use browser back button to return to post

I hope those VMAX users out there who have given up on VSI will give it another go with this release. There are still improvements to be made – like performance when loading the menus – but all in all this is greatly improved and as always is free.


One thought on “VSI 7.0 – the VMAX release

Add yours

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Website Powered by WordPress.com.

Up ↑

%d bloggers like this: