Unsolved
This post is more than 5 years old
53 Posts
1
3302
Unisphere for VMAX Improvement Requests
We have been using Unisphere for VMAX for nine months now (current version is 1.5.0.6) and we have some serious issues with it. Currently it does not meet our needs. We have started using the command line more and more, however, we would like to use the GUI, especially for our less experience team members because it is provides more protections against mistakes. Unfortunately until some things are fixed we cannot use it.
I'm starting a discussion for everyone who uses Unsiphere for VMAX to provide feedback on improvements they would like to see in future releases of the product. I'm start it off wish my current list list:
- Improve the speed of jobs. This is our number one complaint. As our environment grows the performance drops. It now takes 5-10 minutes for the simplest of jobs. Waiting 10 minutes to create a volume is unacceptable. Waiting 5 minutes to add a create volume job to the job list is unacceptable. The GUI is locked during this time, you can't move on and do something else while the job completes in the background (unless you added it to the job list, but then you get less feedback). This has driven us to the command line.
- Provide some form of feedback on job status. The command line provides output every 30 seconds showing step X of Y. This is helpful because it shows the job is still running, it hasn't hung, it hasn't silently failed, and it gives you a feel for how long until it completes. Include something like this in the GUI. Everyone uses and understands progresses bars, please include one in the next release.
- Improve failed job response. To often a job will fail because a prerequisite hasn't been met, such as trying to delete a device that is part of a meta, but the GUI will take up to 10 minutes to report the failure. The command line detects this right away (if you use the preview option). The GUI needs to do a better job of quickly detecting and report failures in jobs.
- Provide a way to annotate alerts. Currently you can acknowledge and/or delete alerts in Unisphere for VMAX. It would be nice to be able to acknowledge the alert, add a note with a SR# or what was done to fix the issue. Also, instead of deleting the alert it would nice to be able to close the alert so it not in the active list but not gone.
- Improve filtering on alerts. When creating a new meta I don't want to see dozens of alerts saying all the devices that make up the meta are offline, then dozens more saying they are on-line. These messages should not be alerts, they should be audit log entries.
If you have a solution any of these issues please pass it along. If you have other features or improvements you would like to see add them to the list.
boydbria
54 Posts
0
July 29th, 2013 09:00
Yes to all the above! Especially the speed thing...
In addition, some usability and consistency concerns:
1. We can right click on some things and a menu comes up, but for other options you need to use the [>>] button to get a list of options. Please do a usability study...consistency is off throughout Unisphere for VMAX.
2. Meta volume creation - We absolutely love the idea of using 'duplicate volume' for metas we've already created but why not have this as a default and have some sort of 'best practice' built in? 240GB volumes are a thing of the past, and I know auto-meta is there but it doesn't always follow best practice. I think a separate "Create Meta volume" would be awesome with member defaults of 2/4/8/16/32 members.... So for example if I create an 8 member 1TB volume it will automatically calculate the volumes it needs behind the scenes rather than me creating them first and then forming the meta.
3. Lun migrations/ Meta volume expanding - Please look at your VNX brethren to have a way to migrate from say a 400GB meta lun to a 600GB lun without using the CLI. The expanding a striped meta using BCV, etc and re-striping back on the fly is awesome but why not have something built into the GUI. VNX does this VERY well... Tier 0/1 VMAX storage should also have this built in.
Hope this helps!
Brian
rawstorage
419 Posts
0
July 31st, 2013 00:00
Brian, in relation to point 2 that you raise is the use of Templates for meta-volume creation, and there have been some improvements in latest version of SE/Unisphere, you can now override the autometa feature to create volumes in a certain way. If you click the advanced TAB you can specify the number of volumes and size of meta members. e.g create a TB volume using 16 x 128GB metas.
here is the update syntax that unisphere relies on from the command line
symconfigure -sid xxx-cmd "create dev count=2 size = 2048 GB , emulation=FBA, config=TDEV meta_member_size = 128 GB;" preview
You can also set up templates for your frequently used device sizes, this would be the most consistent way of doing things.
McCranium
5 Posts
0
August 1st, 2013 08:00
I totally agree with the consistency and useability comments.
AD Integration: Should support nested AD groups when defining which AD group has access to Unisphere. Currently you will need to define multiple groups separated by a comma (and no spaces). Additionally changes made to the Authentication settings are detected. To add groups that are allowed to access Unisphere currently (version 1.6.0.8) you need to disable AD integration and re-enable it for Unisphere to allow you to modify the field. Again, some QA is needed to ensure the logic in the application is working correctly.
SMTP Alerting: Exceptions should be trapped approrpiately. In 1.6.0.8 if you enter an incorrect hostname a java exception is thrown. How about some QA?
ajbarth
14 Posts
0
August 5th, 2013 13:00
More on usability. Unisphere does a good job of navigation via the breadcrumbs, but not all target objects become the new source of an activity. Example: If you have an Initiator Group, you can find a list of initiators in it. But say you want to add/change the initiator alias on one of those. We're already at the perfect spot to right click on the initiator which should present you with the options as found on the bottom of the Host>Initiators page. Instead you must got to the Hosts>Initiators page and scroll through the list to locate the initiator you want in order to Rename Alias. This brings me to the 2nd thing I miss: lack of find/search/filtering on panels. Let's say you want to find out things about volume 10DF. You have to already know what type of volume it is before you can find out more info about it. I think a find/search/filter comes in handy here and anytime one needs to wander through a large list of entities.
Anonymous
5 Practitioner
5 Practitioner
•
274.2K Posts
0
August 7th, 2013 04:00
So just returning back to the original comment, as speed of jobs was a significant part of 1.6 release
1. So in Unisphere for VMAX 1.6 we changed internally how we add/update a job to the overall jobs list. This was done specifically to address environments where the list of jobs grows over time.
We made significant improvements internally to address the performance of the jobs list view. The work we did in this area should address the overall performance of the view itself , and reduce any overhead from UniVMAX in running the job.
2. Again we poll the status of the job frequently, but in the previous version you would have been slowed down by the fact that we were checking all the jobs on the list. Now our check is more granular so again in this case the improvements with 1.6 will help the updating of the view.
3. Here also the 10mins time will be drastically reduced in 1.6 environment, if not I would ask that you contact CS and OPT as we believe we have addressed this issue in 1.6.
4. & 5 I will discuss with product management these ideas will help to improve the product. (If you can enhancement requests are a good way to track those issues. )Thank you very much for the feedback it is much appreciated.