Veeam ONE vNEXT feedback/wishlist

[Vitaliy S.]

Ok, thanks, we will see what we can do.

[HannesK]

I also got an request yesterday for a syslog server in Veeam ONE to correlate all messages in the data center (storage, SAN switches, etc.) but I'm not sure whether this is the focus / scope of Veeam ONE

[Shestakov]

Hello Hannes, what`s the goal of having the syslog server? What was the use case?
Thanks!

[HannesK]

the use case is to have one central monitoring and reporting tool but as far as I know, this is not our goal.

[Vitaliy S.]

This FR is not a priority for 9.5, as all the features have already been defined, but we can consider it for our next releases.

[EugeneK]

Hi guys,

Not sure if that was requested before (if there's a dashboard of already requested features I can check, please point me to it), but having VeeamOne being a standalone instance with no clustering support is one of the things that'd allow the organic growth of the product. Since collector is rather a heavy process while running, being able to distribute it across multiple instances of VeeamOne and still have a centralized console to have a holistic view over the environment would be a great addition (much like the console for B&R).
At the same time, users will not be bound to shovel all monitored environments into a single instance, allowing for smaller failure domain, should an instance fail.

[Shestakov]

Hi Eugene,
There is no such a report, but you can use key words "Feature request" to get most of the requests.
I think the thing you are asking to is pretty similar to one in this discussion.
You can one Monitor Client to work with several servers. The main issue here is databases location and how fast it can be transmitted to the client.

[Vitaliy S.]

Hi Eugene,

Thanks for the feature request, much appreciated. Do you see any issues in terms of performance/load on the server when only one "collector"/ONE server is used?

Thanks!

[EdwardsCP]

I don't thinkthere is any existing functionality to accomplish this, so I'm posting it to the wishlist. Please let me know if there's already a way to do this.

It would be really great if VeeamOne was able to raise alarms for VM hardware utilization anomalies. Today, you can define alarms for when hardware utilization (for VM cpu, memory, etc) are above a certain threshold, which is good for drawing attention to some types of problems. But the types of anomalies I'm referring to are something more along the lines of "This VM typically operates around 5% CPU usage, and for the last 2 hours it has been running at 15% - that's strange and should generate an alarm". The key here would be that VeeamOne could gather data about VMs over time and "learn" what is normal for any given VM, and then generate an alarm when something is causing the VM to use significantly more resources (but not so many resources that the existing "threshold" type of alarms would have gone off anyways). The idea is that resource utilization can be indicative of all kinds of things (both good and bad, but significant unexpected changes are often bad), and an Admins/Engineers responsible for those systems should have tools that help draw their attention to utilization anomalies so that they can investigate the cause and determine if they need to take action.

Obviously, the devil is in the details for detecting anomalies like this, and I can imagine that it would be complicated to tweak the code to get it working really well without too many false positives (where there has not been a significant change but an alarm was raised) or true negatives(where there is a significant change in the operating environment but no alarm was raised).

I'd love to know if Veeam is considering anything like this.

[Vitaliy S.]

Good feedback! I agree that tracking anomalies would be beneficial, but for this you need to at least stay with static thresholds for 3-6 months to collect enough data for analysis. We will discuss this feature for the next major version.

[Shestakov]

Hello Colin,
The request is interesting, but it`s hard to imagine the logic without the mentioned "false positives".
If for instance we create an alarm rule, say "trigger if last 15 minutes utilization is 30% higher than average of last week" we will have false triggers during business hours since the utilization may be higher. Same for holiday weeks etc. when it gives you low weekly average to compare with. There are also vast amount of businesses with peculiar properties and demands.
I would start from your use case, could you explain why is utilization thresholds are not sufficient?
Thanks

[EdwardsCP]

The request is interesting, but it`s hard to imagine the logic without the mentioned "false positives".

You're right - Eliminating excessive false-positives would be difficult. But the admins responsible for the environment could chose to turn this type of monitoring off, or they can choose to have someone investigate each alarm. Think of it a little bit like the way we use Intrusion Detection Systems for network security - the IDS provides you with an alert based on traffic matching some pattern/rule, and then usually an admin/analyst needs to review the traffic and determine the nature (good or bad). False Positives in the IDS world are basically inevitable. Not every alarm will indicate something bad, but drawing the attention of an admin/analyst can help empower them to look into what's really happening.

I would start from your use case, could you explain why is utilization thresholds are not sufficient?

Mostly because most (almost all?) admins aren't going to take the time to analyze all of their VMs and set thresholds that would detect these types of anomalies. They're going to leave the warning/error alarms at the defaults or thresholds that are relatively high (i.e. 75%/90%), and that's probably a really good idea because it's an appropriate type of alarm for drawing attention to a system that has high utilization and the services it provides on top of the OS (app/db/etc) might have a memory leak, or run-away process caught in a loop that's using more and more CPU cycles as time goes on, etc.

The type of anomaly detection I'm describing would alert you to something like this....

If you look at the usage displayed on that graph, it's pretty clear that on 6/22, for some number of hours (maybe 5 or 6 total), that VM was much more busy than it usually is. Yes, CPU usage was only at 4%, but it's usually between 0 and 2. Clearly this server has very little utilization all of the time, and even during those 5-6 hours it was still very low (in terms of total resources available), but something abnormal was happening during that time. If you could imagine that the same VM had a very similar utilization of 0-2% for X number of months prior to the higher utilization on 6/22, then it's even more obvious that something abnormal is occurring. And I can't imagine that any systems admin would set an alarm threshold as low as 4%, even though it may be appropriate in a case like this when that utilization occurs over an extended amount of time (a plateau vs. a peak on the graph).

Hopefully that makes sense.

[EdwardsCP]

Good feedback! I agree that tracking anomalies would be beneficial, but for this you need to at least stay with static thresholds for 3-6 months to collect enough data for analysis. We will discuss this feature for the next major version.

I'm no statistician, and I completely agree that the larger the sample, the more accurately you can start to determine what's normal, but I think in a lot of situations you could trigger some accurate alerts with even a small sample (say, 2 weeks of data). I imagine that you would probably just have more false-positives until you had a more significant sample size.

Customers/Users would need to understand that these types of anomaly alarms wouldn't always be indicative of a problem (even with a 6 month sample size). The alarm would just draw their attention to something that seems to be different.

[Shestakov]

Thanks for the comprehensive reply, Colin!
My point was that the feature may cause lots of unexpected alarms for those who are not aware of it. However it definitely makes sense and we will consider the feature in future product releases.
If you have more details of how the feature should work, that`s the best time and place to share it with us

[BastianP]

It will be great to have an ability to specify the Period units of measure (days, weeks, months, years) and a variable (like N) --> report period = past two month, past two weeks ...

[Vitaliy S.]

Hi Bastian, where exactly would you like to have these settings? Reports, dashboards, Monitor views?

[Shestakov]

If we are talking about reports, please specify exact reports you want this to be implemented in.
Thanks!

[aka]

Hello,
i have updated our veeam one from Version 8 to 9. I think that the overview in "veeam one Bussiness view" was in version 8 better than in version 9 (list of details).
@Veeam Product Management: please see here Case #01903427, perhaps the old few can be releasd with the next patch...

Moreover, it will be nice to have in Veeam Reporter a new Report based on "VMs Configurations" but not sortet on Cluster/ESXi-Hosts rather on Datastore-Cluster/Datastore -> see also Case #01903427

Thanks.

[Vitaliy S.]

Hello,

There are no plans to release a patch for v9. Next release will be 9.5 and it is already in a feature-lock state. Can you please tell me what you're trying to achieve with these BV dashboards?

Thank you!

[aka]

Hello,

ok, is it possible to add this reports in a patch for version 9.5?

Veeam ONE Business View: The Detail-View from the single charts are our monthly report for our management. This was a pretty/fast way for a simple overview for our
- utilization from our datastores
- operating systems
- SLA for planing 24/7 support
- last backup date
The management have an web-access to the business view and have also createt dashboards by itself. so for example for billing of some big customer.
I have try to rebuild this reports with the Veeam One Reporter v9, but no standard report does incloud this simple output how the business view in v8.

The only report i have found to replace this look where the "VM Configurations" in Reporter V9, but there is no way to edit this report (in v8 and v9)- so the second FRQ.
Thanks.

[Shestakov]

Hello,
If I get the request correctly, the report you want to get should contain datastores free space, last backup date of VMs and their OS, all based on SLA category?
If so I would use such reports as Protected VMs, VMs Configuration and Datastore configuration all based on Business View objects, and combine desired parts of the reports using Report Builder.
Thanks!

[Vitaliy S.]

ok, is it possible to add this reports in a patch for version 9.5?

If you're referring to VM Configuration changes, then we will need to estimate the amount of work and available resources. As to the another request above, then Nikita's suggestion should do the trick? You can create this report and schedule it to be delivered on daily basis to your management.

The management have an web-access to the business view and have also created dashboards by itself. so for example for billing of some big customer.

Created a dashboard in Veeam ONE Reporter or Business View? If they used Reporter, then, out of curiosity, which widgets did they use?

[caplaser]

Hello,

Following case ID# 01953540 , the support tells me to post a comment here.

We need to do a report for a customer who needs to know status of these backup and replication jobs.

Unfortunately, there's no predefined report as we wish in the current version of Veeam ONE

The report closer than that i need is "Last Backup Job Status" but I can't run this report only for selected jobs for this customer.
I don't want to send all jobs reports, so I can't do that I need.

I think this can a good way to permit jobs selection like this can be done on "Backup Job Historical Information" report.

Best regards
Laurent

[Shestakov]

Hello Laurent,
If we add an option to choose particular jobs to Latest Backup Job Status report, will it be sufficient for you?
Thanks!

[Vitaliy S.]

Yes, the suggestion (adding a backup job picker) makes sense. We will ask our dev team to add it in the next updates.

[tvandael]

Hi all,

I'm missing the virtual disk capacity usage in the global overview.
When can have a report of it but an extra graph showing the disk space usage of a vm over a certain time (similar like cpu, memory, ...) would make sense to me.
This is an important graph for me and for this I'm currently using a different product to get this.

Anyone else shares this opinion?

Thnx!

Tim

[Vitaliy S.]

Hi Tim,

Do you want to see how VM disk size (thin or thick) is growing over the certain period or you would like to see Guest OS disk usage?

Thanks!

[Shestakov]

Hi Tim and welcome to the community!
Could you also describe your use case, please? Have you checked "VM Guest Disk Space Usage" report by the way?
Thanks!

[tvandael]

Hi,

our use case is that with over 500 virtual machines to manage, diskspace can be hard to manage.
It happens quite often that we need the progress in diskspace usage to identify an issue.
For example a reporting server which triggers a warning on the disk usage on the the system disk.
When we get a graph showing us that this is a slow grow rate we can contact the responsible user and ask him for advise.
If we see that this is a fast grow rate over a limited amount of time we can take the assumption that action is needed and we need to investigate to prevent a full disk.

We know that there are reports for it, but those are not flexible enough in terms of time scaling and flexibility.

Related to the question from Vitaliy S.:
In our case only the guest disk usage is import to us since we only use thick. But I could imagine that for some users this could also be convenient to have the vmdk size itself.

Thnx!

Tim

[Shestakov]

We know that there are reports for it, but those are not flexible enough in terms of time scaling and flexibility.

Hm.. You can specify time period, scope of VMs etc. It would be great for us to know what you find not flexible and scalable in the report.
Thanks!