Part 2 of an in-depth exploration of the uberAgent product – how it performs out-of-the-box
If you haven’t read part #1 of this article yet, then you’re coming in half-cocked. Go and read it first!
We left our overworked systems administrator AdminX having just completed their install of uberAgent and seeing that they had data being collected into the console. Which is always good – nobody likes to have an install go west on them.
So what next? If AdminX is anything like all those other systems administrators in the world, they generally have a laundry list of concerns that they’d like some more clarity on. What are their problems?
The burning issues
Session performance is always high on any EUC admin’s list of issues. Users always complain about performance. AdminX has a particular interest in browser performance, particularly in that of Chrome as it has undergone a lot of adoption recently. There’s also the whinges from the users about performance of audio and video.
Some particular users also complain about logon times. AdminX would like some more visibility into this and ideally baseline them over time, as they personally haven’t noticed any logon problems. Assessing whether this is a problem tied to specific users, or if it is just indeed a perception issue, is another key area for them.
Spotting devices that are in a hung or problematic state is also a major concern. Occasionally devices hit problems and users are left high and dry until it is identified and resolved. Getting some proactive visibility into this is key.
It is also important for AdminX to get an overview of the components that make up their environment. Now, every Citrix infrastructure isn’t solely dependent on the Citrix aspects for optimal performance. There are many other moving parts – Active Directory, networks, storage, maybe multi-factor authentication, etc. It would be nice to have a view into these from a single “pane of glass”, to use one of the industry buzz-phrases.
And of course, every admin would also like to know about problems he doesn’t know about yet. AdminX is no different. But at the initial deployment stage, they are interested in seeing realtime and historical data, and particularly that concerned with the areas they’ve highlighted above. How easy is it to get this information out of uberAgent?
Engaging with the user interface
In my experience, an absolutely vital part of getting an overworked admin like our AdminX on board is allowing them to navigate and use the product quickly and easily. You have a short timeframe in which to get them comfortable – admins are fickle creatures!
A lot of people say that you should have pre-sales engineers engaging in workshops and following up with the admin at this first stage, but I disagree. A lot of admins think that if a product needs handholding through its deployment then it possibly isn’t worth its salt anyway. I, along with many admins, prefer something I can build and test on my own. Of course, if the deployment is particularly large and complicated then it may be worth getting help, but as most people start with small PoCs, I don’t think it is hugely relevant.
There are four stages around this process:-
Honeymoon – after the admin has installed the software, and is looking forward to getting their new product up and running, they are quite willing to dig through the user interface and try and figure out any issues. This stage doesn’t last particularly long (dependent on the individual admin), but let’s not forget that most people are limited in time that they can spend on particular subjects.
Frustration – the second stage comes along when the admin can’t quite do what they want to do, mainly due to unfamiliarity with the interface. At this point, they will generally search help files, documentation and the internet to try and find out what they may have missed. This stage can last a while, but the quality of the documentation is absolutely key. Don’t forget, very few admins read manuals – they just dive straight in, and only fall back to documentation when they become frustrated. Having good documentation is vital because this is the chance to educate them on the interface and possibly even push them back to the Honeymoon stage.
Exasperation – the third stage comes along when they find the documentation and/or internet sites/forums don’t solve their problem, and this is where the danger of losing them begins. At this point, the admin may start to become annoyed and develop a negative perception of the product they are using. It is common at this point for them to engage with support in some way to try and rectify their issues. This is almost the last chance to rescue them – so support needs to be good. You need to make sure that the admin’s issues are solved and that they are happy using the interface going forwards. It is another chance to give them a bit of education, point them at some helpful resources, and generally rebuild their confidence back up in the product.
Rejection – if support doesn’t solve their problems, then you’re moving into the final stage and this is where they are ready to throw in the towel. This is where it’s not just a problem in terms of this particular deployment, but in terms of the negative feedback they will pass on to other admins. In the first part of this article, we made the point about how vital community endorsement is to product adoption – and if an admin has tried to use it, found it challenging, and then received no joy from either documentation, forums or support…well, you’re past the point of no return really. Maybe not entirely – I’ve seen products dragged back from the rejection stage in the past – but for the most part, once AdminX reaches this stage they’re not going to be keen to move forwards.
But let’s make an important point here. No product is 100% easy to use out of the box, especially when we are talking monitoring which is an area where each admin generally has different expectations and requirements. Anyone who tells you that an admin can install a product and immediately start making it do precisely what they require is a liar – for one thing, it depends on what they are proficient in using to start with. Everything has a learning curve – it is how much of a learning curve that is key, and how quickly you can get to a stage where you feel proficient in presenting and organizing your data sets.
Each admin will also compare with any previous tools they are used to dealing with. This isn’t unfair, it’s simply natural. Which means it is all the more important that it becomes easy for the user to navigate and familiarize themselves with.
So how was my experience with uberAgent?
I think it is fair to say that I didn’t find uberAgent particularly frustrating to deal with when I first started using it. Given all of the variables in play (and anyone who knows me can testify that I am one of the most grumpy and impatient users of software in the world!), I’d say this is a pretty good first impression. It helps that it is a web-based console – the interactions don’t feel unnatural and it is easy to navigate around. Once you’re aware that you’re launching the Splunk console and from there the uberAgent app from within it, jumping from screen to screen feels quite relaxed and smooth. I must make the point that it feels much better in Google Chrome than in Internet Explorer, but that should really be a given these days.
Using my sliding scale of “UI engagement” above, I was mainly in “honeymoon” and only had a few limited forays into “frustration”. Whenever I got frustrated, I managed to get what I needed from the helpful uberAgent support people. For a product I was unfamiliar with, this is actually pretty damn good, to be honest! However we will continue to monitor how it performs on this scale as we move on with the set of articles.
But on a more granular level; how easy is it for our admin to get into the bits that they’re interested in?
Visualizing the basics
Navigation of uberAgent is pretty straightforward, as I’ve already pointed out. You can easily see which part of the environment the menus refer to and jump into the submenus for further data.
Logon times were a primary concern – and getting this information couldn’t be easier. Simply hop into Sessions | User Logon Duration and you have lots of data already presented there. Immediately, our AdminX can see if they have bottlenecks around logons and which areas are causing those bottlenecks.
Also, session performance – particularly browser performance – was also of very urgent interest. Again, it is easy to find these metrics. It is worth noting that uberAgent particularly shines here – we have absolutely tons of data available around browser performance. We can separate it by Internet Explorer and Chrome, we can look at the resource utilization on a machine level, or we can dive even deeper into page loading times, data transfers and other browser metrics. For those admins (like me!) who are very interested in seeing how the browser is performing beyond the simple measurements of “how much RAM and CPU is it using?”, this is absolutely invaluable – and I can categorically say that when I revisit this subject for a blog post I will definitely look to use uberAgent to measure it. Being able to see browser performance based around the page URL is awesome. The detail level in here is really really useful – very impressed with this aspect of the product.
AdminX is also interested in pinpointing problematic systems before they start to cause an issue. This is one area where it seemed to be a little more tricky to get it visualized straight “out of the box”. However, if they are Citrix Virtual Apps and Desktops systems, you can see an overview from the SBC/VDI menu item which is helpful – for instance, I can see if machines have become unregistered. What would be more interesting is some sort of view where I can spot machines that potentially are exhibiting behaviour that may indicate an issue – maybe a sequence of events in the logs that we know cause an issue, or a particular Citrix port has stopped responding. We will investigate further whether we can create some custom views around these areas in the next part of this series.
We also mentioned we’d like to see if there are any problems that the admin isn’t yet aware of. uberAgent’s default views give a lot of detail and while browsing through these, we can see that a particular SMB share is performing really badly in terms of latency. That looks like a potential problem right there, so AdminX can head off to do some investigation on that straight away and fix a possible issue.
Also, browsing through the Citrix views available, we also notice that only one user is connected to the Citrix infrastructure using the latest Workspace App version. It’s clear that users aren’t heeding the admin’s advice to try and use the latest client versions where possible! This also warrants some further investigation and can also potentially be a fix for problems that users may be having.
Now, we also mentioned that every AdminX out there doesn’t just want to see all of the SBC/VDI information – every infrastructure is different. Now this isn’t particularly an uberAgent feature, but it does show why adopting Splunk (which uberAgent runs on) can allow you to very quickly extend and customize your monitoring system.
The admin can simply log onto SplunkBase and download Splunk apps that deal with different infrastructure components – for instance, the Microsoft Office 365 app. We can plug this in really quickly so that as well as monitoring our SBC and VDI environments through uberAgent we now have dashboards looking at our other components as well. I was very impressed again by how quick and easy this was to set up – out of the box, we start getting a view into utilization of our Microsoft 365 tenant and particular apps within there (such as Teams).
I think this flexibility is a very important part of what makes uberAgent a very interesting solution. With other monitoring tools that are specifically aimed at the SBC/VDI space, they concentrate very much on looking at those particular elements of the infrastructure. With uberAgent, I get that deep visibility into my EUC estate as well, but I can also use Splunk to add on extra components that give me a much broader and customized overview of the infrastructure under my control. The fact that these extra apps are simple and easy to download, and, like uberAgent, offer valuable insight from the default views, make me warm up to the suite even more.
So, how was the overall impression of using uberAgent and Splunk for the first few weeks?
Overall, we’ve found it has ticked most of the requirements that AdminX came in with.
- We found the console pretty easy to navigate and not frustrating at all – changing the default time segment for some of the views was the only part we changed from the out-of-box settings
- We’ve got extremely detailed information on session performance and particularly browser performance – in fact, there was a huge amount of data available on browser performance
- We’ve got a good detailed breakdown of logon times and we can trend them over time very easily
- We have highlighted some problems we didn’t know we had and can now take proactive action to rectify these
- We can very easily extend the scope of our monitoring to encompass other vital components of the estate, and get a great out-of-box experience with these as well
- We did, admittedly, find it a bit difficult to see a full overview of the Citrix Virtual Apps systems to see if any were approaching a problem state, but we are going to try and address this in the next part by generating more customized views
Let’s also make the point that we’ve approached this simply from the perspective of our own requirements. uberAgent also, out of the box, has many more views that administrators with different requirements may well find particularly useful. There is lots of information on ADCs, application performance, startup and shutdown, storage, networks and licensing that we haven’t even touched on yet.
So far, then, we’ve had good value from uberAgent. The out-of-box experience has given us some good information and a lot of detail around particular areas we are interested in. Next, we shall be looking at how we can start to customize it to get dashboards tailored to the exact requirements we have. Stay tuned!