2 posts tagged with "monitoring"

Splunk recently released Remote Work Insights - Executive Dashboards. An organization can create interesting dashboards by collecting data from tools like Zoom, Okta and the VPN software. This can enable executives to figure out how well their remote tools are working and where people are spending time.

For example you can create dashboards like

1. Number of Zoom meetings active in the org at a time, classified by type of meeting (Scheduled meeting, personal meeting room, etc.)
2. Duration of zoom meetings and a histogram of its duration - This can suggest how long are meetings going, how much time an individual is spending on meetings.
3. Are meetings getting extended from their scheduled time - Can help executives figure out if time in meetings are being utilised well. Zoom meeting etiquettes training can be introduced to check that.
4. Single Sign On systems like Okta can tell which apps are people using most. This could help executives detect which SaaS tools are actually being used and which are lying idle. This can help rationalise their SaaS spending.
5. Integrating with VPNs can also tell number of people logged in via VPN, location from where they are logging in, etc.

Splunk Remote Insights - Dashboard Now, Splunk had to face lots of flak because of this. People got enraged that this is a corporations trying to extract work from their employee, that too when everybody is struggling on their own. But if you look carefully, I am not sure that is the objective. They are not trying to monitor if people are doing their jobs. They are just trying to figure out if people have enough access to do the work, are they facing any difficulties, etc.

It is very similar to how you monitor an app or service. Remote working tools like Zoom, VPNs, etc. are providing a service. How do we figure out if they are working well for the employees? If certain location shows high number of failures in VPN logging, there might be some issue with the VPN provider or network of that area. If people are spending too much time on meetings, then there needs to be some sort of training to be more productive over Zoom.

Since, this is a new experience for many teams, monitoring the tools more closely only helps to figure out issues and solve it.

This is what I call "Tools Observability" and we get more remote friendly, this will get more important.

I run a business which builds products for application monitoring. There are many use cases it solves like:

• Finding how much time are your applications taking
• Which component of the software is slower compared to the other
• Which applications are giving high error rates
• What could be the root cause of the issue
• Can we generate proactive alerts, before the issue actually happens

Screenshots from SigNoz - Monitoring applications While this seems like a very specific use case for monitoring software, the underlying constructs are more universal and applicable in many other domains. It's just that these techniques are first being applied on software monitoring.

Let us take a more real world use case. Suppose you are in charge of delivery in Swiggy or Uber Eats, wouldn't you like similar monitoring graphs. e.g

• How much time is each delivery person taking to deliver the orders
• Which components in the overall delivery chain is taking more time? Are delivery executives taking more time in reaching restaurant, or are restaurants taking more time to cook the food? Which part of the chain is becoming a bottleneck?
• Which restaurants are telling that ordered dishes are not available?
• What is the root cause of the issue? How do we know which part of the chain is cause of the problem
• Can we get proactive alerts if something is about to go wrong in the delivery chain. e.g If less delivery boys have logged in on a day in a zone, if some restaurants are facing any issue.

Aren't the above points as good as monitoring a delivery business? They map one-to-one with what we do currently with applications & cloud infrastructure. It is a thing of wet dreams for a ops manager in food delivery companies. Of course, some parts of the above are already being captured in some analytics dashboards, but it is not to the level of sophistication of software monitoring. Root cause of analysis, proactive alerting, are still a matter of dreams. Business Processes like Delivery should be monitored Monitoring is a much broader concept than we realize today. Rather than just software stacks, we can monitor business metrics, we can monitor utilization of resources in any factory. This is what control systems was when factories were new things. Researchers would devise new & ingenuous way to monitor different machines & components in the factory. Any process within an organization can be monitored in a continuous way.

"So, why it is not already being done?", you ask. Well the key pre-requisite for setting up a good monitoring system is granularity of data. You should have high granularity of data on a continuous basis for each component of the system to get meaningful insights from monitoring.

Unfortunately, most meat space processes don't satisfy that requirement. Monitoring is actively being applied in software and applications, because that is one of the few places where we have good granularity of data.

But, as we are moving to a more data centric world, when each of our activities are tracked or logged somewhere, I am sure that the day is not far when we would be able to monitor processes in meat space also. Monitor different real world processes. Monitor business metrics.

Heck, why didn't the author get an alert when his cheese was moved? 😆  We would have escaped going through a whole book on it.