DevOps

DevOps and Monitoring, hmm

Nobius blue logoNobius was formed from the need to assist customers build enterprise grade IT Monitoring solutions without the costs typically associated.  But isare Monitoring and Devops comatible?
For some people IT monitoring isn’t needed anymore.

My conversation with a young developer.

She claimed infrastructure monitoring is old hat and not needed anymore with the adoption of cloud and DevOps.
She said her new applications were deployed in containers (#Docker or #Azure containers I don’t recall) so if they had issues, the host would restart them.Banana skin
Hmm. I could see her point. I smiled and asked what happens if they just keep being incessantly restarted?
Well she claimed that IT operations would know and fix it.

Hmm. I asked what information would they have to diagnose what was causing that issue? What tools would they have to stop it happening? She said she gave them native tools she developed with her application.

Hmm I asked her how many applications she and her team mates roll out in the last year. Probably 30 or so, she said.
Hmm I asked her how many tools did she think IT operations could handle, remember and use?
She saw my point.

In any case I said, the applications are meant to serve users and the business they generate. As long as the application keeps being rebooted incessantly, the service her application is supposed to provide is out of order.
She agreed and asked what would I do?

Focus on excellent service delivery.

Service by WaiterTherein lies a great question, and I hope answers are on the lips of DevOps guys out there because what she was asking defines the need to integrate Dev and Ops and how monitoring must evolve.

To answer her I said change is always upon us, and the choices you developers can make of languages, tools, environments is vast. Then there’s choice of deployment or implementation too especially with hybrid cloud. Whatever choices people make, IT operations should always be focused on service delivery i.e. making sure that users can consume.
They’re measured on it and the MTTR of issues.
Focus on service delivery should one of the key common design constraints of anything new.
ALL new projects MUST include the means to allow excellent service delivery.

She was nodding enthusiastically now, but asked how do operators know if the service IS being delivered, and which application is faulty if it isn’t?

Hmm, I continued. The root cause of issues affecting service delivery spans all components. These include elements, resources that applications, but also the container host, depend on, right down to most meagre fan in the power supply. If that stops working and the power supply fails, chances are contingency such as batteries or generator will maintain service but only for a while.Best Practice
To make things even more involved, what if a purchase can’t be made because the credit check on the user’s account is failing. That’s probably a service dependency outside the control of IT Operations because its furnished by another organisation.
All these factors add up and should be considered when deploying new applications and services.

Ok she said, sure but you didn’t answer my question, how does an IT operator know.
MONITORING I answered

End user experience monitoring

She started to roll her eyes, but I asked her. How do you test your application before releasing it?

We run sessions against it, testing navigation, page execution etc.

Hmm I continued, well IT operators need the same thing, and that’s purpose of end user experience monitoring. This capability injects traffic across the web to invoke your service just like a user, and then replays instructions just like he would when he consumes the service. These synthetic sessions help to continuously test service delivery, capture slow pages and other issues. With this you can then alert when it under performs or simply doesn’t answer.

Tilting her head, she said so at least that kind of monitoring sounds useful.

Yes I agreed but what does the IT operator do if all he has is an alert concerning the application. He can’t remember how your little tool works, with 30 other apps and their tools that’s not going to be easy. You can see that he is going to need to correlate that alert with the elements we talked about earlier. His analysis needs to lead him to the root cause, whether it’s the app, the DB, the virtual server, the cloud instance, the network or that pesky fan in the power supply. Infrastructure impacts service delivery. Do you want him to escalate to you that your application isn’t behaving, if in fact the reason is that fan?

She frowned and said so that’s why infrastructure monitoring can’t be ignored?
But all those monitoring tools are expensive to buy and own.

finger pointingSomeone else’s problem?

Putting everything in the cloud means those infrastructure issues are somebody else’s problem.

Hmm I added, but if it’s impacting your service, your business, that’s your problem.
Can you afford the loss of business when the cloud vendor’s infrastructure impacts your service?
How do you connect to the cloud? Does your network provider have issues?
If you only have one cloud vendor, you can always point the finger. Do you use more than one?
What happens when you have several, 2, 5 maybe even more. I know from talking to many of your colleagues in development, that a new project may easily subscribe and consume different cloud vendors, AND different services from each of them .

Measuring service delivery is vital

Working out what it is influencing it is a multivariable equation to solve. Did you ever here the phrase “You can’t manage what you don’t measure”.
If you agree with that, then you know why monitoring is necessary.

Monitoring must evolve

Well since that conversation moons ago, I think the growth of tele working has alertedpeople even more to the the real need to monitor.  But like everything, monitoring must evolve. Has yours?

Nobius recommends using #Zabbix an enterprise grade IT monitoring software.
Guess what, its open source so its totally free !!
And it just evolved, with version 5.2 where importantly, #Zabbix has added synthetic monitoring. Nice. Well done #Zabbix

Nobius is here to assist you

25% Discount

If you need better IT Monitoring, on premise or hosted, why not reach out to Nobius.

did you attend the Zabbix summit 2020? If so, then you’ll qualify for a 25% discount on our services ordered by 31st December 2020

Our services are described on our services page and our video here, our website or just reach out to info@nobius.co.uk.