“We will see an intersection between of observability and DevSecOps”
We spoke with Stela Udovicic, Product Marketing Director at VMware Tanzu about the State of Observability Report 2021. Stela discusses increasing cloud complexity and its drawbacks, what observability red flags to watch out for, predictions for 2022, and more.
JAXenter: First things, first: What is your role at VMWare Tanzu?
Stela Udovicic: I’m a Product Marketing Director, leading marketing for the observability function. I work closely with DevOps teams and across multiple business areas at the front and backend. I have considerable experience in the AIOPs field – I’ve been with Tanzu since day one when it became part of the VMware brand. Prior to that I was with Wavefront which became part of VMware four years ago. Before that I led software defined networking and other DevOps marketing initiatives at Splunk.
JAXenter: According to the State of Observability 2021, only 16% of IT practitioners currently use observability tools. But this number is changing. Why are more and more IT teams planning to deploy observability practices?
Stela Udovicic: If we take a closer look at the adoption data for Tanzu’s recent State of Observability 2021 report – yes, it is true that only 16% of IT practitioners are currently using observability tools, but another 16% are in the process of implementation, with a further 18% set to join in the next six months. That puts us at 50% who will use observability tools in the next six months – this is an area that is actively growing in uptake for IT practitioners.
As well as this, our research shows only 8% haven’t heard about these tools, and the popularity of containers, microservices, and APIs is skyrocketing. 77% of survey respondents report hundreds or even thousands of containers and microservices across their app portfolios – and as overall use of modern application architectures increase with more than 50% report having highly distributed applications, organisations need to rely more heavily on observability tools.
JAXenter: What red flags should teams look for when considering observability?
Stela Udovicic: There are several red flags teams can look for when considering observability.
Firstly, teams must carefully assess the kinds of performance metrics they need to collect, and what they must do to measure these at high granularity. Observability platforms such as ours collect lots of metrics and offer flexibility in fine-tuning the metrics collection.
Secondly, teams should consider which observability platform will integrate well with existing tooling. Open source monitoring is very popular and will require integration. Our recent research found easy integration with open source and existing tooling is essential for 51% of respondents – and not only integration, but also preferred deployment.
Many teams try open source monitoring which is great for small deployments or for really savvy development shops – but what most people don’t think about, is what happens if they you lose their observability data. In this case, looking into things such as high availability of platforms or how to release code in production if my monitoring tool is down is crucial.
Last but by no means least, teams should consider what the cost of unoptimised infrastructures that run the open source observability platforms they are maintaining is. From this, it’s vital to factor in infrastructure costs.
JAXenter: Cloud environments are more complex than ever. Could you tell us a little more about that?
Stela Udovicic: Yes, based on our findings, the complexity of cloud environments is rising. For instance, 84% of IT practitioners report hundreds of thousands of compute instances across their portfolios.
App requests are now going through dozens of technologies and third-party APIs in the context of distributed tracing, cloud databases, and digital queues.
To be more precise, nearly 50% report that app requests touch over 25 different technologies, and 10% of them state that number is even above 100 technologies. The need to deploy troubleshooting has skyrocketed across highly distributed applications, increasing risks of bottlenecks across so many overlapping systems.
These bottlenecks can impact suppliers all along the chain and impact your overall cloud service availability for which your team is responsible.
JAXenter: Are there any drawbacks to cloud complexity?
Stela Udovicic: Certainly, there are some drawbacks to cloud complexity – 96% of survey respondents report challenges due to complexity and there’s a range of these.
The more complex the environment, the harder troubleshooting becomes. A more complex environment has distributed applications that are updated frequently, which means they teams will need to act fast when issues arise. With so many endpoints to monitor, tools must be able to scale to avoid incomplete coverage.
But now, due to DevOps culture and practices, more than ever, you need to have the complete picture, which is difficult.
Another issue is having too many disparate monitoring tools for different elements of the cloud. We can see that more than 51% of respondents use more than five monitoring tools across apps and infrastructure, from traditional style logging and NPM tools, to siloed container tools. And the larger the enterprise, the issue becomes even more prevalent. Drawbacks include a lack of visibility or siloed visibility, which prolongs troubleshooting and makes it difficult to maintain existing tooling.
JAXenter: Any predictions for the future? What changes will we see in the State of Observability 2022?
Stela Udovicic: Although it is already happening to an extent, over the next year, we will likely see an increase in businesses linking observability to their outcomes. Indeed, we have customers that are doing that today, I discussed this in a blog on our customer, a well-known ride sharing tech company.
Indeed, if your app is your business, then observability had direct impact on your business, and so we will increasingly see data-driven business decisions becoming more and more common. For many of our customers, seconds and sub-seconds are highly important, because sub-optimal performance or critical delays can translate into revenue losses. If on a Friday evening you cannot order a lift or food delivery through the app, then there is a revenue impact for the sales team.
Secondly, we will see more self-healing. One of our customers recently said that if you instrument your observability and your CI/CD properly, by the time SRE teams react, the alert is probably already resolved. Similarly, implementations of automated technologies, such as AI/ML will increase over the next year. For example, ‘surfacing’ which helps automatically spot problems for DevOps teams to troubleshoot. This is particularly important when you look at highly distributed environments with millions of traces and metrics. AI/ML algorithms can assist human operators in providing all necessary points to explore where the issues lie.
Finally, as with lots of software technology at the moment, we will see an intersection between of observability and DevSecOps. As developers are adding observability early into applications, they need to define which metrics to use and their code must be secure.
JAXenter: Did anything surprise you during your research for the State of Observability 2021?
Stela Udovicic: It was surprising at first to see ‘synthetic and real-user monitoring’ come in as the third highest valuable observability capability. But if you have in mind that we’re currently still in a pandemic where remote working is prevalent, then it makes sense that user-perceived experience is of high importance.
JAXenter: What industries, in particular, leaning more heavily towards observability?
Stela Udovicic: The industries leaning more heavily towards observability are those that we say have ‘tech DNA’ – so largely, software companies. Right now, we’re seeing a trend in fintech companies leaning towards observability – they’ve already adopted DevOps processes and automated deployments, so those companies in the space that are looking to scale up are now using monitoring tools to do so.
Thank you so much!