At this year’s Cisco Live event in Amsterdam, I decided to prioritize the Full Stack Observability (FSO) sessions to learn as much as possible about Cisco’s new plans and developments. Although we at Flint already have experience with Cisco Full Stack Observability and have written about it, I went to the event with the feeling that the folks at Cisco were preparing something special for the future of FSO, and I was not let down.
On the first day of the Cisco Live conference, I attended a session, the title of which indicated I should not miss it – so I did not. By the end of the 4-hour technical seminar, I felt that the room we sat in would be too small next year – I started my Cisco Live week with excitement. This article is an attempt to summarize the information I have gathered in my search for the truth about what Cisco Full Stack Observability even is.
Cisco Live Amsterdam brought a lot of new development to the Full Stack Observability
There were several interesting developments around Cisco FSO. One was the new security feature with Business Risk Score, Kenna and Panoptica. We saw how the Grafana plugin works for AppDynamics Cloud, and how the new pairwise integration between AppDynamics and ThousandEyes adds Business Application Context to ThousandEyes and network information to AppDynamics. This year we will also hear and learn more about OpenTelemetry as Cisco embraces it at full speed and uses the telemetry with Metrics, Events, Logs and Traces (MELT) in its products.
The star of the Observability theme at the Cisco Live event, however, was the shiny new product that forms the foundation and vision for the future of full stack observability – the FSO Platform.
The FSO Platform
Up to this point, I have used the term Cisco FSO Platform to describe the combined effort of the Cisco AppDynamics, ThousandEyes, and Intersight platforms to achieve what we believe is observability across the entire application and infrastructure stack. It’s time to rewire our brains and associate “FSO platform” with something much bigger and more exciting than I initially thought.
The Cisco FSO Platform is the name of Cisco’s new solution for collecting, storing, processing, and modelling telemetry data. It is intended to serve as the foundation for building full-stack observability solutions. And believe me, it’s not another marketing attempt by Cisco, but something with its own APIs, graphical user interface, query language and even a traditional command line interface (called fsoc). We got a technical preview of the platform at Cisco Live, and the release date is set for the next six months.
How does it work?
The platform is open to receive telemetry signals from any external source as long as they are ingested in the form of standardized OpenTelemetry Protocol (OTLP) signals (metrics, logs, traces). This can be directly from your applications instrumented by OpenTelemetry (OTel) agents, through OTel collectors, cloud data sources such as AWS Cloud Watch, or any type of custom integration that uses the OpenTelemetry protocol or can convert telemetry data to OTLP.
You may have heard that some of Cisco AppDynamics’ proprietary agents now support exporting telemetry data using OTLP, and it has also been announced that ThousandEyes will support exporting some of its metrics in OTel format.
High-level architecture of the CISCO FSO platform
Once the data is fed into the platform, a set of RBAC rules can be applied, and the telemetry data is then further processed and stored so that it can be easily queried with a unified query engine (UQE), retrieved via HTTP API, or observed via a customized user interface. And the key is customization. Most of the moving parts of the FSO platform can be customized to meet the needs of a particular business. To streamline this process, the team behind the platform has developed the concept of extensibility through custom applications, also called solutions, and enrichments that can add needed functionality to existing solutions.
Build your own observability solutions
Full stack observability is only useful if organizations can easily understand the performance and availability of their systems according to their own business and system perspective. Being limited to a predetermined set of rules for observing applications and their systems is not optimal for many organizations. On the other hand, raw telemetry data is also difficult to use and make sense of. The developers of the FSO platform came up with the idea of solutions, domain models, entities, attributes, and associations to describe the observable world of an organization. This can be different for each organization, and the extensibility of the platform allows developers to customize their solutions accordingly. A solution package is ultimately a directory of files that describe the solution with a manifest and descriptors in JSON format.
The anatomy of a solution package
A solution package allows organizations to create their own observability applications on the FSO platform. Such applications can use a customized UI, selected cloud OTLP collectors, MELT processing rules, APIs, and RBAC rules. But perhaps the most interesting part is the domain models. Domain models enable easier interpretation, correlation, and traversal of MELT telemetry data. This is done through the concept of entities and their attributes and associations. In other words, a domain model allows your solution to describe the equivalent of a graph schema of your MELT telemetry data.
An example of the domain model can be Kubernetes, with entities such as Cluster, Node, Pod, and Service. Such entities are usually further represented by their attributes and associations. The domain model can literally be anything. In the lab exercise at Cisco Live, we modeled a starship domain with entities such as warp drive, torpedo tube, torpedo, etc. and associations between them, e.g., a starship has a torpedo tube and a torpedo tube has a torpedo.
Customizing the UI using our domain models
Within the domain models, you can describe metrics, events, and logs and associate them with entities, e.g., a starship entity has a speed metric, a torpedo tube entity has a torpedo count metric, and so on. When the metrics, logs, and traces arrive on the platform, they are mapped to your domain model so you can query them with UQE or observe them in the customized UI.
AppDynamics Cloud solution
We can think of the solution as an application whose backend relies on the FSO platform. Need a more concrete example of what a solution is? No problem. The first solution built on the FSO platform is AppDynamics Cloud. The domain model of this solution focuses on the observability of cloud-native applications and infrastructure.
This is a solution created and maintained by Cisco, but you can see the solution package and review how the domain model was created. Along with acquired products from companies like Replex, the AppDynamics Cloud solution is extended to create a cost breakdown for your cloud-native infrastructure.
The idea, of course, is that anyone can create a solution, not just Cisco’s internal development team. How Cisco envisions monetizing solutions developed by outside companies is not yet known.
Is the FSO Platform replacing AppDynamics cSaaS and ThousandEyes for Full Stack Observability?
The FSO Platform will not replace AppDynamics or ThousandEyes. Both are used today in Cisco’s FSO vision, and the new integration capabilities between the two only strengthen the case for their use. The maturity of OpenTelemetry’s agents cannot be compared to AppDynamics’ proprietary agents. Advanced tracing and snapshot deep dives are simply not yet possible using OpenTelemetry alone. Enterprises are likely to take a hybrid approach and use AppDynamics and ThousandEyes alongside the FSO platform, at least until OpenTelemetry’s signals mature and they develop domain models in the platform that cover all aspects of their enterprises’ observability.
So far, we have talked about Cisco’s FSO in the context of the 7 Use Cases. Now this concept is extended to include the business context, services, and capabilities of the new platform.
The FSO use cases are still here, but the vision has expanded
As we previously stated, Cisco FSO is like a Swiss Army knife for application observability. Now that we can see that they are taking all the necessary camping gear along with them, in addition to the knife, I think the trip will be enjoyable.
If you have questions about the FSO Platform, AppDynamics, AppDynamics Cloud, OpenTelemetry, or what’s new in the world of Cisco Full Stack Observability, Flint’s team of knowledgeable and always up-to-date experts will have the solution.
Flint professional and mentoring services
The Flint team has continued to invest in leading-edge technologies to help drive the adoption and benefits that they can provide. Our team uniquely works directly with the vendor to support the creation and release of training for these technologies as well as provide professional and mentored installation services to support partners by supplementing their teams and guiding them through early deployments successfully.
Should your business be looking for support with an FSO opportunity, please contact the team at [email protected].
We also offer support on the following related solutions:
ACI, Hyperflex, SDAccess, SDWAN, DNA Center, DNA Spaces, ISE, Stealthwatch, 1000 Eyes, IWE and AppDynamics.
Get in touch, we are looking forward to hearing from you!