This guest blog post is by Mansour Karam, CEO and Founder at Apstra. We thank Apstra for being a sponsor.
At Cisco Live, Apstra will be showcasing exciting features of AOS 1.2 that our early-adopter customers have defined and embraced, as well as experimental features they have requested. We’re sharing a video teaser by Derick Winkworth (@cloudtoad) of a headroom demo, and a video by Jeremy Schulman (@nwkautomaniac) demonstrating the power of AOS’ extensible telemetry in advance of Cisco Live.
Our focus on the operational aspects of data center networks, both in terms of extensible, intent-based, real-time analytics and enhanced change operation is resonating with customers. Those early adopters who embarked on the journey with Apstra learned in the process the 5 Secrets of Intent-Based Networking.
While anyone with a need to automate their infrastructure is a prospective customer for Apstra, early adopters have been working with us from Day 1. In a series of upcoming blogs, we plan to share those use cases with our readers.
In this blog, I initiate this effort by describing the use cases of two early adopter types: the first is a typical Fortune 500 large enterprise, and the second a large webscale organization. I refer to them as LargeCo and WebCo. Their use cases urgently need Apstra to maximize their business velocity. And while they recognize that no single vendor can address their requirements, they also acknowledge they cannot build it themselves.
With AOS 1.2, we are progressing with these two initial customer types towards large-scale production deployments.
Is one of these you?
LargeCo Use Case
You’re leading the network architecture or operations team for LargeCo. Your infrastructure is a core asset to your business. You have been buying your networking equipment, software, and services from one vendor only. You are spending tens of millions on capex every year, yet you are spending at least an order of magnitude more on operating this equipment. You perform hundreds if not thousands of change operations every month to support your business. You must abide by stringent compliance requirements.
Your CIO is acutely aware of the criticality of the network infrastructure for your business. He understands that to compete in the digital economy—which requires key initiatives such as IOT, Machine Learning, or Big Data—the organization needs to scale data center operations, starting with the network. It’s clear that the culture of over-dependence on one hardware vendor has had an adverse role in the ever-increasing capital and operational expenses.
As a result of this, your CIO identified three major initiatives:
- The need for game-changing automation – the type of automation that enables you to focus on business intent, and have the infrastructure autonomously deliver on this intent. In short, a Self Operating Network – a network that configures itself, fixes itself, documents itself, and defends itself.
- Avoid over-dependence on one hardware vendor. As a corollary, your chosen operational model cannot be dependent on a single hardware vendor or proprietary features.
- Move to a new data center network architecture that leverages standard protocols to provide vast amounts of cost-effective bandwidth, owned and operated by the networking team.
To meet these goals, you need a network transformation strategy that moves away from traditional 3-tier architectures, and embraces 2-tier leaf/spine designs. And to support your L2 apps, you need to layer L2 services on top. These L2 virtual networks need to be spun up and down on demand, and you should be able to “drag and drop” physical or virtual endpoints into these virtual networks dynamically and on demand.
You recognize that the networking teams are separate from the compute teams, and conclude that networking policies should ideally be owned by the networking teams.
As a team, you realize that you need a hardware-independent, vendor-inclusive, Intent-Based Networking System (IBNS) focused on the modern data center network as a first use case. You research the market and stumble on Apstra – no surprise, since Apstra pioneered Intent-Based Networking Systems and the concept of a Self-Operating Network; and that Apstra is the only production-ready vendor-independent IBNS on the market. And, Apstra is squarely focused on the data center network as a first use case.
WebCo Use Case
You are leading the tooling team at a Web-scale organization. Your business is growing very quickly, and your network teams are building out their network infrastructure at a record pace. You plan to increase the number of devices in your network by orders of magnitude in the next few years. Your CIO has a plan to scale, and it involves the following points:
- Build out leaf-spine network designs, and leverage open standards such as BGP.
- Create a tooling team of developers whose goals are (a) to automate the infrastructure aggressively – anything that can be automated needs to be automated; and (b) build an extensive platform for monitoring – anything that can be measured should be monitored, or available on demand. You are heading that team.
- Use multiple vendors, and recognize that hyperscale vendors are able to create a unique advantage by leveraging commodity hardware and network open-source software. You need to do the same. Buying a 100Gbps ToR switch for $6,000 instead of $50,000 is a big deal.
In leading the tooling team, you have had initial successes by effectively automating initial use cases. You automated the leaf-spine network configuration management for your first two vendors. You were also able to integrate your IP management system with your configuration management system.
However, in order to keep up with the massive network buildup, you need to develop a solution from scratch. You’re struggling to find the right developers at the right rate to support this effort.
You realize that what you need is a fully intent-based, autonomous, closed-loop system and that it’ll take a massive development effort to get it built. Your efforts are falling behind, threatening the ability of the organization to build out their network, and in turn, the organization’s critical business objectives.
Enter Apstra. You watch an Apstra webinar on Intent-Based Networking Systems and decide to bring them in for a discussion. You realize that Apstra has built a platform, AOS, that integrates all the critical components for a fully autonomous operational model: including an intent layer, device and resource management functionality, and the ability to build and customize blueprints from intent, modeling and programing all the relationships in a graph model.
You love the idea of being able to generate configurations and extract telemetry using device drivers that are available for a variety of both established and commodity NetOS and hardware offerings. You learn that at the core of AOS is a state-driven data store that supports both sharding and hierarchy, hence supporting your scale and reliability requirements.
You also learn that all of the state within AOS can be accessed through RESTful APIs, Graph queries, and a streaming interface that integrates with telegraph, influxDB, Grafana, and any other time series database or visualization tool you choose. You find out that alerts are integrated with popular tools such as PagerDuty.
Then it becomes clear that Apstra has built a platform that can be used as the foundational component for your automation efforts. And that your valuable software developers can focus on the higher-value tasks of automating the specific use cases that are critical to your business, to develop the device drivers for the specific devices that you care about, extract the telemetry that you care about, and program and build continuous validation tests that are relevant to your business.
You realize that Apstra has the platform to support the automation efforts at the speed and scale of your business.
Solving Customer Problems
At Apstra we are passionate about building technology that addresses critical customer operational pain points. We are obsessed with customer success and are delivering new features at breakneck speed. As CEO I am committed to partnering with you on the network transformation “journey” as @nwkautomaniac calls it.
If your use case is comparable to that of LargeCo or WebCo, and you are facing the same pain points, please join @cloudtoad for an AOS 1.2 webinar next week.
You can schedule a 1:1 meeting to see the latest demos at Cisco Live and stop by our booth #2925 to request an invitation to our joint party with @PacketPushers. Or read the Intent-Based Networking blog post from my co-founder @SashaRatkovic.