The Stirring Simplicity of SaaS for Network Visibility

Summary

Let’s face it – we live in a world of increasingly complex software products and systems, upon which organizations are absolutely dependent to function. While software-centricity has allowed phenomenal rates of innovation and growth in productivity, software systems can be notoriously expensive to own and operate. Research going back to Barry Boehm’s seminal text on Software Economics estimates that up-front costs of software comprise as little as 25% of the long-term cost of ownership (TCO). Much of the cost equation involves deployment, configuration, and ongoing maintenance of the software and the hardware to host/run it.

Network and security monitoring solutions are a prime example. Keeping today’s networks up, running, and safe is a major challenge given their scale, traffic growth, and criticality. This has resulted in a wide array of complex, highly functional and highly capable solutions coming to market, which carry with them not only large up-front licensing price tags, but often heavy costs for installation and administration.

To simplify deployment, many such products are offered in an appliance model, where the software comes preloaded and preconfigured (to a degree) on a hardware platform. This is helpful, but comes with some common downsides – namely the high up-front capital outlay required, the limited lifetime of hardware (can you say “forklift upgrade?”) and the fact that such approaches do little to reduce the pain of systems administration for software updates and patches.

How can we do better? The public/private cloud approach can help reduce hardware costs, if the product of interest can run in a VM and if you can get your management data to your cloud provider of choice. A promising alternative is Software-as-a-Service (SaaS). SaaS is a proven and accepted model for many software product sectors, including CRM, APM, ERP, and more. SaaS offers some compelling advantages, including:

• Blazingly fast deployment: Our experience is that basic monitoring can take as little as 10-15 minutes from first keystroke to first data graph.
• OPEX instead of CAPEX: Taking subscription-based bites at license costs instead of a big upfront swallow is a great option for most shops.
• Better/faster support: The SaaS provider looks after the viability and resilience of the system 24x7, applying patches and new features with little or no delay. No more “sorry, we can’t reproduce your problem” support black holes and more more “we recommend you upgrade your operating system” brick walls.
• Easy scalability: Again, the SaaS provider is responsible for making sure adequate compute and storage resources are available to deliver committed levels of service. No more worrying about whether you have enough RAM or enough disk for the system.

To be fair, there are some disadvantages as well, or at least perceived disadvantages. The two we hear about most often are:

• Data control: Many shops worry about their sensitive data being held and protected by an external entity. At Kentik, we have overcome such concerns with most organizations by providing secure data transport and assured multi-tenant partitioning. Thinking of analogous services helps too. Consider Salesforce here – if your CRM data is considered to be safe with an external provider, then in most cases your network flow data will be too. And if it’s just too scary, we bring out private hosting options (though few teams end up needing to go that far).
• OPEX instead of CAPEX: While most organizations like the subscription operational expense option for license fees, some only have the budget available for software systems on an occasional capital basis. As with most SaaS providers, we’re glad to get creative on those situations.

The Time for SaaS Network Monitoring is Now

Before joining Kentik, I spent a couple of decades in the network management sector, largely focused on traditional product architectures that required local deployment and administration. In the past several years, I’ve been intrigued by the utter lack of SaaS options among high-end network management solutions. As an industry analyst at EMA, I conducted direct primary research on the topic in 2014, and found that on average, 80% of enterprises were amenable to SaaS models for network management – there just weren’t many choices available.

There are a few SaaS options out there already for basic network infrastructure monitoring, such as Auvik and LogicMonitor. But choices are fewer and farther between when it comes to network traffic monitoring. AppNeta offers a hybrid SaaS solution that does network and application performance monitoring, but it is designed for small-medium enterprise scale only. Similarly, Polygraph.io offers SaaS-based flow monitoring, but is limited in its ability to scale for large managed environments.

Kentik is the first to take on ultra high-volume flow monitoring using the SaaS approach and make it work for massive scale and near-real-time (sub-minute) results. On the scale front, we are already monitoring environments that regularly generate in excess of 1 million flows per second, and our back end SaaS platform is currently ingesting over 40 billion flow records per day. Our clients are getting immediate results when they start using the service, and our operations team constantly monitors the health and success of each managed environment.

There is beauty in simple answers to complex problems. As long as you don’t have to sacrifice capability or assume uncomfortable levels of risk, the real question around SaaS is no longer “why?” The real question is “why not?”