June 19, 2013
We all now the drill. The rise of cloud application has meant that performance monitoring, formerly all about delivering metrics from one stack, has been blown apart and needs to become much more modular and distributed in nature. Companies like NewRelic and ExtraHop have arisen to deliver upon this new distributed stack monitoring opportunity. But what happens when outside factors (read – the internet) have an impact on application performance? Maybe management tools should include this in their scope of monitoring.
That’s the premise of ThousandEyes, a company coming out of stealth at Structure today. Their elevator pitch rams it home:
What if you could get a single pane of glass that would give you a view beyond your edge firewall? With the adoption of cloud applications, the internet is becoming the backbone for the modern enterprise, and the IT team is left to deal with the hard performance past the infrastructure perimeter.
Their solution allows organizations to view the internet like their own network. Using ThousandEyes troubleshooting can occur across the application stack and the delivery network – already in beta use by companies like Evernote, ServiceNow, Twitter and Zendesk, thousand eyes is backed to the tune of
$3.5M $5.5M by Sequoia.
The theory goes that traditional performance management products were designed for more tightly controlled and constrained environments such as individual data centers and corporate networks – with the rise of all things cloud, this constraint is largely removed resulting in far more fractured architectural topologies in use within organizations.
ThousandEyes therefor takes a network centric approach and sees the entire internet as one large network. It also tips its hat in the direction of more distributed development and operation teams and includes a collaboration platform for data sharing across all parties who “touch” an application. ThousandEyes describes the monitoring continuum in the diagram below and suggests that they are the only vendor that delivers on all three parts of the continuum.
As the company says:
There are three segments involved in the consumption of cloud applications, the enterprise network, the Internet and the cloud application network. Traditional tools operate either only on the enterprise side (e.g. netscout) or on the data center side (e.g. AppDynamics) but nothing focuses on the entire picture from a network perspective to understand the entire delivery path.
The core elements of the ThousandEyes solution include:
- X-Layer provides visibility into each layer of application delivery and a connecting thread between these layers, making it possible to navigate from layer to layer to find the root cause of problems
- Deep Path Analysis provides a view of the end-to-end path between the client and server, including localization of loss per interface, link delays, route changes and measurement of capacity and available bandwidth
- Interactive Sharing enables enterprises and application providers to share live data with each other and resolve problems rapidly, eliminate ad-hoc tests, reduce manual processes and cut multiple conversation round trips between IT teams
It’s a logical offering, but one which has a few arguments against it. Operations teams have the ability to change factors within their control. However “fixing the internet” is largely beyond their capabilities. While monitoring the entire network is worthwhile, management can really only occur within the confines of the particular applications virtual stack. As Lew Cirne, CEO of NewRelic pointed out to me:
The problem with public internet performance monitoring is that remediation is rarely under the control of the customer (the owner of the application)… Deep diagnostics aren’t as valuable to the customer because they really don’t have much they can do about a problem in somebody else’s network gear
There is always a bit of tension between raw monitoring and actual management – many so called management offerings are really simply a monitoring solution – that’s not to say they’re not worthwhile, just that they deliver only one side of the coin. This startup seems to have a degree of that syndrome, monitoring called management – it’s not a negative per se, but somewhat limits their utility. ThousandEyes seems like an attractive proposition, but potentially one which would sit well within a traditional monitoring vendors that’s looking for a deeper distributed stack and public network story.