Powered by patents to collect, store, and analyze long-term time-series data, MARS is a technology platform providing a high-performance analytics engine that can work with a massively scalable data repository.

Issued Patents:


Revolutionary NOSQL ™ Technology

Cumulus’ proprietary NOSQL technology allows you to retain data at the lowest time granularity, providing constant time analysis with no need to roll up data. It is specifically designed to provide an expert the information to quickly find issues that need to be addressed, or an IT generalist the capabilities to quickly identify the source of the problems, without the added trouble and expense of hiring an external expert to diagnose the issue.



Proven Results

Applications powered by the MARS platform capture data in near-real-time from devices in a Data Center such as Servers, Virtual Machines, Storage Systems, Switches, etc. This data is processed continuously to provide analysis on performance as well as adherence to vendor best practices. Built-in recommendations help data center managers to improve the performance and reliability of the overall virtual infrastructure.

Products powered by MARS have been implemented at more than 700 customers world-wide.


MARS ‘Probes’ collect metrics from target devices using industry-standard protocols. Since the data is gathered from production environments, these probe has been carefully designed to be secure, non-intrusive and with minimal impact on the production resources in a data center.

MARS gathers data at an unprecedented granularity from all target devices: from the applications to the virtual machines down to the individual volume on a storage system. This level of granularity can be maintained for as long as the customer wants. This becomes critical for capacity planning and forecasting.




The biggest challenge in virtual infrastructure analysis is the amount of performance data that has to be processed. The increase in the number of layers in a virtual infrastructure, along with the increased number of components in each layer, means the quantum of data that has to be analyzed increases dramatically.

To analyze the data management of large infrastructures, traditional performance analysis tools have typically relied on relational databases. Despite improvements in such technologies compromises have been required. For example, restrictions on the length of time data for which lower time granularity data can be retained, limits on the number of components of an infrastructure processed, or on the number of metrics used for analysis.

MARS takes a refreshingly different approach to handle large amounts of historical time-series data without any external relational database dependencies. MARS analytics engine is closer in approach to popular NOSQL technologies and Big Data engines used by the global technology leaders. The platform allows analytics that the traditional tools cannot without using very large, performance-warehouse class machines.


Being able to analyze large amounts of data is great but it is equally important to use the analytics to provide useful recommendations. MARS uses industry-standard best practices to create recommendations include all areas of the virtual infrastructure.

One unique aspect of MARS is its powerful correlation capability. This makes it easy to view and analyze not only resources that are directly correlated such as storage systems and virtual machines, but also in situations such as for storage subsystems independent of the server infrastructures. This means, if the storage subsystem is experiencing issues due to elements outside the server infrastructure, it can still be measured and analyzed.




MARS is delivered as a package comprising the Probe and Server components. The probe provides an option to communicate securely with the Server through firewalls. This provides for a flexible deployment model for the virtual server, either on site or to a service provider.

For large customers with distributed multi-site environments, MARS provides for a centralized performance data store that is low cost and highly scalable. This allows organizations to create a centralized method for consistently viewing and managing their multi-site infrastructures.