Druid is the first open source, distributed analytical database designed to scale horizontally. Its distributed architecture, read orientation, and support for semi-structured, multidimensional data, allows Druid to be the perfect database to power the current revolution in big data. By providing Druid to the the open source community, Metamarkets is contributing a critical, missing piece to the big data ecosystem.
Our platform is ready for business
Metamarkets has invested heavily in its platform so our customers don’t have to. Metamarkets supports a big data solution in order to protect clients and provide them with the following:
- Verification of data center capacity
- Acquisition and provisioning of new hardware
- Setup of a Hadoop cluster
- Written data processing scripts
- Installation of data warehousing appliances
- Provision of complete performance tuning
- Installation of desktop visualization tools
- Management of patches and upgrades
It is easier for customers to use our big data service which has taken care of the critical details. Clients no longer need to waste time or money on the daunting efforts of building their own big data solution.
The Stack
The Metamarkets analytics solution is built on a big data stack required for the processing, querying and visualization of high volume, high frequency event streams. How does Metamarkets do this?
- By our people The most critical feature of Metamarkets is the talent that develops and supports our product: we have assembled an incredible team of digital advertising experts, PhDs in statistical analysis, gurus in machine learning, experts of visualization and unrivaled professional service specialists.
- By our data pipes Our customized Hadoop pipeline for parallel data processing is a critical component of the Metamarkets analytic solution provided to its client base.
- By Druid A distributed, in-memory data engine which can slice, dice and drill data 1000 times faster than conventional, disk-backed databases.
- By DVL Dynamic Visualization LEGOs, a Javascript framework for interactive data visualizations.
- By utilizing an out-of-the-box modeling framework that generates critical analytic insights.
- By leveraging a cloud-based architecture which enables elastic scaling and lowers the total cost of ownership.
Metamarkets provides all of these unique capabilities in a single and easy to use deployment service that delivers a focused solution which is both affordable and scalable.
Speed at scale
Query speed is critical for real-time business analytics. Users want to ask questions, get answers and iterate as quickly as possible. This leads to the necessity of deeper data exploration and better understanding to support this fast turnaround of data. At the center of the Metamarkets solution is Druid, a distributed, in-memory, columnar database. Druid enables us to deliver a dynamic, interactive user experience across billions of records with faster response times.
While several vendors acclaim the use of in-memory database technology to support fast, ad-hoc querying, many of these tools are constrained by fixed hardware capacity. As data and usage increases, companies must acquire larger, more expensive, proprietary servers to maintain reasonable query performance.
Metamarkets can quickly grow and shrink capacity on-demand horizontally across many cloud instances. This enables us to deliver sub-second response times on sophisticated queries whether data is measured in gigabytes, terabytes, or even petabytes.
Smart performance
Druid is designed to operate in different configurations to optimize price for performance. New data can be stored in-memory while historical data can be memory-mapped. This way, recent data is always hot and ready to use, while historical data is available but at a lower cost.
Keeping it real-time
When other technology vendors talk about real-time access to new data, they typically mean one of two things:
- Data can be loaded immediately (to be processed later).
- Data can be processed immediately once loaded.
Usually, a client can access only one of these functions. At Metamarkets, clients can have both. Event data can be streamed live by aggregations and performing real-time calculations. If a client isn’t set up for the use of real-time data, Metamarkets’ professional services can also support batches of data of any size and also help clients on board to real-time data use.