GPU-powered database pioneer MapD Technologies today announced that customers can now get its software through the Google Cloud Launcher, the online marketplace for Google Cloud Platform. The news comes on the heals of a major version 4.0 launch, which adds more geospatial capabilities.
Although the majority of users run on-premise, MapD Technologies is no stranger to the cloud. The San Francisco company has offered its software on Amazon Web Services for some time. And while this is not the first time that MapD has been used on GCP, making the software available through Google’s marketplace lowers the barrier to entry for prospective customers, according to Todd Mostak, MapD’s
“We have customer deploying on GCP, but they do that via an enterprise build,” Mostak tells Datanami. “This will be in Google Cloud Launcher, and it just allows your average user to get access to us easily without going through an enterprise sales process.”
In terms of support for GPUs, Mostak says GCP offers the full gamut of Nvidia processor, including the Kepler-based K80, the Pascal-based P100, and the Volta-based V100 GPUs. AWS, by comparison, doesn’t offer the Pascal card, he says.
“Not only that, but GCP offers a lot of flexibility in how you can combine with CPU resources,” Mostak adds. “That’s a boon for a lot of their customers. We can tailor the instance a little better for the right CPU-core, CPU-RAM, and GPU ratio.”
What’s more, GCP offers more fine-grained pricing options than AWS. While most of MapD customers leave their GPU database environments up continuously, the capability to rent GPUs by the minute (or even possibly down to thesecond) could allows them to better match their GPU requirements with their budget needs.
In 2013, Mostak spun MapD out of an MIT CSAIL research project to focus on high-end analytics. The MapD platform, which can also run on CPUs, also includes an interactive visual layer. Together, these elements allow customers to leverage supercomputer processing power to deliver SQL-powered interactive analytics upon hundreds of billions of pieces of data.
The company, which has raised $37.1 million in venture funding, has found a sweet spot in industries that have traditionally struggled to make sense of large data sets, including telecommunications, financial services, hedge funds, governmental agencies, automotive companies, and oil and gas firms.
In many cases, MapD complements or replaces previously built analytic system based on massively parallel processing (MPP) column-oriented databases, such as Oracle Exadata, Teradata, and IBM‘s Netezza, as well as newer data lakes running Hadoop, Hive, Spark, and related technology. While the performance of these solutions wasn’t horrible, Mostak says the volume and velocity of data posed big challenges for analytic endevors.
“Verizon told us their queries could take hours sometimes across tens of millions of records, trying to find the needle in the haystack of why a call is being dropped. Is it a bad cell tower, or a handset with a broken firmware update?” Mostak says. “When they got MapD, all of a sudden those queries shot up to sub-second response, or a within a few seconds. And not only that, they have the integrated visualization where they can see all this data on a map.”
MapD recently released 4.0 of its eponymous database, which brought a host of new enterprise features like role-based access control and full create, read, update, and delete (CRUD) capabilities. Previously, MapD was an append-only database.
But the biggest features in 4.0 were the new geo-spatial capabilities. About 40% of MapD’s existing customer base used the database for geospatial analytics, but the customers were pushing for more even more geo-spatial capabilities, Mostak says. So the company responded by incorporating a range of Open Geospatial Consortium data types, including point types, line types, polygon typos, and multi-polygon types – as well as their associated operators – into the MapD product.
This release should make it easier for customers to take advantage of geospatial analytics, MapD Vice President of Product Management Venkat Krishnamurthy wrote on the MapD blog.
“With MapD 4.0, we’re taking the first big step toward making geospatial analytics available to any user, from an experienced GIS analyst or data scientist who can write complex SQL queries over location enriched data to a business user of MapD Immerse who wants to go a step beyond mainstream BI tools to derive insight quickly and visually from the same data,” he wrote.