Follow Datanami:
August 11, 2017

Huawei Adds MapD GPUs to Latest Servers


Data analytics and visualization specialist MapD Technologies has inked a deal with Huawei, the multinational networking and telecommunications giant, to juice the Chinese vendor’s server hardware with scalable GPU-supported analytics.

The partners announced an agreement this week intended to improve compatibility between Huawei’s latest FusionServer and MapD database and visualization platforms. Huawei (SHE: 002502) announced earlier this summer it was converting its FusionServer line to Intel’s (NASDAQ: INTC) Xeon Scalable Processor.

Among the anticipated outcomes of the R&D initiative is improving hardware and software integration between servers and analytics while supporting more third-party applications that increasingly include analytics.

San Francisco-based MapD said Thursday (Aug. 10) the deal allows the Chinese telecom giant to sell it analytics platform in the nascent Chinese server market. Huawei also will roll out MapD’s flagship database and visual analytics client on its G series FusionServer described as a heterogeneous server built around 32 GPUs that can serve as the basis of a supercomputing cluster.

The latest generation Huawei server is reportedly optimized for cloud computing and network function virtualization along with HPC.

But the key focus of the agreement is GPU acceleration of analytics, especially in the Asian server market that is among the few expanding regions for hardware vendors. “We are seeing stronger and stronger demand for real-time data queries, analysis and visualization of hundreds of billions or trillions of rows of machine-generated logs,” Qiu Long, president of Huawei’s IT server product line, noted in a statement announcing the analytics agreement.

“Achieving ad-hoc data analytics at this scale in real-time is extremely challenging,” the Huawei executive added. “We believe that GPU acceleration will play a pivotal role in the future of analytics.”

The agreement also would help Huawei differentiate its server line in the Asian market where it competes head-on with Lenovo, which acquired IBM’s server business in 2014 for $2.1 billion. Huawei, Lenovo (HKG: 0992) and other server makers are targeting the Asian IT and analytics markets as key growth regions as server sales slump elsewhere.

The Huawei deal also underscores the steady shift to converged GPU-based analytics and accelerated access to databases. MapD released the latest version of its Core database and Immerse visual analytics tools this spring, both reemphasizing GPU acceleration as a means of scaling data volumes while boosting query performance.

The analytics startup touts its distributed architecture as enabling larger data volumes and faster queries across multiple servers. Huawei and other server makers are moving quickly to incorporate GPU acceleration into servers processing ever-larger data volumes as a way to speed up searches.

MapD CEO Todd Mostak noted that the partnership with Huawei would help the Chinese server vendor deliver “a linearly scalable analytics platform [that] runs natively on [its] hardware.” MapD’s approach also compresses data to create multiple billions of rows in the GPU memory.

Recent items:

MapD Upgrades GPU-Based Database Performance

Why America’s Spy Agencies Are Investing in MapD