Dotazy
Fuel autonomous driving with OceanStor Pacific series storage.
• Upgraded autonomous driving models generate massive amounts of road data and at L4 — considered to be fully autonomous — there are three to five times more test data compared to L3: hundreds of TB need to be imported and stored every day.
• Of all collected data, approximately 15% needs to be pre-processed, with PBs of data that needs to be archived for over a decade.
• Test data transmission, import, preprocessing, training, simulation, and result analysis require different protocols, including object storage, Network Attached Storage (NAS), and Hadoop Distributed File System (HDFS).
• Diverse Data is radically siloed, so much so that copying data takes longer than processing and analyzing it, dragging down efficiency.
• There are various types of road test data sensors and complex service Input/Output (I/O) models. Extremely high performing Information Technology (IT) infrastructure is required for vehicle model algorithm training and function simulation.
• A single OceanStor Pacific 9550 chassis with two nodes accommodates road data collected by 500 vehicles.
• One copy of metadata and one copy of data can be accessed by multiple services, breaking down silos.
• As results a result, efficiency is improved by 25%, lowering Total Cost of Ownership (TCO) by 20%.
• A 5U chassis delivers bandwidth of 160 Gbit/s and 2 million IOPS.
• Performance density is 30% higher than the industry average.
