US 12,475,216 B2
Cloud solution for rowhammer detection
Jean Pierre, Brockton, MA (US); Azzam Tannous, Cambridge, MA (US); Jochen De Smet, Shrewsbury, MA (US); Ananthanarayanan Balachandran, Framingham, MA (US); Huijun Xie, Hopkinton, MA (US); and Massarrah Tannous, Cambridge, MA (US)
Assigned to Dell Products L.P., Round Rock, TX (US)
Filed by Dell Products, L.P., Round Rock, TX (US)
Filed on Apr. 22, 2022, as Appl. No. 17/727,541.
Prior Publication US 2023/0342454 A1, Oct. 26, 2023
Int. Cl. G06F 21/55 (2013.01)
CPC G06F 21/554 (2013.01) 20 Claims
OG exemplary drawing
 
1. A method comprising:
generating, by a data services computing system comprising at least one processor, metrics associated with at least one enterprise computing system other than the data services computing system, wherein the data services computing system and the at least one enterprise computing system are distinct computing systems coupled by a network, wherein the at least one enterprise computing system is not part of the data services computing system, and wherein the data services computing system is not part of the at least one enterprise computing system;
transmitting, by the data services computing system via the network, the metrics to an artificial intelligence (AI) service, wherein the AI service is not part of, and is not executed by, the data services computing system;
receiving, by the data services computing system from the AI service via the network, an attack confirmation message that a memory of at least one of the at least one enterprise computing system has been subjected to a row-hammer attack, wherein the attack confirmation message was generated based on a determination that the metrics correspond to at least one row-hammer attack characteristic; and
responsive to the attack confirmation message, initiating, by the data services computing system, a remediation action with respect to the row-hammer attack to reduce at least one effect of the row-hammer attack at the at least one of the at least one enterprise computing system,
wherein the AI service comprises a time series classification model trained using labeled data that corresponds to, based on previous human prediction data, the at least one row-hammer attack characteristic that is threshold likely to correspond to the at least one effect to a processor or to the memory of the at least one of the at least one enterprise computing system being caused by the row-hammer attack.