Abstract: A data service periodically monitors parameters of a computing system and periodically transmits metrics related to the monitored parameters to an AI service over a secure channel. The AI service uses a Time Series Classification model trained on training data that results from simulated row-hammer attacks to a computing system. The AI service processes the metrics using the trained Time Series Classification model. If the Time Series Classification model determines that the received metrics correspond to a row-hammer attack, the AI services transmits an attack confirmation message to the computing the data service. A remediation action is implemented responsive to receiving the attack confirmation message.

Abstract: Boot integrity of a storage platform is verified by comparing observed boot data values with expected boot data values stored in a secure remote cloud. The boot data values include hashes of software that runs at each stage of a boot sequence, e.g., BIOS, bootloader, kernel, and runlevel programs. The observed boot data values may be provided by a TPM using an AIK and nonce. If the observed boot data values fail to match the expected boot data values then a boot integrity service running on the storage platform limits functionality such as by disabling IO services, disabling remote data replication, enabling a diagnostic service, enabling a data collection service, disabling access by non-service accounts, and protecting a management database.

Abstract: A topology service may include a topology database and topology service interface used by clients to query the topology database. The topology database may describe a topology of resources of an infrastructure that may include racks of resources such as network, compute and storage resources. The topology database may be populated using discovery information returned as a result of issuing discovery requests to switches. The discovery information may identify connectivity between switches and other resources. The discovery information may be processed to determine assigned switch roles and additional derived information such as regarding instances of topological entities in the infrastructure. The derived information may identify instances of racks, rows and data centers in the infrastructure. The topology database may identify particular resources included in the different racks, rows, and data centers as well as connectivity between resources and communication capabilities of resource interfaces.