Detection of ransomware attack at object store
The technology disclosed herein provides batchwise metric-based malware detection by receiving a plurality of input/output (IO) requests, generating IO trace sequences by combining a number of IO requests, generating batchwise metrics, each of the batchwise metrics corresponding to one of the IO trace sequences and generated based on an operation performed on values of parameters within IO trace sequences, generating inferential model input feature vectors based on the generated batchwise metrics, and training an inferential model based on a plurality of the generated inferential model input feature vectors and ground truth values corresponding to the IO trace sequences, each of the ground truth values indicating whether a corresponding one of the IO trace sequences represents a ransomware attack.
1 . A method, comprising:
receiving a plurality of input/output (IO) requests at an object store;
generating IO trace sequences by combining a predetermined number of the plurality of IO requests;
generating batchwise metrics based on values of parameters within IO trace sequences corresponding to the batchwise metrics, each of the batchwise metrics corresponding to one of the IO trace sequences;
generating inferential model input feature vectors based on the generated batchwise metrics, each of the generated inferential model input feature vectors corresponding to one of the IO trace sequences; and
training an inferential model based on a plurality of the inferential model input feature vectors and ground truth values corresponding to the IO trace sequences, each of the ground truth values indicating whether a corresponding one of the IO trace sequences represents a ransomware attack.
2 . The method of claim 1 , wherein generating the batchwise metrics further comprising generating the batchwise metrics including a batch of data per request, the batch of data representing an amount of data upon which requests represented in one of the IO trace sequences are configured to operate.
3 . The method of claim 1 , wherein generating the batchwise metrics further comprising generating at least one of:
a batch read data size representing a size of data upon which read requests in one of the IO trace sequences are configured to operate; and
a batch write data size representing a size of data upon which write requests in one of the IO trace sequences are configured to operate.
4 . The method of claim 1 , wherein generating the batchwise metrics further comprising generating at least one of:
a batch read count representing a number of the predetermined number of the plurality of requests in one of the IO trace sequences that are read requests; and
a batch read count representing a number of the predetermined number of the plurality of requests in the one of the IO trace sequences that are write requests.
5 . The method of claim 1 , wherein generating the batchwise metrics further comprising generating a batch read/write comparison metric representing a comparison between read requests in a corresponding one of the IO trace sequences and write requests in the corresponding one of the IO trace sequences.
6 . The method of claim 5 , wherein generating the batch read/write comparison metric further comprising generating at least one of:
a read/write request ratio representing a ratio of data on which requests in one of the IO trace sequences including write operations are configured to operate data on which requests in the one of the IO trace sequences including read operations are configured to operate, and
a read/write parity metric representing whether the read/write operation ratio is within a predefined threshold difference from a value of one.
7 . The method of claim 1 , wherein generating the batchwise metrics further comprising generating a batch location metric representing a localization in memory of data upon which requests in one of the IO trace sequences are configured to operate.
8 . The method of claim 7 , wherein the localization in memory is based on sector identification metric values representing locations in memory of data upon which requests represented in the one of the IO trace sequences are configured to operate.
9 . The method of claim 1 , wherein generating batchwise metrics based on values of parameters within IO trace sequences further comprising determining an average or a median of the values of parameters within IO trace sequences.
10 . The method of claim 1 , wherein generating batchwise metrics based on values of parameters within IO trace sequences further comprising determining a variance or a standard deviation of the values of parameters within IO trace sequences.
11 . In a computing environment, a method performed at least in part on at least one processor, the method comprising:
receiving a plurality of input/output (IO) requests at an object store;
generating IO trace sequences by combining a predetermined number of the plurality of IO requests;
generating batchwise metrics based on values of parameters within IO trace sequences corresponding to the batchwise metrics, each of the batchwise metrics corresponding to one of the IO trace sequences;
generating inferential model input feature vectors based on the generated batchwise metrics, each of the generated inferential model input feature vectors corresponding to one of the IO trace sequences; and
training an inferential model based on a plurality of the inferential model input feature vectors and ground truth values corresponding to the IO trace sequences, each of the ground truth values indicating whether a corresponding one of the IO trace sequences represents a ransomware attack.
12 . The method of claim 11 , wherein generating the batchwise metrics further comprising generating the batchwise metrics including a batch of data per request, the batch of data representing an amount of data upon which requests represented in one of the IO trace sequences are configured to operate.
13 . The method of claim 11 , wherein generating the batchwise metrics further comprising generating a batch read/write comparison metric representing a comparison between read requests in a corresponding one of the IO trace sequences and write requests in the corresponding one of the IO trace sequences.
14 . The method of claim 13 , wherein generating the batch read/write comparison metric further comprising generating at least one of:
a read/write request ratio representing a ratio of data on which requests in one of the IO trace sequences including write operations are configured to operate data on which requests in the one of the IO trace sequences including read operations are configured to operate, and
a read/write parity metric representing whether the read/write operation ratio is within a predefined threshold difference from a value of one.
15 . The method of claim 11 , wherein generating the batchwise metrics further comprising generating a batch location metric representing a localization in memory of data upon which requests in one of the IO trace sequences are configured to operate and wherein the localization in memory is based on sector identification metric values representing locations in memory of data upon which requests represented in the one of the IO trace sequences are configured to operate.
16 . One or more non-transitory computer-readable storage media encoding computer-executable instructions for executing on a computer system a computer process, the computer process comprising:
receiving a plurality of input/output (IO) requests at an object store;
generating IO trace sequences by combining a predetermined number of the plurality of IO requests;
generating batchwise metrics based on values of parameters within IO trace sequences corresponding to the batchwise metrics, each of the batchwise metrics corresponding to one of the IO trace sequences;
generating inferential model input feature vectors based on the generated batchwise metrics, each of the generated inferential model input feature vectors corresponding to one of the IO trace sequences; and
training an inferential model based on a plurality of the inferential model input feature vectors and ground truth values corresponding to the IO trace sequences, each of the ground truth values indicating whether a corresponding one of the IO trace sequences represents a ransomware attack.
17 . One or more tangible computer-readable storage media of claim 16 , wherein generating the batchwise metrics further comprising generating the batchwise metrics including a batch of data per request, the batch of data representing an amount of data upon which requests represented in one of the IO trace sequences are configured to operate.
18 . One or more tangible computer-readable storage media of claim 16 , wherein generating the batchwise metrics further comprising generating a batch read data size representing a size of data upon which read requests in one of the IO trace sequences are configured to operate.
19 . One or more tangible computer-readable storage media of claim 16 , wherein generating the batchwise metrics further comprising generating a batch write data size representing a size of data upon which write requests in one of the IO trace sequences are configured to operate.
20 . One or more tangible computer-readable storage media of claim 16 , wherein generating the batchwise metrics further comprising generating a batch read count representing a number of the predetermined number of the plurality of requests in one of the IO trace sequences that are read requests.