Abstract: In one implementation, a system for multi-column statistic generation of a multi-dimensional tree includes a sampling engine to generate a sample of values from a set of data values, wherein the data values include multi-dimensional inputs. In addition, the system includes generating engine to generate a multi-dimensional tree utilizing the sample values. In addition, the system includes a filtering engine to determine a number of unique values at each node of the multi-dimensional tree utilizing a filter for the set of data values. Furthermore, the system includes an implementing engine to implement the multi-dimensional tree for a query of the set of data values.

Abstract: A method of managing a database including creating an initial counting bloom filter (CBF) instance having an array of counters and hash functions that map an inserted value to the array of counters, and designating the initial CBF instance as a current CBF instance, and sequentially inserting each value of a sample data set of a table column into the hash functions of the current CBF instance and incrementing counters of the array of counters to which the value is mapped. The method further includes, prior to inserting each value into the hash functions of the current CBF instance, when a number of counters of the array of counters having non-zero values is at least at a threshold level, designating the current CBF instance as an old CBF instance, creating a new CBF instance having an array of counters and hash functions that map an inserted value to the array counters, and designating the new CBF instance as the current CBF instance.

Abstract: Incrementally updating statistics includes sampling rows from a database column in a database to generate a first sample, sampling a subset of modified rows from the database column after generating the first sample to generate a second sample, determining whether distribution changes occurred to the database column based on the first and second samples, and updating a database statistic about the database column in response to determining that a distribution change exists.

Abstract: A method for improving database query classification includes reducing a predetermined plurality of features, generated by an optimizer, to a learned model of features by using a machine learning method. Classification is performed based on features of the query and features of operators executed by the query. The method also includes assigning an execution classification to a query based on the learned model of features. The execution classification is associated with a timeout threshold for execution of the query.

Abstract: A method of indexing in a skip list is disclosed. Key/value pairs are randomly inserted at an appropriate page in a skip list. A new page is created at the lowest level in the skip list. When creating the new page, the page is incremented to a higher level with a write probability. Reading the new page during a search. When reading the new page, the page is incremented to a higher level with a read probability. The read probability is not equal to the write probability.

Abstract: In one implementation, a system for multi-column statistic generation of a multi-dimensional tree includes a sampling engine to generate a sample of values from a set of data values, wherein the data values include multi-dimensional inputs. In addition, the system includes generating engine to generate a multi-dimensional tree utilizing the sample values. In addition, the system includes a filtering engine to determine a number of unique values at each node of the multi-dimensional tree utilizing a filter for the set of data values. Furthermore, the system includes an implementing engine to implement the multi-dimensional tree for a query of the set of data values.

Abstract: A method of managing a database including creating an initial counting bloom filter (CBF) instance having an array of counters and hash functions that map an inserted value to the array of counters, and designating the initial CBF instance as a current CBF instance, and sequentially inserting each value of a sample data set of a table column into the hash functions of the current CBF instance and incrementing counters of the array of counters to which the value is mapped. The method further includes, prior to inserting each value into the hash functions of the current CBF instance, when a number of counters of the array of counters having non-zero values is at least at a threshold level, designating the current CBF instance as an old CBF instance, creating a new CBF instance having an array of counters and hash functions that map an inserted value to the array counters, and designating the new CBF instance as the current CBF instance.

Abstract: A method of estimating a number of unique entry counts of an attribute in a database comprises, with a processor: identifying a sample of entries from an attribute database, determining frequencies of a number of input observations of the sample of entries, determining a number of high frequency values of the sample of entries, and estimating a number of unique entry counts of an attribute within the attribute database using a counting Bloom filter and based on the frequencies of the input observations and the high frequency values.

Abstract: Incrementally updating statistics includes sampling rows from a database column in a database to generate a first sample, sampling a subset of modified rows from the database column after generating the first sample to generate a second sample, determining whether distribution changes occurred to the database column based on the first and second samples, and updating a database statistic about the database column in response to determining that a distribution change exists.

Abstract: A method of estimating a number of unique entry counts of an attribute in a database comprises, with a processor: identifying a sample of entries from an attribute database, determining frequencies of a number of input observations of the sample of entries, determining a number of high frequency values of the sample of entries, and estimating a number of unique entry counts of an attribute within the attribute database using a counting Bloom filter and based on the frequencies of the input observations and the high frequency values.

Abstract: A method of indexing in a skip list is disclosed. Key/value pairs are randomly inserted at an appropriate page in a skip list. A new page is created at the lowest level in the skip list. When creating the new page, the page is incremented to a higher level with a write probability. Reading the new page during a search. When reading the new page, the page is incremented to a higher level with a read probability. The read probability is not equal to the write probability.

Abstract: Estimation of unique values in a database can be performed where a data field having multiple information values is provided in the database. The data field can be partitioned into multiple intervals such that each interval includes a range of information values. An interval specific Bloom filter can be calculated for each of the multiple intervals. A binary Bloom filter value can be calculated for an information value within an interval specific Bloom filter. The binary Bloom filter value can represent whether the information value is unique. A number of unique values in the database can be determined based on calculated binary Bloom filter values.

Abstract: The present disclosure relates to computing techniques. Data arrays are processed using Bloom filters to determine aggregate count, maximum, and minimum. These methods can be used on different types of data, including data groups, partial groups, data cubes, hypercubes, and others.

Abstract: The present disclosure generally relates to data processing. Bloom filters are used to process data at high speed. A Bloom filter that is initialized based on a source string can be used to quickly determine the similarity between the source string and a query string.

Abstract: One embodiment is a method that determines skew to predict costs for query plans for a database query. The method selects a query plan that accounts for the skew to evenly distribute processing load among multiple processors and then executes the query with the selected query plan.

Abstract: Estimation of unique values in a database can be performed where a data field having multiple information values is provided in the database. The data field can be partitioned into multiple intervals such that each interval includes a range of information values. An interval specific Bloom filter can be calculated for each of the multiple intervals. A binary Bloom filter value can be calculated for an information value within an interval specific Bloom filter. The binary Bloom filter value can represent whether the information value is unique. A number of unique values in the database can be determined based on calculated binary Bloom filter values.

Abstract: There is provided a computer-implemented method of executing a query. An exemplary method comprises optimizing the query. The query may specify multiple operations on corresponding multiple column sets of a table in a database. The exemplary method further comprises partitioning data of the table based on the operations within a single pass of the table. Additionally, the exemplary method comprises performing the operations on the partitioned data in parallel.

Abstract: Methods and systems receive a query, receive a control statement that represents at least a portion of a desired execution plan associated with the query, and produce an execution plan that conforms to the desired plan to the extent that the produced plan preserves the semantics of the query, without requiring an exhaustive search. These methods also enable realization of bushy shaped plans that are outside the scope of the optimizer search space.

Abstract: Methods and systems receive a query, receive a control statement that represents at least a portion of a desired execution plan associated with the query, and produce an execution plan that conforms to the desired plan to the extent that the produced plan preserves the semantics of the query, without requiring an exhaustive search. These methods also enable realization of bushy shaped plans that are outside the scope of the optimizer search space.