ajantha-bhat commented on a change in pull request #2814: [CARBONDATA-3001] configurable page size in MB
URL: https://github.com/apache/carbondata/pull/2814#discussion_r275268381 ########## File path: processing/src/main/java/org/apache/carbondata/processing/store/CarbonFactDataHandlerColumnar.java ########## @@ -227,50 +268,120 @@ public void addDataToStore(CarbonRow row) throws CarbonDataWriterException { /** * Check if column page can be added more rows after adding this row to page. + * only few no-dictionary dimensions columns (string, varchar, + * complex columns) can grow huge in size. * - * A varchar column page uses SafeVarLengthColumnPage/UnsafeVarLengthColumnPage to store data - * and encoded using HighCardDictDimensionIndexCodec which will call getByteArrayPage() from - * column page and flatten into byte[] for compression. - * Limited by the index of array, we can only put number of Integer.MAX_VALUE bytes in a page. - * - * Another limitation is from Compressor. Currently we use snappy as default compressor, - * and it will call MaxCompressedLength method to estimate the result size for preparing output. - * For safety, the estimate result is oversize: `32 + source_len + source_len/6`. - * So the maximum bytes to compress by snappy is (2GB-32)*6/7≈1.71GB. * - * Size of a row does not exceed 2MB since UnsafeSortDataRows uses 2MB byte[] as rowBuffer. - * Such that we can stop adding more row here if any long string column reach this limit. - * - * If use unsafe column page, please ensure the memory configured is enough. - * @param row - * @return false if any varchar column page cannot add one more value(2MB) + * @param row carbonRow + * @return false if next rows can be added to same page. + * true if next rows cannot be added to same page */ - private boolean isVarcharColumnFull(CarbonRow row) { - //TODO: test and remove this as now UnsafeSortDataRows can exceed 2MB - if (model.getVarcharDimIdxInNoDict().size() > 0) { + private boolean needToCutThePage(CarbonRow row) { + List<DataType> noDictDataTypesList = model.getNoDictDataTypesList(); + int totalNoDictPageCount = noDictDataTypesList.size() + model.getNoDictAllComplexColumnDepth(); + if (totalNoDictPageCount > 0) { + int currentElementLength; + int bucketCounter = 0; + if (configuredPageSizeInBytes == 0) { + // no need to cut the page + // use default value + /*configuredPageSizeInBytes = + CarbonCommonConstants.TABLE_PAGE_SIZE_INMB_DEFAULT * 1024 * 1024;*/ + return false; + } Object[] nonDictArray = WriteStepRowUtil.getNoDictAndComplexDimension(row); - for (int i = 0; i < model.getVarcharDimIdxInNoDict().size(); i++) { - if (DataTypeUtil - .isPrimitiveColumn(model.getNoDictAndComplexColumns()[i].getDataType())) { - // get the size from the data type - varcharColumnSizeInByte[i] += - model.getNoDictAndComplexColumns()[i].getDataType().getSizeInBytes(); - } else { - varcharColumnSizeInByte[i] += - ((byte[]) nonDictArray[model.getVarcharDimIdxInNoDict().get(i)]).length; - } - if (SnappyCompressor.MAX_BYTE_TO_COMPRESS - - (varcharColumnSizeInByte[i] + dataRows.size() * 4) < (2 << 20)) { - LOGGER.debug("Limited by varchar column, page size is " + dataRows.size()); - // re-init for next page - varcharColumnSizeInByte = new int[model.getVarcharDimIdxInNoDict().size()]; - return true; + for (int i = 0; i < noDictDataTypesList.size(); i++) { + DataType columnType = noDictDataTypesList.get(i); + if ((columnType == DataTypes.STRING) || (columnType == DataTypes.VARCHAR)) { + currentElementLength = ((byte[]) nonDictArray[i]).length; + noDictColumnPageSize[bucketCounter] += currentElementLength; + canSnappyHandleThisRow(noDictColumnPageSize[bucketCounter]); + // If current page size is more than configured page size, cut the page here. + if (noDictColumnPageSize[bucketCounter] + dataRows.size() * 4 + >= configuredPageSizeInBytes) { + if (LOGGER.isDebugEnabled()) { + LOGGER.debug("cutting the page. Rows count in this page: " + dataRows.size()); + } + // re-init for next page + noDictColumnPageSize = new int[totalNoDictPageCount]; + return true; + } + bucketCounter++; + } else if (columnType.isComplexType()) { + // this is for depth of each complex column, model is having only total depth. + GenericDataType genericDataType = complexIndexMapCopy + .get(i - model.getNoDictionaryCount() + model.getPrimitiveDimLens().length); + int depth = genericDataType.getDepth(); + List<ArrayList<byte[]>> flatComplexColumnList = (List<ArrayList<byte[]>>) nonDictArray[i]; + for (int k = 0; k < depth; k++) { + ArrayList<byte[]> children = flatComplexColumnList.get(k); + // Add child element from inner list. + int complexElementSize = 0; + for (byte[] child : children) { + complexElementSize += child.length; + } + noDictColumnPageSize[bucketCounter] += complexElementSize; + canSnappyHandleThisRow(noDictColumnPageSize[bucketCounter]); + // If current page size is more than configured page size, cut the page here. + if (noDictColumnPageSize[bucketCounter] + dataRows.size() * 4 + >= configuredPageSizeInBytes) { + LOGGER.info("cutting the page. Rows count: " + dataRows.size()); + // re-init for next page + noDictColumnPageSize = new int[totalNoDictPageCount]; + return true; + } + bucketCounter++; + } } } } return false; } + private int setFlatCarbonRowForComplex(CarbonRow row) { Review comment: WriteStepRowUtil is in core module, this is in processing. Also This code logic is from TablePage, so it is coupled with model and GenericDataType. We can keep it here only for now. ---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [hidden email] With regards, Apache Git Services |
Free forum by Nabble | Edit this page |