View Javadoc

1   /*
2    *
3    * Licensed to the Apache Software Foundation (ASF) under one
4    * or more contributor license agreements.  See the NOTICE file
5    * distributed with this work for additional information
6    * regarding copyright ownership.  The ASF licenses this file
7    * to you under the Apache License, Version 2.0 (the
8    * "License"); you may not use this file except in compliance
9    * with the License.  You may obtain a copy of the License at
10   *
11   *     http://www.apache.org/licenses/LICENSE-2.0
12   *
13   * Unless required by applicable law or agreed to in writing, software
14   * distributed under the License is distributed on an "AS IS" BASIS,
15   * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
16   * See the License for the specific language governing permissions and
17   * limitations under the License.
18   */
19  
20  package org.apache.hadoop.hbase.client;
21  
22  import java.io.IOException;
23  import java.util.ArrayList;
24  import java.util.HashMap;
25  import java.util.List;
26  import java.util.Map;
27  import java.util.NavigableSet;
28  import java.util.TreeMap;
29  import java.util.TreeSet;
30  
31  import org.apache.commons.logging.Log;
32  import org.apache.commons.logging.LogFactory;
33  import org.apache.hadoop.hbase.classification.InterfaceAudience;
34  import org.apache.hadoop.hbase.classification.InterfaceStability;
35  import org.apache.hadoop.hbase.HConstants;
36  import org.apache.hadoop.hbase.filter.Filter;
37  import org.apache.hadoop.hbase.filter.IncompatibleFilterException;
38  import org.apache.hadoop.hbase.io.TimeRange;
39  import org.apache.hadoop.hbase.util.Bytes;
40  
41  /**
42   * Used to perform Scan operations.
43   * <p>
44   * All operations are identical to {@link Get} with the exception of
45   * instantiation.  Rather than specifying a single row, an optional startRow
46   * and stopRow may be defined.  If rows are not specified, the Scanner will
47   * iterate over all rows.
48   * <p>
49   * To scan everything for each row, instantiate a Scan object.
50   * <p>
51   * To modify scanner caching for just this scan, use {@link #setCaching(int) setCaching}.
52   * If caching is NOT set, we will use the caching value of the hosting {@link HTable}.  See
53   * {@link HTable#setScannerCaching(int)}. In addition to row caching, it is possible to specify a
54   * maximum result size, using {@link #setMaxResultSize(long)}. When both are used,
55   * single server requests are limited by either number of rows or maximum result size, whichever
56   * limit comes first.
57   * <p>
58   * To further define the scope of what to get when scanning, perform additional
59   * methods as outlined below.
60   * <p>
61   * To get all columns from specific families, execute {@link #addFamily(byte[]) addFamily}
62   * for each family to retrieve.
63   * <p>
64   * To get specific columns, execute {@link #addColumn(byte[], byte[]) addColumn}
65   * for each column to retrieve.
66   * <p>
67   * To only retrieve columns within a specific range of version timestamps,
68   * execute {@link #setTimeRange(long, long) setTimeRange}.
69   * <p>
70   * To only retrieve columns with a specific timestamp, execute
71   * {@link #setTimeStamp(long) setTimestamp}.
72   * <p>
73   * To limit the number of versions of each column to be returned, execute
74   * {@link #setMaxVersions(int) setMaxVersions}.
75   * <p>
76   * To limit the maximum number of values returned for each call to next(),
77   * execute {@link #setBatch(int) setBatch}.
78   * <p>
79   * To add a filter, execute {@link #setFilter(org.apache.hadoop.hbase.filter.Filter) setFilter}.
80   * <p>
81   * Expert: To explicitly disable server-side block caching for this scan,
82   * execute {@link #setCacheBlocks(boolean)}.
83   */
84  @InterfaceAudience.Public
85  @InterfaceStability.Stable
86  public class Scan extends Query {
87    private static final Log LOG = LogFactory.getLog(Scan.class);
88  
89    private static final String RAW_ATTR = "_raw_";
90  
91    /**
92     * EXPERT ONLY.
93     * An integer (not long) indicating to the scanner logic how many times we attempt to retrieve the
94     * next KV before we schedule a reseek.
95     * The right value depends on the size of the average KV. A reseek is more efficient when
96     * it can skip 5-10 KVs or 512B-1KB, or when the next KV is likely found in another HFile block.
97     * Setting this only has any effect when columns were added with
98     * {@link #addColumn(byte[], byte[])}
99     * <pre>{@code
100    * Scan s = new Scan(...);
101    * s.addColumn(...);
102    * s.setAttribute(Scan.HINT_LOOKAHEAD, Bytes.toBytes(2));
103    * }</pre>
104    * Default is 0 (always reseek).
105    * @deprecated without replacement
106    *             This is now a no-op, SEEKs and SKIPs are optimizated automatically.
107    */
108   @Deprecated
109   public static final String HINT_LOOKAHEAD = "_look_ahead_";
110 
111   private byte [] startRow = HConstants.EMPTY_START_ROW;
112   private byte [] stopRow  = HConstants.EMPTY_END_ROW;
113   private int maxVersions = 1;
114   private int batch = -1;
115 
116   private int storeLimit = -1;
117   private int storeOffset = 0;
118   private boolean getScan;
119 
120   // If application wants to collect scan metrics, it needs to
121   // call scan.setAttribute(SCAN_ATTRIBUTES_ENABLE, Bytes.toBytes(Boolean.TRUE))
122   static public final String SCAN_ATTRIBUTES_METRICS_ENABLE = "scan.attributes.metrics.enable";
123   static public final String SCAN_ATTRIBUTES_METRICS_DATA = "scan.attributes.metrics.data";
124   
125   // If an application wants to use multiple scans over different tables each scan must
126   // define this attribute with the appropriate table name by calling
127   // scan.setAttribute(Scan.SCAN_ATTRIBUTES_TABLE_NAME, Bytes.toBytes(tableName))
128   static public final String SCAN_ATTRIBUTES_TABLE_NAME = "scan.attributes.table.name";
129   
130   /*
131    * -1 means no caching
132    */
133   private int caching = -1;
134   private long maxResultSize = -1;
135   private boolean cacheBlocks = true;
136   private boolean reversed = false;
137   private TimeRange tr = new TimeRange();
138   private Map<byte [], NavigableSet<byte []>> familyMap =
139     new TreeMap<byte [], NavigableSet<byte []>>(Bytes.BYTES_COMPARATOR);
140   private Boolean loadColumnFamiliesOnDemand = null;
141 
142   /**
143    * Set it true for small scan to get better performance
144    * 
145    * Small scan should use pread and big scan can use seek + read
146    * 
147    * seek + read is fast but can cause two problem (1) resource contention (2)
148    * cause too much network io
149    * 
150    * [89-fb] Using pread for non-compaction read request
151    * https://issues.apache.org/jira/browse/HBASE-7266
152    * 
153    * On the other hand, if setting it true, we would do
154    * openScanner,next,closeScanner in one RPC call. It means the better
155    * performance for small scan. [HBASE-9488].
156    * 
157    * Generally, if the scan range is within one data block(64KB), it could be
158    * considered as a small scan.
159    */
160   private boolean small = false;
161 
162   /**
163    * Create a Scan operation across all rows.
164    */
165   public Scan() {}
166 
167   public Scan(byte [] startRow, Filter filter) {
168     this(startRow);
169     this.filter = filter;
170   }
171 
172   /**
173    * Create a Scan operation starting at the specified row.
174    * <p>
175    * If the specified row does not exist, the Scanner will start from the
176    * next closest row after the specified row.
177    * @param startRow row to start scanner at or after
178    */
179   public Scan(byte [] startRow) {
180     this.startRow = startRow;
181   }
182 
183   /**
184    * Create a Scan operation for the range of rows specified.
185    * @param startRow row to start scanner at or after (inclusive)
186    * @param stopRow row to stop scanner before (exclusive)
187    */
188   public Scan(byte [] startRow, byte [] stopRow) {
189     this.startRow = startRow;
190     this.stopRow = stopRow;
191     //if the startRow and stopRow both are empty, it is not a Get
192     this.getScan = isStartRowAndEqualsStopRow();
193   }
194 
195   /**
196    * Creates a new instance of this class while copying all values.
197    *
198    * @param scan  The scan instance to copy from.
199    * @throws IOException When copying the values fails.
200    */
201   public Scan(Scan scan) throws IOException {
202     startRow = scan.getStartRow();
203     stopRow  = scan.getStopRow();
204     maxVersions = scan.getMaxVersions();
205     batch = scan.getBatch();
206     storeLimit = scan.getMaxResultsPerColumnFamily();
207     storeOffset = scan.getRowOffsetPerColumnFamily();
208     caching = scan.getCaching();
209     maxResultSize = scan.getMaxResultSize();
210     cacheBlocks = scan.getCacheBlocks();
211     getScan = scan.isGetScan();
212     filter = scan.getFilter(); // clone?
213     loadColumnFamiliesOnDemand = scan.getLoadColumnFamiliesOnDemandValue();
214     TimeRange ctr = scan.getTimeRange();
215     tr = new TimeRange(ctr.getMin(), ctr.getMax());
216     reversed = scan.isReversed();
217     small = scan.isSmall();
218     Map<byte[], NavigableSet<byte[]>> fams = scan.getFamilyMap();
219     for (Map.Entry<byte[],NavigableSet<byte[]>> entry : fams.entrySet()) {
220       byte [] fam = entry.getKey();
221       NavigableSet<byte[]> cols = entry.getValue();
222       if (cols != null && cols.size() > 0) {
223         for (byte[] col : cols) {
224           addColumn(fam, col);
225         }
226       } else {
227         addFamily(fam);
228       }
229     }
230     for (Map.Entry<String, byte[]> attr : scan.getAttributesMap().entrySet()) {
231       setAttribute(attr.getKey(), attr.getValue());
232     }
233   }
234 
235   /**
236    * Builds a scan object with the same specs as get.
237    * @param get get to model scan after
238    */
239   public Scan(Get get) {
240     this.startRow = get.getRow();
241     this.stopRow = get.getRow();
242     this.filter = get.getFilter();
243     this.cacheBlocks = get.getCacheBlocks();
244     this.maxVersions = get.getMaxVersions();
245     this.storeLimit = get.getMaxResultsPerColumnFamily();
246     this.storeOffset = get.getRowOffsetPerColumnFamily();
247     this.tr = get.getTimeRange();
248     this.familyMap = get.getFamilyMap();
249     this.getScan = true;
250     for (Map.Entry<String, byte[]> attr : get.getAttributesMap().entrySet()) {
251       setAttribute(attr.getKey(), attr.getValue());
252     }
253   }
254 
255   public boolean isGetScan() {
256     return this.getScan || isStartRowAndEqualsStopRow();
257   }
258 
259   private boolean isStartRowAndEqualsStopRow() {
260     return this.startRow != null && this.startRow.length > 0 &&
261         Bytes.equals(this.startRow, this.stopRow);
262   }
263   /**
264    * Get all columns from the specified family.
265    * <p>
266    * Overrides previous calls to addColumn for this family.
267    * @param family family name
268    * @return this
269    */
270   public Scan addFamily(byte [] family) {
271     familyMap.remove(family);
272     familyMap.put(family, null);
273     return this;
274   }
275 
276   /**
277    * Get the column from the specified family with the specified qualifier.
278    * <p>
279    * Overrides previous calls to addFamily for this family.
280    * @param family family name
281    * @param qualifier column qualifier
282    * @return this
283    */
284   public Scan addColumn(byte [] family, byte [] qualifier) {
285     NavigableSet<byte []> set = familyMap.get(family);
286     if(set == null) {
287       set = new TreeSet<byte []>(Bytes.BYTES_COMPARATOR);
288     }
289     if (qualifier == null) {
290       qualifier = HConstants.EMPTY_BYTE_ARRAY;
291     }
292     set.add(qualifier);
293     familyMap.put(family, set);
294     return this;
295   }
296 
297   /**
298    * Get versions of columns only within the specified timestamp range,
299    * [minStamp, maxStamp).  Note, default maximum versions to return is 1.  If
300    * your time range spans more than one version and you want all versions
301    * returned, up the number of versions beyond the defaut.
302    * @param minStamp minimum timestamp value, inclusive
303    * @param maxStamp maximum timestamp value, exclusive
304    * @throws IOException if invalid time range
305    * @see #setMaxVersions()
306    * @see #setMaxVersions(int)
307    * @return this
308    */
309   public Scan setTimeRange(long minStamp, long maxStamp)
310   throws IOException {
311     tr = new TimeRange(minStamp, maxStamp);
312     return this;
313   }
314 
315   /**
316    * Get versions of columns with the specified timestamp. Note, default maximum
317    * versions to return is 1.  If your time range spans more than one version
318    * and you want all versions returned, up the number of versions beyond the
319    * defaut.
320    * @param timestamp version timestamp
321    * @see #setMaxVersions()
322    * @see #setMaxVersions(int)
323    * @return this
324    */
325   public Scan setTimeStamp(long timestamp)
326   throws IOException {
327     try {
328       tr = new TimeRange(timestamp, timestamp+1);
329     } catch(IOException e) {
330       // This should never happen, unless integer overflow or something extremely wrong...
331       LOG.error("TimeRange failed, likely caused by integer overflow. ", e);
332       throw e;
333     }
334     return this;
335   }
336 
337   /**
338    * Set the start row of the scan.
339    * @param startRow row to start scan on (inclusive)
340    * Note: In order to make startRow exclusive add a trailing 0 byte
341    * @return this
342    */
343   public Scan setStartRow(byte [] startRow) {
344     this.startRow = startRow;
345     return this;
346   }
347 
348   /**
349    * Set the stop row.
350    * @param stopRow row to end at (exclusive)
351    * Note: In order to make stopRow inclusive add a trailing 0 byte
352    * @return this
353    */
354   public Scan setStopRow(byte [] stopRow) {
355     this.stopRow = stopRow;
356     return this;
357   }
358 
359   /**
360    * Get all available versions.
361    * @return this
362    */
363   public Scan setMaxVersions() {
364     this.maxVersions = Integer.MAX_VALUE;
365     return this;
366   }
367 
368   /**
369    * Get up to the specified number of versions of each column.
370    * @param maxVersions maximum versions for each column
371    * @return this
372    */
373   public Scan setMaxVersions(int maxVersions) {
374     this.maxVersions = maxVersions;
375     return this;
376   }
377 
378   /**
379    * Set the maximum number of values to return for each call to next()
380    * @param batch the maximum number of values
381    */
382   public void setBatch(int batch) {
383     if (this.hasFilter() && this.filter.hasFilterRow()) {
384       throw new IncompatibleFilterException(
385         "Cannot set batch on a scan using a filter" +
386         " that returns true for filter.hasFilterRow");
387     }
388     this.batch = batch;
389   }
390 
391   /**
392    * Set the maximum number of values to return per row per Column Family
393    * @param limit the maximum number of values returned / row / CF
394    */
395   public void setMaxResultsPerColumnFamily(int limit) {
396     this.storeLimit = limit;
397   }
398 
399   /**
400    * Set offset for the row per Column Family.
401    * @param offset is the number of kvs that will be skipped.
402    */
403   public void setRowOffsetPerColumnFamily(int offset) {
404     this.storeOffset = offset;
405   }
406 
407   /**
408    * Set the number of rows for caching that will be passed to scanners.
409    * If not set, the default setting from {@link HTable#getScannerCaching()} will apply.
410    * Higher caching values will enable faster scanners but will use more memory.
411    * @param caching the number of rows for caching
412    */
413   public void setCaching(int caching) {
414     this.caching = caching;
415   }
416 
417   /**
418    * @return the maximum result size in bytes. See {@link #setMaxResultSize(long)}
419    */
420   public long getMaxResultSize() {
421     return maxResultSize;
422   }
423 
424   /**
425    * Set the maximum result size. The default is -1; this means that no specific
426    * maximum result size will be set for this scan, and the global configured
427    * value will be used instead. (Defaults to unlimited).
428    *
429    * @param maxResultSize The maximum result size in bytes.
430    */
431   public void setMaxResultSize(long maxResultSize) {
432     this.maxResultSize = maxResultSize;
433   }
434 
435   @Override
436   public Scan setFilter(Filter filter) {
437     super.setFilter(filter);
438     return this;
439   }
440 
441   /**
442    * Setting the familyMap
443    * @param familyMap map of family to qualifier
444    * @return this
445    */
446   public Scan setFamilyMap(Map<byte [], NavigableSet<byte []>> familyMap) {
447     this.familyMap = familyMap;
448     return this;
449   }
450 
451   /**
452    * Getting the familyMap
453    * @return familyMap
454    */
455   public Map<byte [], NavigableSet<byte []>> getFamilyMap() {
456     return this.familyMap;
457   }
458 
459   /**
460    * @return the number of families in familyMap
461    */
462   public int numFamilies() {
463     if(hasFamilies()) {
464       return this.familyMap.size();
465     }
466     return 0;
467   }
468 
469   /**
470    * @return true if familyMap is non empty, false otherwise
471    */
472   public boolean hasFamilies() {
473     return !this.familyMap.isEmpty();
474   }
475 
476   /**
477    * @return the keys of the familyMap
478    */
479   public byte[][] getFamilies() {
480     if(hasFamilies()) {
481       return this.familyMap.keySet().toArray(new byte[0][0]);
482     }
483     return null;
484   }
485 
486   /**
487    * @return the startrow
488    */
489   public byte [] getStartRow() {
490     return this.startRow;
491   }
492 
493   /**
494    * @return the stoprow
495    */
496   public byte [] getStopRow() {
497     return this.stopRow;
498   }
499 
500   /**
501    * @return the max number of versions to fetch
502    */
503   public int getMaxVersions() {
504     return this.maxVersions;
505   }
506 
507   /**
508    * @return maximum number of values to return for a single call to next()
509    */
510   public int getBatch() {
511     return this.batch;
512   }
513 
514   /**
515    * @return maximum number of values to return per row per CF
516    */
517   public int getMaxResultsPerColumnFamily() {
518     return this.storeLimit;
519   }
520 
521   /**
522    * Method for retrieving the scan's offset per row per column
523    * family (#kvs to be skipped)
524    * @return row offset
525    */
526   public int getRowOffsetPerColumnFamily() {
527     return this.storeOffset;
528   }
529 
530   /**
531    * @return caching the number of rows fetched when calling next on a scanner
532    */
533   public int getCaching() {
534     return this.caching;
535   }
536 
537   /**
538    * @return TimeRange
539    */
540   public TimeRange getTimeRange() {
541     return this.tr;
542   }
543 
544   /**
545    * @return RowFilter
546    */
547   public Filter getFilter() {
548     return filter;
549   }
550 
551   /**
552    * @return true is a filter has been specified, false if not
553    */
554   public boolean hasFilter() {
555     return filter != null;
556   }
557 
558   /**
559    * Set whether blocks should be cached for this Scan.
560    * <p>
561    * This is true by default.  When true, default settings of the table and
562    * family are used (this will never override caching blocks if the block
563    * cache is disabled for that family or entirely).
564    *
565    * @param cacheBlocks if false, default settings are overridden and blocks
566    * will not be cached
567    */
568   public void setCacheBlocks(boolean cacheBlocks) {
569     this.cacheBlocks = cacheBlocks;
570   }
571 
572   /**
573    * Get whether blocks should be cached for this Scan.
574    * @return true if default caching should be used, false if blocks should not
575    * be cached
576    */
577   public boolean getCacheBlocks() {
578     return cacheBlocks;
579   }
580 
581   /**
582    * Set whether this scan is a reversed one
583    * <p>
584    * This is false by default which means forward(normal) scan.
585    * 
586    * @param reversed if true, scan will be backward order
587    * @return this
588    */
589   public Scan setReversed(boolean reversed) {
590     this.reversed = reversed;
591     return this;
592   }
593 
594   /**
595    * Get whether this scan is a reversed one.
596    * @return true if backward scan, false if forward(default) scan
597    */
598   public boolean isReversed() {
599     return reversed;
600   }
601 
602   /**
603    * Set the value indicating whether loading CFs on demand should be allowed (cluster
604    * default is false). On-demand CF loading doesn't load column families until necessary, e.g.
605    * if you filter on one column, the other column family data will be loaded only for the rows
606    * that are included in result, not all rows like in normal case.
607    * With column-specific filters, like SingleColumnValueFilter w/filterIfMissing == true,
608    * this can deliver huge perf gains when there's a cf with lots of data; however, it can
609    * also lead to some inconsistent results, as follows:
610    * - if someone does a concurrent update to both column families in question you may get a row
611    *   that never existed, e.g. for { rowKey = 5, { cat_videos => 1 }, { video => "my cat" } }
612    *   someone puts rowKey 5 with { cat_videos => 0 }, { video => "my dog" }, concurrent scan
613    *   filtering on "cat_videos == 1" can get { rowKey = 5, { cat_videos => 1 },
614    *   { video => "my dog" } }.
615    * - if there's a concurrent split and you have more than 2 column families, some rows may be
616    *   missing some column families.
617    */
618   public void setLoadColumnFamiliesOnDemand(boolean value) {
619     this.loadColumnFamiliesOnDemand = value;
620   }
621 
622   /**
623    * Get the raw loadColumnFamiliesOnDemand setting; if it's not set, can be null.
624    */
625   public Boolean getLoadColumnFamiliesOnDemandValue() {
626     return this.loadColumnFamiliesOnDemand;
627   }
628 
629   /**
630    * Get the logical value indicating whether on-demand CF loading should be allowed.
631    */
632   public boolean doLoadColumnFamiliesOnDemand() {
633     return (this.loadColumnFamiliesOnDemand != null)
634       && this.loadColumnFamiliesOnDemand.booleanValue();
635   }
636 
637   /**
638    * Compile the table and column family (i.e. schema) information
639    * into a String. Useful for parsing and aggregation by debugging,
640    * logging, and administration tools.
641    * @return Map
642    */
643   @Override
644   public Map<String, Object> getFingerprint() {
645     Map<String, Object> map = new HashMap<String, Object>();
646     List<String> families = new ArrayList<String>();
647     if(this.familyMap.size() == 0) {
648       map.put("families", "ALL");
649       return map;
650     } else {
651       map.put("families", families);
652     }
653     for (Map.Entry<byte [], NavigableSet<byte[]>> entry :
654         this.familyMap.entrySet()) {
655       families.add(Bytes.toStringBinary(entry.getKey()));
656     }
657     return map;
658   }
659 
660   /**
661    * Compile the details beyond the scope of getFingerprint (row, columns,
662    * timestamps, etc.) into a Map along with the fingerprinted information.
663    * Useful for debugging, logging, and administration tools.
664    * @param maxCols a limit on the number of columns output prior to truncation
665    * @return Map
666    */
667   @Override
668   public Map<String, Object> toMap(int maxCols) {
669     // start with the fingerpring map and build on top of it
670     Map<String, Object> map = getFingerprint();
671     // map from families to column list replaces fingerprint's list of families
672     Map<String, List<String>> familyColumns =
673       new HashMap<String, List<String>>();
674     map.put("families", familyColumns);
675     // add scalar information first
676     map.put("startRow", Bytes.toStringBinary(this.startRow));
677     map.put("stopRow", Bytes.toStringBinary(this.stopRow));
678     map.put("maxVersions", this.maxVersions);
679     map.put("batch", this.batch);
680     map.put("caching", this.caching);
681     map.put("maxResultSize", this.maxResultSize);
682     map.put("cacheBlocks", this.cacheBlocks);
683     map.put("loadColumnFamiliesOnDemand", this.loadColumnFamiliesOnDemand);
684     List<Long> timeRange = new ArrayList<Long>();
685     timeRange.add(this.tr.getMin());
686     timeRange.add(this.tr.getMax());
687     map.put("timeRange", timeRange);
688     int colCount = 0;
689     // iterate through affected families and list out up to maxCols columns
690     for (Map.Entry<byte [], NavigableSet<byte[]>> entry :
691       this.familyMap.entrySet()) {
692       List<String> columns = new ArrayList<String>();
693       familyColumns.put(Bytes.toStringBinary(entry.getKey()), columns);
694       if(entry.getValue() == null) {
695         colCount++;
696         --maxCols;
697         columns.add("ALL");
698       } else {
699         colCount += entry.getValue().size();
700         if (maxCols <= 0) {
701           continue;
702         } 
703         for (byte [] column : entry.getValue()) {
704           if (--maxCols <= 0) {
705             continue;
706           }
707           columns.add(Bytes.toStringBinary(column));
708         }
709       } 
710     }       
711     map.put("totalColumns", colCount);
712     if (this.filter != null) {
713       map.put("filter", this.filter.toString());
714     }
715     // add the id if set
716     if (getId() != null) {
717       map.put("id", getId());
718     }
719     return map;
720   }
721 
722   /**
723    * Enable/disable "raw" mode for this scan.
724    * If "raw" is enabled the scan will return all
725    * delete marker and deleted rows that have not
726    * been collected, yet.
727    * This is mostly useful for Scan on column families
728    * that have KEEP_DELETED_ROWS enabled.
729    * It is an error to specify any column when "raw" is set.
730    * @param raw True/False to enable/disable "raw" mode.
731    */
732   public void setRaw(boolean raw) {
733     setAttribute(RAW_ATTR, Bytes.toBytes(raw));
734   }
735 
736   /**
737    * @return True if this Scan is in "raw" mode.
738    */
739   public boolean isRaw() {
740     byte[] attr = getAttribute(RAW_ATTR);
741     return attr == null ? false : Bytes.toBoolean(attr);
742   }
743 
744   /**
745    * Set whether this scan is a small scan
746    * <p>
747    * Small scan should use pread and big scan can use seek + read
748    * 
749    * seek + read is fast but can cause two problem (1) resource contention (2)
750    * cause too much network io
751    * 
752    * [89-fb] Using pread for non-compaction read request
753    * https://issues.apache.org/jira/browse/HBASE-7266
754    * 
755    * On the other hand, if setting it true, we would do
756    * openScanner,next,closeScanner in one RPC call. It means the better
757    * performance for small scan. [HBASE-9488].
758    * 
759    * Generally, if the scan range is within one data block(64KB), it could be
760    * considered as a small scan.
761    * 
762    * @param small
763    */
764   public void setSmall(boolean small) {
765     this.small = small;
766   }
767 
768   /**
769    * Get whether this scan is a small scan
770    * @return true if small scan
771    */
772   public boolean isSmall() {
773     return small;
774   }
775 }