CN105893373A - Method and device for lowering Hash collision probability - Google Patents
Method and device for lowering Hash collision probability Download PDFInfo
- Publication number
- CN105893373A CN105893373A CN201410725232.3A CN201410725232A CN105893373A CN 105893373 A CN105893373 A CN 105893373A CN 201410725232 A CN201410725232 A CN 201410725232A CN 105893373 A CN105893373 A CN 105893373A
- Authority
- CN
- China
- Prior art keywords
- hash
- units
- bucket
- keyword
- current
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a method and a device for lowering a Hash collision probability. One Hash table with M pieces of Hash units is evenly divided into a plurality of Hash buckets of which the capacity is N pieces of Hash units, wherein the number of the Hash buckets is M, and two adjacent Hash buckets is provided with the overlap of (N-1) pieces of Hash units. The Hash collision probability can be effectively lowered.
Description
Technical field
The present invention relates to the network equipment, particularly to the method using Hash lookup key word in the network device
And device.
Background technology
In the network device, lookup method based on Hash is very universal, most typically MAC
The lookup of (Medium/MediaAccess Control, medium access control) address table.Using Hash
During lookup, the keyword message searched is carried out Hash operation, obtains a cryptographic Hash, as accessing Hash
The content of hash units corresponding in Hash table is gone to read in the address of table, and enters with keyword message to be found
Row compares, if the hash units of correspondence is empty, then for not mate, if the hash units of correspondence is not empty, and
Identical with keyword to be found, then for coupling, otherwise, necessarily corresponding hash units be sky but with
Keyword to be found differs, then be hash-collision.Hash lookup is a kind of high-speed searching method, resonable
In the case of thinking, it is possible to achieve complexity is the search performance of o (1).
The shortcoming of Hash lookup is, due to hash algorithm (hash function) set up from set of keywords to Kazakhstan
Mapping between uncommon address is the mapping of more to, and the existence of hash-collision is inevitable, the most different keys
The Hash Round Robin data partition that word obtains after hash function computing is identical.When after hash-collision, search performance can notable under
Fall.
One complete Hash lookup implementation, it should include the treatment measures of hash-collision, conventional side
Method includes: public overflow area method, chain technique, open addressing method, then Hash method etc.;No matter use which kind of side
Method, when hash-collision, search performance the most inevitably to decline, and therefore, reduces the general of hash-collision
Rate has great meaning.
Reducing the probability of hash-collision, can be realized by the size increasing Hash table, its cost is to use
More memory space.What the method that can also use Hash bucket, i.e. Hash Round Robin data partition were corresponding is no longer a Kazakhstan
Uncommon unit, but the Hash bucket being made up of several hash units, adding keyword in Hash table
Time, as long as bucket has an idle hash units, conflict would not occur.Its cost is, when searching,
Need the multiple hash units in bucket are compared.It can be seen that when the hash units quantity in Hash bucket
The most, hash-collision probability is the least, but now every time Hash lookup time, need the Hash reading and comparing
Element number is the most, and this obviously can affect the overall performance of Hash lookup, so, it is achieved time, need power
Weighing apparatus hash-collision probability and the capacity of Hash bucket, compromise, and relatively common is to store in a Hash bucket
2,4,6 or 8 hash units.
Traditional lookup based on Hash bucket, the result of Hash operation is the address (n) as Hash bucket, and
It it not the address of hash units.It is 4 examples with the size of Hash bucket, as it is shown in figure 1, each in figure is real
The little square frame of line, represents a hash units, 4 the continuous print hash units being framed with dotted line, forms one
Individual Hash bucket.If 4 the most occupied situations of hash units in the n-th Hash bucket (Bucket (n))
Under, and if the Hash operation result of new keyword is also equal to n, no matter in other adjoining Hash bucket
The most available free, all process only as hash-collision.
Visible, it is necessary existing lookup based on Hash bucket is further improved in fact.
Summary of the invention
The technical problem to be solved in the present invention is, for the drawbacks described above of prior art, the present invention proposes
A kind of technology reducing hash-collision probability, it is possible to efficiently reduce the probability of hash-collision.
The technical solution adopted for the present invention to solve the technical problems includes: provide one to reduce hash-collision general
The method of rate, a Hash table with M hash units is evenly dividing into capacity is N number of hash units
Multiple Hash buckets in;Wherein, the number of Hash bucket is M, has between two adjacent Hash buckets
The overlap of N-1 hash units.
In certain embodiments, when adding new keyword in this Hash table, if this new keyword
The corresponding hash units Kazakhstan that is the most occupied and that close on mutually with this current Hash bucket in current Hash bucket
Hash units the most available free in uncommon bucket, then can take hash units in this current Hash bucket
Content transfer to an idle hash units in the Hash bucket that this phase is closed on, so that this current Hash bucket
In can vacate a hash units to take for this new keyword.
In certain embodiments, when searching this Hash table, need to read and keyword institute the most to be found
The corresponding whole hash units in current Hash bucket.
The technical solution adopted for the present invention to solve the technical problems also includes: provide one to reduce hash-collision
The device of probability, comprising: multiple Hash bucket, the capacity of each Hash bucket is N number of hash units, these
Hash bucket constitutes the Hash table with M hash units;One Hash table operating unit, in order to this Hash
Table operates;Wherein, the number of Hash bucket is M, has N-1 between two adjacent Hash buckets
The overlap of individual hash units.
In certain embodiments, this Hash table operating unit when adding new keyword in this Hash table,
If the hash units in the current Hash bucket corresponding to this new keyword is the most occupied and current with this
Hash units the most available free in the Hash bucket that Hash bucket closes on mutually, then can be in this current Hash bucket
The individual content having taken hash units transfers to an idle hash units in the Hash bucket that this phase is closed on, from
And a hash units can be vacated in making this current Hash bucket to take for this new keyword.
In certain embodiments, this Hash table operating unit, when searching this Hash table, needs to read and compare
The whole hash units in current Hash bucket corresponding to keyword to be found.
In the present invention, the Hash bucket that this phase is closed on includes at least N-1 Hash bucket, in these Hash buckets
The overlapping of at least 1 hash units is there is in any one with this current Hash bucket.
In the present invention, N is 2,4,6 or 8;M is the integral multiple of N.
Compared with prior art, the method and device reducing hash-collision probability of the present invention, by introducing weight
The concept of folded Hash bucket, and can not increase each Hash in the case of the size of identical Hash bucket and looks into
In the case of the hash units number read when looking for and compare, it is possible to efficiently reduce the probability of hash-collision.
Accompanying drawing explanation
Fig. 1 is existing Hash table and the structural representation of Hash bucket.
Fig. 2 is Hash table and the structural representation of Hash bucket of the present invention.
Fig. 3 is the structural representation of the device reducing hash-collision probability of the present invention.
Fig. 4 is the interpolation keyword flow process signal to Hash table of the present invention.
Fig. 5 is the flow process signal deleting keyword from Hash table of the present invention.
Fig. 6 is the flow process signal of the lookup that Hash table carries out keyword of the present invention.
Detailed description of the invention
In conjunction with accompanying drawing, presently preferred embodiments of the present invention is elaborated.
The present invention proposes a kind of method reducing hash-collision probability, will have a Kazakhstan of M hash units
Uncommon table is evenly dividing in multiple Hash buckets that capacity is N number of hash units, and wherein, the number of Hash bucket is
M, there is between two adjacent Hash buckets the overlap of N-1 hash units.Wherein, M is the whole of N
Several times, need depending on actual application, can be very big, can be the least.N is 2,4,6 or 8.With N equal to 4
As a example by, owing to the result of Hash operation is the address of hash units, also it is the address of Hash bucket, corresponding to it
Hash bucket comprise continuous 4 hash units started from this hash units address.Visible, at overlapping Hash
In the method for bucket, the quantity of Hash bucket is equal to the quantity of hash units address, is in tradition Hash bucket method
4 times of the quantity of Hash bucket.
The inventive method, when adding new keyword in this Hash table, if this new keyword institute is right
Hash units in the current Hash bucket answered Hash bucket that is the most occupied and that close on mutually with this current Hash bucket
In the most available free hash units, then in this current Hash bucket can have been taken in hash units
Hold the idle hash units transferring in the Hash bucket that this phase is closed on, so that energy in this current Hash bucket
Enough vacate a hash units to take for this new keyword.
The inventive method, when searching this Hash table, needs to read and corresponding to keyword the most to be found
Current Hash bucket in whole hash units.
It is 4 illustrative with the capacity N of Hash bucket below,
As in figure 2 it is shown, the Hash table that Fig. 2 is the present invention and the structural representation of Hash bucket.That closes on mutually is multiple
Hash barrel section overlaps.Specifically, there are between the most adjacent Hash bucket 3 hash units
Overlap.Now, in the n-th Hash bucket (Bucket (n)), 4 hash units are the most occupied
In the case of, if the Hash operation result of keyword to be added is also equal to n, although the n-th Hash
4 hash units of bucket Bucket (n) are by the first keyword Key1, the second keyword Key2, the 3rd
Keyword Key3 and the 4th keyword Key4 takies, and the most not necessarily can conflict.Multiple situation is below divided to enter
Row describes in detail.For the sake of for ease of, the Hash operation result of keyword key is designated as Hash (key).
Situation one, if the hash units in Fig. 2 (n-3) is idle, if Hash (key1)=(n-3),
Then the first keyword key1 can move to hash units (n-3) from hash units (n), newly frees out
Hash units (n), uses to keyword to be added.
Situation two, if the hash units in Fig. 2 (n-2) is idle, if Hash (key1)=(n-3) or
Hash (key1)=(n-2), then the first keyword key1 can move to Hash list from hash units (n)
Unit (n-2), the hash units (n) newly freed out, use to keyword to be added.Otherwise, if
Hash (key2)=(n-2), then the second keyword key2 can move to Hash from hash units (n+1)
Unit (n-2), the hash units newly freed out (n+1), use to keyword to be added.
Situation three, if the hash units in Fig. 2 (n-1) is idle, if Hash (key1)=(n-3) or
Hash (key1)=(n-2) or Hash (key1)=(n-1), then the first keyword key1 can be from Kazakhstan
Uncommon unit (n) moves to hash units (n-1), the hash units (n) newly freed out, and can give to be added
Keyword use.Otherwise, if Hash (key2)=(n-2) or Hash (key2)=(n-1), then
Two keyword key2 can move to hash units (n-1), the Kazakhstan newly freed out from hash units (n+1)
Uncommon unit (n+1), uses to keyword to be added.Otherwise, if Hash (key3)=(n-2),
Then the 3rd keyword key3 can move to hash units (n-1) from hash units (n+2), newly frees out
Hash units (n+2), use to keyword to be added.
Situation four, if the hash units in Fig. 2 (n+6) is idle, if Hash (key4)=(n+3),
Then the 4th keyword key4 can move to hash units (n+6) from hash units (n+3), newly frees out
Hash units (n+3), use to keyword to be added.
Situation five, if the hash units in Fig. 2 (n+5) is idle, if Hash (key4)=(n+3) or
Hash (key4)=(n+2), then the 4th keyword key4 can move to Hash from hash units (n+3)
Unit (n+5), the hash units newly freed out (n+3), use to keyword to be added.Otherwise,
If Hash (key3)=(n+2), then the 3rd keyword key3 can move to breathe out from hash units (n+2)
Uncommon unit (n+5), the hash units newly freed out (n+2), use to keyword to be added.
Situation six, if the hash units in Fig. 2 (n+4) is idle, if Hash (key4)=(n+3) or
Hash (key4)=(n+2) or Hash (key4)=(n+1), then the 4th keyword key4 can be from Kazakhstan
Uncommon unit (n+3) moves to hash units (n+4), the hash units newly freed out (n+3), can be treated
The keyword added is used.Otherwise, if Hash (key3)=(n+2) or Hash (key3)=(n+1),
Then the 3rd keyword key3 can move to hash units (n+4) from hash units (n+2), newly frees out
Hash units (n+2), use to keyword to be added.Otherwise, if Hash (key2)=(n+1),
Then the second keyword key2 can move to hash units (n+4) from hash units (n+1), newly frees out
Hash units (n+1), use to keyword to be added.
Under other circumstances, the first keyword key1, the second keyword key2, the 3rd keyword key3,
The hash units of the 4th keyword key4 is the most irremovable, then hash-collision is inevitable.Visible, this
It is bright by the keyword in certain hash units in current Hash bucket is moved in the Hash bucket closed on mutually
Idle hash units, make the hash units being newly available in current Hash bucket can be used to store new key
Word, such that it is able to reach to reduce the purpose of the probability of hash-collision.It is noted that alleged by the present invention
The Hash bucket closed on mutually with current Hash bucket, depends on current Hash bucket position in Hash table, can have
3 to 6 Hash buckets, any one in these Hash buckets are had to there is at least 1 Kazakhstan with current Hash bucket
The overlap of uncommon unit.
As it is shown on figure 3, the structural representation of the device reducing hash-collision probability that Fig. 3 is the present invention.This dress
Putting and include: multiple Hash buckets, the capacity of each Hash bucket is N number of hash units, and these Hash buckets constitute tool
There is the Hash table 101 of M hash units;One Hash table operating unit 102, in order to this Hash table 101
Operate;Wherein, the number of Hash bucket is M, has N-1 Kazakhstan between two adjacent Hash buckets
The overlap of uncommon unit.This Hash table operating unit 102 can be divided into three functional modules: the interpolation mould of Hash table
Block, the removing module of Hash table, the lookup module of Hash table.Below in conjunction with flow chart, to above three
Module, is described in detail respectively.
As shown in Figure 4, Fig. 4 is the interpolation keyword flow process signal to Hash table of the present invention.Add and close
Key word generally comprises to the flow process of Hash table:
Step A01, new key value is carried out Hash operation, obtains Hash Round Robin data partition (n=hash (key)),
Read the content from (n-3) to (n+6) totally 10 hash units.
The Hash ground corresponding for key of storage in step A02, calculating n to (n+3) these 4 hash units
Location, and it is designated as hash (key1), hash (key2), hash (key3), hash (key4).
Step A03, judge n to (n+3) these 4 hash units the most at least 1 for sky, if being
Go to step A07, otherwise go to step A04.
Step A04, (n-3) hash units that judges whether are sky, and hash (key1)=(n-3), yes
Words go to step A08, otherwise go to step A05.
Step A05, judge whether (n-2) hash units for sky, and hash (key1)=(n-3) or
Hash (key1)=(n-2) or hash (key2)=(n-2), goes to step A09, otherwise turns step if being
Rapid A06.
Step A06, judge whether (n-1) hash units for sky, and hash (key1)=(n-3) or
Hash (key1)=(n-2) or hash (key1)=(n-1) or hash (key2)=(n-2) or
Hash (key2)=(n-1) or hash (key3)=(n-1), goes to step A10, otherwise goes to step A07 if being.
Step A07, the hash units of that free time that new key value is added to.
Step A08, key1 is moved on to (n-3) hash units, then new key value is added to that new
The hash units freed out.
If step A09 hash (key1)=(n-3) or hash (key1)=(n-2), then key1
Move on to (n-2) hash units;Otherwise key2 is moved on to (n-2) hash units, then new key value is added
It is added to that hash units newly freed out.
If step A10 hash (key1)=(n-3) or hash (key1)=(n-2) or hash (key1)
=(n-1), then move on to (n-1) hash units key1, otherwise, if or hash (key2)=(n-2) or
Hash (key2)=(n-1), then move on to (n-1) hash units key2, otherwise key3 moved on to (n-1)
Hash units, then adds that hash units newly freed out to new key value.
Step A11, judge whether (n+6) hash units for sky, and hash (key4)=(n+3) be if turn
Step A14, otherwise goes to step A12.
Step A12, judge whether (n+5) hash units for sky, and hash (key4)=(n+3) or
Hash (key4)=(n+2) or hash (key3)=(n+2) goes to step A15 if being, otherwise goes to step A13.
Step A13, judge whether (n+4) hash units for sky, and hash (key4)=(n+3) or
Hash (key4)=(n+2) or hash (key4)=(n+1) or hash (key3)=(n+2) or
Hash (key3)=(n+1) or hash (key2)=(n+1) goes to step A16 if being, otherwise goes to step A17.
Step A14, key4 is moved on to (n+6) hash units, then new key value is added to that new
The hash units freed out.
If step A15 hash (key4)=(n+3) or hash (key4)=(n+2), then key4 is moved on to (n+5)
Hash units, otherwise moves on to (n+5) hash units key3, then new key value is added to that new
The hash units freed out.
If step A16 hash (key4)=(n+3) or hash (key4)=(n+2) or
Hash (key4)=(n+1), then move on to (n+4) hash units key4, if otherwise hash (key3)=(n+2)
Or hash (key3)=(n+1), then key3 is moved on to (n+4) hash units, otherwise key2 is moved on to (n+4)
Hash units, then adds that hash units newly freed out to new key value.
Step A17, hash-collision occurs, new key value is saved in hash-collision table.
As it is shown in figure 5, the flow process signal deleting keyword from Hash table that Fig. 5 is the present invention.From Kazakhstan
The flow process deleting keyword in uncommon table generally comprises:
Step D01, key value to be deleted is carried out Hash operation, obtain Hash Round Robin data partition (n=
Hash (key)), read the content from n to (n+3) totally 4 hash units.
Step D02, judge hash units n to key that is the most empty and that wherein store equal to be deleted
Key, goes to step D06, otherwise goes to step D03 if being.
Step D03, judge hash units (n+1) to key that is the most empty and that wherein store equal to be deleted
Key, go to step D07 if being, otherwise go to step D04.
Step D04, judge hash units (n+2) to key that is the most empty and that wherein store equal to be deleted
Key, go to step D08 if being, otherwise go to step D05.
Step D05, judge hash units (n+3) to key that is the most empty and that wherein store equal to be deleted
Key, go to step D09 if being, otherwise go to step D10.
Step D06, the key value deleted in hash units n.
Step D07, the key value deleted in hash units (n+1).
Step D08, the key value deleted in hash units (n+2).
Step D09, the key value deleted in hash units (n+3).
There is hash-collision in step D10, this key value, need to delete this key value in hash-collision table.
As shown in Figure 6, Fig. 6 be the present invention the lookup that Hash table is carried out keyword flow process signal.
The step of the lookup that Hash table carries out keyword is generally comprised:
Step F01, key value to be found is carried out Hash operation, obtain Hash Round Robin data partition (n=
Hash (key)), read the content from n to (n+3) totally 4 hash units.
Step F02, judge hash units n to key that is the most empty and that wherein store equal to be found
Key, goes to step F06, otherwise goes to step F03 if being.
Step F03, judge hash units (n+1) to key that is the most empty and that wherein store equal to be found
Key, go to step F07 if being, otherwise go to step F04.
Step F04, judge hash units (n+2) to key that is the most empty and that wherein store equal to be found
Key, go to step F08 if being, otherwise go to step F05.
Step F05, judge hash units (n+3) to key that is the most empty and that wherein store equal to be found
Key, go to step F09 if being, otherwise go to step F10.
Lookup result in step F06, output hash units n.
Lookup result in step F07, output hash units (n+1).
Lookup result in step F08, output hash units (n+2).
Lookup result in step F09, output hash units (n+3).
There is hash-collision in step F10, this key value, need to search this ke in hash-collision tableyValue
Lookup result.
The beneficial effect of the method and device reducing hash-collision probability of the present invention includes: by introducing overlap
The concept of Hash bucket a so that hash units can belong simultaneously to multiple Hash bucket, at certain Hash bucket
In the case of Man, by certain hash units of the fullest Hash bucket move on to affiliated other less than Hash
In Tong, make the fullest Hash bucket become discontented, thus reach to reduce the purpose of the probability of hash-collision.
It is noted that the invention is not restricted to certain specific hash algorithm (hash function), can be with many
Kind of hash algorithm with the use of, therefore, above-mentioned flow chart does not the most describe the realization of concrete hash algorithm
Flow process.The invention is not restricted to certain specific hash-collision treatment mechanism, can be with multiple hash-collision datatron
Make with the use of, therefore, after above-mentioned flow chart does not the most describe conflict, realize flow process.
It should be appreciated that above example is only in order to illustrate technical scheme, it is not intended to limit,
It will be understood by those skilled in the art that the technical scheme described in above-described embodiment can be modified, or
Wherein portion of techniques feature is carried out equivalent;And these amendments and replacement, all should belong to appended by the present invention
Scope of the claims.
Claims (10)
1. the method reducing hash-collision probability, uniformly draws a Hash table with M hash units
It is divided in multiple Hash buckets that capacity is N number of hash units;It is characterized in that, the number of Hash bucket is
M, there is between two adjacent Hash buckets the overlap of N-1 hash units.
Method the most according to claim 1, it is characterised in that adding new key in this Hash table
During word, if the hash units in the current Hash bucket corresponding to this new keyword is the most occupied and
Hash units the most available free in the Hash bucket closed on mutually with this current Hash bucket, then can be current this
A content having taken hash units in Hash bucket transfers in the Hash bucket that this phase is closed on
Idle hash units, so that can vacate a hash units to new for this in this current Hash bucket
Keyword takies.
Method the most according to claim 2, it is characterised in that the Hash bucket that this phase is closed on includes at least N-1
Individual Hash bucket, there is at least 1 Hash list with this current Hash bucket in any one in these Hash buckets
The overlap of unit.
Method the most according to claim 2, it is characterised in that when searching this Hash table, needs to read
And the whole hash units in the current Hash bucket corresponding to keyword the most to be found.
Method the most according to claim 1, it is characterised in that N is 2,4,6 or 8;M is N's
Integral multiple.
6. reduce a device for hash-collision probability, comprising: multiple Hash bucket, the capacity of each Hash bucket
For N number of hash units, these Hash buckets constitute the Hash table with M hash units;One Hash
Table handling unit, in order to operate this Hash table;It is characterized in that, the number of Hash bucket is M
Individual, there is between two adjacent Hash buckets the overlap of N-1 hash units.
Device the most according to claim 6, it is characterised in that this Hash table operating unit is to this Hash
When table adds new keyword, if the Hash in the current Hash bucket corresponding to this new keyword
Unit is the most occupied and Hash list the most available free in the Hash bucket that closes on mutually with this current Hash bucket
Unit, then can transfer to this content having taken hash units in this current Hash bucket adjacent
An idle hash units near Hash bucket, so that can vacate one in this current Hash bucket
Hash units takies for this new keyword.
Device the most according to claim 7, it is characterised in that the Hash bucket that this phase is closed on includes at least N-1
Individual Hash bucket, there is at least 1 Hash list with this current Hash bucket in any one in these Hash buckets
The overlap of unit.
Device the most according to claim 7, it is characterised in that this Hash table operating unit is searching this Kazakhstan
During uncommon table, need the whole Kazakhstan reading and comparing in the current Hash bucket corresponding to keyword the most to be found
Uncommon unit.
Device the most according to claim 6, it is characterised in that N is 2,4,6 or 8;M is N
Integral multiple.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410725232.3A CN105893373A (en) | 2014-12-04 | 2014-12-04 | Method and device for lowering Hash collision probability |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410725232.3A CN105893373A (en) | 2014-12-04 | 2014-12-04 | Method and device for lowering Hash collision probability |
Publications (1)
Publication Number | Publication Date |
---|---|
CN105893373A true CN105893373A (en) | 2016-08-24 |
Family
ID=56699363
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201410725232.3A Pending CN105893373A (en) | 2014-12-04 | 2014-12-04 | Method and device for lowering Hash collision probability |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105893373A (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105550208A (en) * | 2015-12-02 | 2016-05-04 | 南京邮电大学 | Similarity storage design method based on spectral hashing |
CN106708438A (en) * | 2016-12-16 | 2017-05-24 | 盛科网络(苏州)有限公司 | Method and device for lowering Hash conflict probability |
CN107330047A (en) * | 2017-06-28 | 2017-11-07 | 华信塞姆(成都)科技有限公司 | A kind of FPGA training and enquiry circuit implementation method based on perfect hash algorithm |
CN111857982A (en) * | 2019-04-25 | 2020-10-30 | 浙江大学 | Data processing method and device |
-
2014
- 2014-12-04 CN CN201410725232.3A patent/CN105893373A/en active Pending
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105550208A (en) * | 2015-12-02 | 2016-05-04 | 南京邮电大学 | Similarity storage design method based on spectral hashing |
CN105550208B (en) * | 2015-12-02 | 2019-04-02 | 南京邮电大学 | Similitude design Storage method based on spectrum Hash |
CN106708438A (en) * | 2016-12-16 | 2017-05-24 | 盛科网络(苏州)有限公司 | Method and device for lowering Hash conflict probability |
CN107330047A (en) * | 2017-06-28 | 2017-11-07 | 华信塞姆(成都)科技有限公司 | A kind of FPGA training and enquiry circuit implementation method based on perfect hash algorithm |
CN107330047B (en) * | 2017-06-28 | 2020-06-30 | 华信塞姆(成都)科技有限公司 | FPGA training and inquiring circuit implementation method based on perfect Hash algorithm |
CN111857982A (en) * | 2019-04-25 | 2020-10-30 | 浙江大学 | Data processing method and device |
CN111857982B (en) * | 2019-04-25 | 2023-10-27 | 浙江大学 | Data processing method and device |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US10346432B2 (en) | Compaction policy | |
CN105893373A (en) | Method and device for lowering Hash collision probability | |
CN103473239B (en) | A kind of data of non relational database update method and device | |
CN102542071B (en) | Distributed data processing system and method | |
US20190065546A1 (en) | Multi stage aggregation using digest order after a first stage of aggregation | |
US9710505B1 (en) | Systems and methods for reducing database index contention and generating unique database identifiers | |
US10657066B2 (en) | Method and device for storage management | |
CN105205082A (en) | Method and system for processing file storage in HDFS | |
CN105989076A (en) | Data statistical method and device | |
CN104111936B (en) | Data query method and system | |
CN107704202B (en) | Method and device for quickly reading and writing data | |
US20110040761A1 (en) | Estimation of postings list length in a search system using an approximation table | |
CN103778202A (en) | Enterprise electronic document managing server side and system | |
US10998918B2 (en) | Reducing the amount of data stored in a sequence of data blocks by combining deduplication and compression | |
US11294805B2 (en) | Fast and safe storage space reclamation for a data storage system | |
KR101426673B1 (en) | Method of Distributed Indexing and Searching for Effective Analysis of Time Series Data in Search System | |
Busch et al. | Packing of graphic n‐tuples | |
CN103918230A (en) | Synchronizing forwarding databases in a network device background | |
US20200379948A1 (en) | Indexes and queries for files by indexing file directories | |
CN104933054B (en) | The URL storage methods and device of cache resource file, cache server | |
CN101236564A (en) | Mass data high performance reading display process | |
CN104038423A (en) | Open flow flow table refreshing method and router device | |
CN104166661A (en) | Data storage system and method | |
JP2021534505A (en) | Data storage method, data acquisition method, and equipment | |
CN108920613A (en) | A kind of metadata management method, system and equipment and storage medium |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
WD01 | Invention patent application deemed withdrawn after publication | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20160824 |