CN107579926A - The QoS methods to set up of Ceph cloud storage systems based on token bucket algorithm - Google Patents
The QoS methods to set up of Ceph cloud storage systems based on token bucket algorithm Download PDFInfo
- Publication number
- CN107579926A CN107579926A CN201710986526.5A CN201710986526A CN107579926A CN 107579926 A CN107579926 A CN 107579926A CN 201710986526 A CN201710986526 A CN 201710986526A CN 107579926 A CN107579926 A CN 107579926A
- Authority
- CN
- China
- Prior art keywords
- request
- token
- token bucket
- sub
- queue
- 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.)
- Granted
Links
Landscapes
- Data Exchanges In Wide-Area Networks (AREA)
- Computer And Data Communications (AREA)
Abstract
The invention discloses a kind of QoS methods to set up of the Ceph cloud storage systems based on token bucket algorithm, the Ceph cloud storage systems include multiple client and the service end with multiple client communication connection, the service end is provided with total token bucket module, include more sub- token bucket modules in total token bucket module, it is in one-to-one relationship between each client and each sub- token bucket module, the invention provides the limitations of the limit in QoS, in the presence of all when multi-client carries out I/O operation, its maximum IOPS can be limited, some clients can be avoided to take the phenomenon of a large amount of IOPS and other clients starvation；Provide the ability of processing burst simultaneously, the depth of bucket is that the highest number of tokens that keeps of sub- token bucket module can be to the larger IO processing to arrive out of the blue, the problems such as avoiding shake.
Description
Technical field
The invention belongs to field of distributed storage, and in particular to a kind of Ceph cloud storage systems based on token bucket algorithm
QoS methods to set up.
Background technology
Cloud storage is the extension of cloud computing concept and the emerging concept developed, and it refers to by cluster application, network
A large amount of various types of storage devices gather association by application software in the network such as technology or distributed file system
With the common system for externally providing data storage and Operational Visit function of work, ensure the security of data, and save and deposit
Store up space.
QoS（Quality of Service）It is a kind of controlling mechanism, it is provided for different user or different pieces of information
Stream uses corresponding different priorities, or the requirement according to program, ensures that the performance of data flow reaches certain level.
QoS refers to the probability that network meets given business contract in packet switching network field；In telephone service field, refer to circuit and make an uproar
The measurement of the index such as sound and sound, appropriate audio volume level；The IOPS accessed storage is then mainly shown as in field of storage
The control of (Input/Output Operations Per Second).The QoS guarantee utilization of resources is provided in each professional domain
All it is critically important.
Ceph eliminates the dependence to single Centroid, is truly realized a distributed storage system of non-stop layer structure
System, this is that other distributed memory systems can not be than.Ceph is used as current most popular distributed memory system of increasing income,
Openstack one of storage scheme of increasing income is turned into, it has been included among linux kernel.And QoS is in storage
Quite important, due to the relation of resource contention, a variety of workloads fight for mutually IOPS, to ensure that the bandwidth of mission critical can
With, there is provided QoS technology is necessary.Especially in the case of using SSD, with the steady increase of SSD capacity, a variety of work
Load competition SSD turns into normality；And SSD can be safeguarded automatically, this regular maintenance can influence performance, and now QoS can be with
Priority ranking is carried out to memory bandwidth so that mission critical can also obtain required bandwidth when resource contention fierceness,
Otherwise mission critical may occur to compete less than enough bandwidth, and the execution with other tasks slowly together.Ceph is more at present
New version is proposed the optimization architecture that bluestore etc. is directed to SSD, but there has been no the support to QoS at present by Ceph.
Token bucket（token bucket）Algorithm is that traffic shaping and speed limit are realized in packet switched computer network
One of the most frequently used algorithm of system, it tackles the packet of arrival by the way of fixed rate produces token, made according to residue
The granting or discarding of the quantity determination data bag of board.And token bucket algorithm not only can be used in network technology, storage etc. its
Still there is very big application potential in its field.
The content of the invention
The technical problems to be solved by the invention are to provide a kind of token bucket that is based on for above-mentioned the deficiencies in the prior art to calculate
The QoS methods to set up of the Ceph cloud storage systems of method, based on the QoS methods to set up of the Ceph cloud storage systems of token bucket algorithm
The limit limitations in QoS are provided, when multi-client carries out I/O operation in the presence of all, its maximum IOPS can be limited, certain can be avoided
A little clients take the phenomenon of a large amount of IOPS and other clients starvation；The ability of processing burst, the depth of bucket are provided simultaneously
The highest number of tokens that i.e. sub- token bucket module is kept can avoid shake etc. and ask to the larger IO processing to arrive out of the blue
Topic.
To realize above-mentioned technical purpose, the technical scheme that the present invention takes is：
A kind of QoS methods to set up of the Ceph cloud storage systems based on token bucket algorithm, the Ceph cloud storage systems include more
Individual client and the service end with multiple client communication connection, the service end are provided with total token bucket module, total order
Include more sub- token bucket modules in board bucket module, in one by one between each client and each sub- token bucket module
Corresponding relation, the QoS methods to set up include：
（1）Maximum and token is set to produce speed：Preset the maximum and son of the token quantity included in total token bucket module
Token bucket module produces the speed of token；
（2）Produce I/O request：Client produces I/O request, and I/O request is stored in corresponding I/O request set queue by service end,
It is in one-to-one relationship between its neutron token bucket module and I/O request set queue；
（3）Obtain token：Sub- token bucket module in service end constantly produces token according to token generation speed and puts token
Enter in sub- token bucket module, while the token total quantity for ensureing to include in all sub- token bucket modules is not more than total token bucket module
The maximum of the token quantity inside included；
（4）Take head of the queue I/O request：Judge whether I/O request set queue is empty, if so, then terminate, otherwise sub- token bucket module root
The I/O request positioned at head of the queue is taken out according to putting in order in I/O request set queue；
（5）Ask token：Sub- token bucket module produces speed according to the size, type and token of the I/O request positioned at head of the queue and set
Token quantity required for the I/O request, if the order that the token quantity in sub- token bucket module is more than or equal to required for the I/O request
Board quantity, then take out the token of identical quantity out of sub- token bucket module and handle the I/O request, the I/O request is exited into IO please
Ask set queue and return to execution step（4）；If the token quantity in sub- token bucket module is less than the order required for the I/O request
Board quantity, then remaining all tokens in sub- token bucket module are taken out, the token quantity required for the I/O request is subtracted into sub- order
Remaining all token quantity in board bucket module, is updated to by the result queue being calculated to the I/O request and by the result
New token quantity required for the I/O request, the I/O request is placed back in the Head-of-line of I/O request set queue, performed
Next step；
（6）Check I/O request：The sub- token bucket module is regular and is examined in the IO for being located at head of the queue in I/O request set queue
Whether the new token quantity required for request is less than or equal to the token quantity in sub- token bucket module, if so, then from sub- token
Bucket module in take out with positioned at head of the queue I/O request required for new token quantity identical token and handle the I/O request, will
The I/O request exits I/O request set queue and returns to execution step（4）, otherwise sub- token bucket module end check.
The method of every processing I/O request between sub- token bucket module and each client is according to above-mentioned steps（1）Extremely
Step（6）Operated.
Further, I/O request is stored in corresponding I/O request set queue and included by the service end：
（1）I/O request is sequentially stored into I/O request set queue by the time sequencing that service end sends I/O request according to client；
（2）The dispatching priority of each I/O request of the service end in I/O request set queue to I/O request put in order into
The corresponding adjustment of row.
Further, the I/O request set queue includes read operation I/O request set queue and write operation I/O request set
Queue, the read operation I/O request set queue are used for the read operation for storing I/O request, the write operation I/O request set queue
For storing the write operation of I/O request.
The present invention has following technique effect：The invention provides the limit in QoS to limit, in the presence of all multi-client（With
Family）, can be by setting token generation rate in every sub- token bucket module to limit each client when carrying out I/O request operation
Maximum IOPS, the phenomenon that certain user takes a large amount of IOPS and other users are hungry can be avoided；It is prominent that processing is provided simultaneously
The ability of hair, the depth of bucket is that the highest number of tokens that sub- token bucket module is kept can be entered to the larger I/O request to arrive out of the blue
Row processing, the problems such as avoiding shake.The present invention formulates priority for different I/O requests, and enters row major point according to the priority
With IOPS, urgent, vital task is allowed also to obtain enough IOPS when resource contention is violent；It is and loose in resource contention
Shi Jinliang maximizes IOPS utilization rate.
Brief description of the drawings
Fig. 1 is the structural representation of the present invention.
Fig. 2 is the workflow diagram of the present invention.
Embodiment
The embodiment of the present invention is further illustrated below according to Fig. 1 and Fig. 2：
Referring to Fig. 1, a kind of QoS methods to set up of the Ceph cloud storage systems based on token bucket algorithm, there is provided in QoS guarantee
limit（Limitation）It is required that and can handle certain emergency case, it is allowed to client IOPS is temporarily higher than limits value.The Ceph clouds
Storage system includes multiple client and the service end with multiple client communication connection, and the service end is provided with total token bucket
Module, total token bucket module is interior to include more sub- token bucket modules, each client and each sub- token bucket
It is in one-to-one relationship between module, wherein, total token bucket module is a total token bucket, includes many small token buckets
（Sub- token bucket module）, each client is corresponded to, the speed summation that client produces token should be not more than total token bucket most
Big speed, sub- IOPS sums are represented as total IOPS.
Wherein described sub- token bucket module is corresponded in each client, sub- token bucket module have provide obtain token,
Ask token, maximum is set（The depth of bucket）, the generation speed of token is set and checks the function such as I/O request.
Described acquisition token is the unit time to produce speed n/t, t according to the token of setting, then can be when by T
Between after produce n* (T/t) individual token and be put into sub- token bucket module, and ensure in sub- token bucket module that token quantity is no more than and set
The maximum put.
When described request token is handles an I/O request, its institute is judged according to the size of its I/O request, type etc.
Need token quantity, be then attempt to from sub- token bucket module take out respective numbers token, if token number enough if take out order
Board simultaneously handles the I/O request；Remaining all tokens are taken out if token lazy weight and are marked and give the IO, continue waiting for token number
Reprocessed when measuring enough.
How many most tokens in total token bucket module set for described setting maximum, when in total token bucket module
When token quantity is more than the maximum, the maximum can be automatically regulated to be.
The token generation rate is the token number increased newly in unit time interior bucket, i.e., the value is used for providing QoS's
Limit ensures that the more big then limit values of the value are bigger, and the maximum IOPS obtained by user is bigger；Otherwise the IOPS of user can quilt
It is restricted to some less average value.
The inspection I/O request is examined in the medium pending I/O request of I/O request set queue, if it asks token
It is successful then handle and check next；Exited if failure.
The I/O request set queue is the queue of a storage I/O request, whenever user has new I/O request to add
Into the queue, wait to be treated.
There is comparatively large number of IO to be unexpected in a short time in the burst, the ability of this algorithm process burst with
The token generation rate v of setting is relevant with bucket depth degree h, is h/ (M-v), and wherein M is the maximum processing speed of physical machine.
In summary, the QoS methods to set up specifically include：Referring to Fig. 2,
（1）Maximum and token is set to produce speed：Preset the maximum and son of the token quantity included in total token bucket module
Token bucket module produces the speed of token；
（2）Produce I/O request：Client produces I/O request, and I/O request is stored in corresponding I/O request set queue by service end,
It is in one-to-one relationship between its neutron token bucket module and I/O request set queue；
（3）Obtain token：Sub- token bucket module in service end constantly produces token according to token generation speed and puts token
Enter in sub- token bucket module, while the token total quantity for ensureing to include in all sub- token bucket modules is not more than total token bucket module
The maximum of the token quantity inside included, if exceed default maximum, unnecessary order is abandoned according to default maximum
Board；
（4）Take head of the queue I/O request：Judge whether I/O request set queue is empty, if so, then terminate, otherwise sub- token bucket module root
The I/O request positioned at head of the queue is taken out according to putting in order in I/O request set queue；
（5）Ask token：Sub- token bucket module produces speed according to the size, type and token of the I/O request positioned at head of the queue and set
Token quantity required for the I/O request, if the order that the token quantity in sub- token bucket module is more than or equal to required for the I/O request
Board quantity, then take out the token of identical quantity out of sub- token bucket module and handle the I/O request, the I/O request is exited into IO please
Ask set queue and return to execution step（4）；If the token quantity in sub- token bucket module is less than the order required for the I/O request
Board quantity, then remaining all tokens in sub- token bucket module are taken out, the token quantity required for the I/O request is subtracted into sub- order
Remaining all token quantity in board bucket module, is updated to by the result queue being calculated to the I/O request and by the result
New token quantity required for the I/O request, the I/O request is placed back in the Head-of-line of I/O request set queue, performed
Next step；
（6）Check I/O request：The sub- token bucket module is regular and is examined in the IO for being located at head of the queue in I/O request set queue
Whether the new token quantity required for request is less than or equal to the token quantity in sub- token bucket module, if so, then from sub- token
Bucket module in take out with positioned at head of the queue I/O request required for new token quantity identical token and handle the I/O request, will
The I/O request exits I/O request set queue and returns to execution step（4）And step（5）, the sub- next IO of token bucket module check
Request, otherwise sub- token bucket module end check, after end check, in certain predetermined period（Within this cycle, service end
Interior sub- token bucket module constantly produces token according to token generation speed and token is put into sub- token bucket module）, sub- order
Board bucket module is checked again.
I/O request is stored in corresponding I/O request set queue by the service end to be included：
（1）I/O request is sequentially stored into I/O request set queue by the time sequencing that service end sends I/O request according to client；
（2）The dispatching priority of each I/O request of the service end in I/O request set queue to I/O request put in order into
The corresponding adjustment of row.The I/O request set queue can select to be Priority Queues, if so having different priority if I/O request
Enqueue can be prioritized to, priority scheduling can be achieved.
The I/O request set queue includes read operation I/O request set queue and write operation I/O request set queue, described
Read operation I/O request set queue is used for the read operation for storing I/O request, and the write operation I/O request set queue is used to store IO
The write operation of request；So when there is write operation obstruction, the normal processing of read operation is not interfered with.
The I/O request is probably point different types of, is such as simply divided into reading and writing, or scrub（Erasing）、
client（From client）Deng function distribution is pressed, I/O request may possess different priority or entrance according to the difference of type
Different subqueues.I/O request is waited after entering I/O request set queue.
As described in Figure 1, user is worked as（Or other structures, process etc.）Need to produce I/O request during read-write, at this moment I/O request meeting
Ranked into corresponding I/O request set queue, wait service end processing；The sub- token bucket module of service end can timing
Carry out sequence of operations（Including adding certain amount（Determined by user configuration）Token and as far as possible processing I/O request set queue
In I/O request）, the result most handled at last returns to client.Subsequent operation, the son of the service end are made for its judgement
Token bucket module can timing progress sequence of operations；The addition token refers to and unnecessary order can be abandoned according to the maximum of setting
Board.It is described to handle as far as possible, it is intended that as long as there is enough tokens to meet the I/O request positioned at head of the queue, then constantly to handle the IO
Ask and it is gone out O request set queue, until I/O request set queue is empty or can not meet head of the queue IO, when can not meet,
Remaining token is taken out and gives I/O request, marks the token number needed for I/O request to subtract corresponding quantity, then again by the IO
Head of the queue is put back to, waits the processing of next round.The result, including the information such as processing success or failure, if failure, client
Consideration resends request etc.；If success, by corresponding information（The information such as read）Return to client.
Due to the hereditability of bucket, when there is a number of token in bucket, token bucket can tackle a certain degree of prominent
Hair, the speed that the ability of reply produces token with token quantity and bucket that bucket accumulates are relevant.
In this example, pass through the implementation of token bucket so that Ceph cloud storages support the limit in QoS（Limitation）, and
Burst IO to a certain degree can be handled so that Ceph more adapts to production environment.
Protection scope of the present invention includes but is not limited to embodiment of above, and protection scope of the present invention is with claims
It is defined, any replacement being readily apparent that to those skilled in the art that this technology is made, deformation, improvement each fall within the present invention's
Protection domain.
Claims (3)
1. a kind of QoS methods to set up of the Ceph cloud storage systems based on token bucket algorithm, it is characterised in that the Ceph clouds are deposited
Storage system includes multiple client and the service end with multiple client communication connection, and the service end is provided with total token bucket mould
Block, total token bucket module is interior to include more sub- token bucket modules, each client and each sub- token bucket mould
It is in one-to-one relationship between block, the QoS methods to set up include：
（1）Maximum and token is set to produce speed：Preset the maximum and son of the token quantity included in total token bucket module
Token bucket module produces the speed of token；
（2）Produce I/O request：Client produces I/O request, and I/O request is stored in corresponding I/O request set queue by service end,
It is in one-to-one relationship between its neutron token bucket module and I/O request set queue；
（3）Obtain token：Sub- token bucket module in service end constantly produces token according to token generation speed and puts token
Enter in sub- token bucket module, while the token total quantity for ensureing to include in all sub- token bucket modules is not more than total token bucket module
The maximum of the token quantity inside included；
（4）Take head of the queue I/O request：Judge whether I/O request set queue is empty, if so, then terminate, otherwise sub- token bucket module root
The I/O request positioned at head of the queue is taken out according to putting in order in I/O request set queue；
（5）Ask token：Sub- token bucket module produces speed according to the size, type and token of the I/O request positioned at head of the queue and set
Token quantity required for the I/O request, if the order that the token quantity in sub- token bucket module is more than or equal to required for the I/O request
Board quantity, then take out the token of identical quantity out of sub- token bucket module and handle the I/O request, the I/O request is exited into IO please
Ask set queue and return to execution step（4）；If the token quantity in sub- token bucket module is less than the order required for the I/O request
Board quantity, then remaining all tokens in sub- token bucket module are taken out, the token quantity required for the I/O request is subtracted into sub- order
Remaining all token quantity in board bucket module, is updated to by the result queue being calculated to the I/O request and by the result
New token quantity required for the I/O request, the I/O request is placed back in the Head-of-line of I/O request set queue, performed
Next step；
（6）Check I/O request：The sub- token bucket module is regular and is examined in the IO for being located at head of the queue in I/O request set queue
Whether the new token quantity required for request is less than or equal to the token quantity in sub- token bucket module, if so, then from sub- token
Bucket module in take out with positioned at head of the queue I/O request required for new token quantity identical token and handle the I/O request, will
The I/O request exits I/O request set queue and returns to execution step（4）, otherwise sub- token bucket module end check.
2. the QoS methods to set up of the Ceph cloud storage systems according to claim 1 based on token bucket algorithm, its feature exist
In I/O request is stored in corresponding I/O request set queue by the service end to be included：
（1）I/O request is sequentially stored into I/O request set queue by the time sequencing that service end sends I/O request according to client；
（2）The dispatching priority of each I/O request of the service end in I/O request set queue to I/O request put in order into
The corresponding adjustment of row.
3. the QoS methods to set up of the Ceph cloud storage systems according to claim 1 based on token bucket algorithm, its feature exist
In the I/O request set queue includes read operation I/O request set queue and write operation I/O request set queue, described to read behaviour
Make the read operation that I/O request set queue is used to store I/O request, the write operation I/O request set queue is used to store I/O request
Write operation.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710986526.5A CN107579926B (en) | 2017-10-20 | 2017-10-20 | QoS setting method of Ceph cloud storage system based on token bucket algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710986526.5A CN107579926B (en) | 2017-10-20 | 2017-10-20 | QoS setting method of Ceph cloud storage system based on token bucket algorithm |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107579926A true CN107579926A (en) | 2018-01-12 |
| CN107579926B CN107579926B (en) | 2021-02-09 |
Family
ID=61038010
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710986526.5A Active CN107579926B (en) | 2017-10-20 | 2017-10-20 | QoS setting method of Ceph cloud storage system based on token bucket algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107579926B (en) |
Cited By (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108494703A (en) * | 2018-03-08 | 2018-09-04 | 腾讯科技（深圳）有限公司 | A kind of access frequency control method, device and storage medium |
| CN108848038A (en) * | 2018-08-30 | 2018-11-20 | 华为技术有限公司 | Flow managing method and token bucket node based on token bucket |
| CN109542643A (en) * | 2018-11-16 | 2019-03-29 | 北京金山云网络技术有限公司 | The restoration methods and device of message in a kind of OpenStack system |
| CN110417677A (en) * | 2019-07-29 | 2019-11-05 | 北京易捷思达科技发展有限公司 | A kind of QoS control method based on Ceph distributed storage Osd end data Recovery |
| CN110891089A (en) * | 2019-11-29 | 2020-03-17 | 中经社控股有限公司 | Token bucket algorithm-based Ceph block memory mirror image synchronization QoS optimization method |
| CN111290858A (en) * | 2020-05-11 | 2020-06-16 | 腾讯科技（深圳）有限公司 | Input/output resource management method, device, computer equipment and storage medium |
| CN111314238A (en) * | 2020-02-03 | 2020-06-19 | 网银在线（北京）科技有限公司 | Token management method and device, storage medium and electronic device |
| CN111343102A (en) * | 2020-02-03 | 2020-06-26 | ***股份有限公司 | Flow rate control method, server-side server, client-side server and system |
| CN112363673A (en) * | 2020-11-12 | 2021-02-12 | 深圳大普微电子科技有限公司 | Flow management method and system, solid state disk controller and solid state disk |
| CN112698790A (en) * | 2020-12-31 | 2021-04-23 | 浪潮云信息技术股份公司 | QoS control method and system for processing burst I/O in storage system |
| CN112751878A (en) * | 2020-12-30 | 2021-05-04 | 北京天融信网络安全技术有限公司 | Page request processing method and device |
| CN113296717A (en) * | 2021-07-26 | 2021-08-24 | 紫光恒越技术有限公司 | Optimization method of object storage distributed service quality, server and storage equipment |
| CN113472681A (en) * | 2020-03-30 | 2021-10-01 | 阿里巴巴集团控股有限公司 | Flow rate limiting method and device |
| CN113645150A (en) * | 2021-06-11 | 2021-11-12 | 中国电信股份有限公司 | Transmission rate control method, device, electronic equipment and readable storage medium |
| CN113709063A (en) * | 2021-08-05 | 2021-11-26 | 浪潮云信息技术股份公司 | Fine-grained QoS control method of Rados Gateway |
| CN114500406A (en) * | 2022-02-11 | 2022-05-13 | 浪潮云信息技术股份公司 | Cloud file storage fine-grained QoS control method and system |
| CN114760327A (en) * | 2022-03-30 | 2022-07-15 | 阿里巴巴（中国）有限公司 | Method and device for adjusting cloud disk resource allocation |
| US11979341B2 (en) | 2021-09-28 | 2024-05-07 | Red Hat, Inc. | Adaptive message queue timeouts for message queues related to storage systems |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104111800A (en) * | 2013-04-18 | 2014-10-22 | 阿里巴巴集团控股有限公司 | Virtual disk IO port scheduling method and scheduling device |
| CN105144161A (en) * | 2013-03-13 | 2015-12-09 | 亚马逊科技公司 | Token-based admission control for replicated writes |
| CN106462601A (en) * | 2014-03-31 | 2017-02-22 | 亚马逊科技公司 | Atomic writes for multiple-extent operations |
-
2017
- 2017-10-20 CN CN201710986526.5A patent/CN107579926B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105144161A (en) * | 2013-03-13 | 2015-12-09 | 亚马逊科技公司 | Token-based admission control for replicated writes |
| CN104111800A (en) * | 2013-04-18 | 2014-10-22 | 阿里巴巴集团控股有限公司 | Virtual disk IO port scheduling method and scheduling device |
| WO2014172475A1 (en) * | 2013-04-18 | 2014-10-23 | Alibaba Group Holding Limited | Method and device for scheduling virtual disk input and output ports |
| CN106462601A (en) * | 2014-03-31 | 2017-02-22 | 亚马逊科技公司 | Atomic writes for multiple-extent operations |
Non-Patent Citations (2)
| Title |
|---|
| YUBIN WU 等: "A New Disk I/O Scheduler Module with QoS Guarantee for Cloud Platform", 《2012 FOURTH INTERNATIONAL SYMPOSIUM ON INFORMATION SCIENCE AND ENGINEERING》 * |
| 小史: "ceph基于令牌桶算法的image的流量IOPS上限限制", 《HTTPS://BLOG.CSDN.NET/A1454927420/ARTICLE/DETAILS/72865016?LOCATIONNUM=7&FPS=1》 * |
Cited By (27)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108494703A (en) * | 2018-03-08 | 2018-09-04 | 腾讯科技（深圳）有限公司 | A kind of access frequency control method, device and storage medium |
| CN108848038B (en) * | 2018-08-30 | 2021-01-29 | 华为技术有限公司 | Token bucket-based traffic management method and token bucket node |
| CN108848038A (en) * | 2018-08-30 | 2018-11-20 | 华为技术有限公司 | Flow managing method and token bucket node based on token bucket |
| CN109542643A (en) * | 2018-11-16 | 2019-03-29 | 北京金山云网络技术有限公司 | The restoration methods and device of message in a kind of OpenStack system |
| CN110417677A (en) * | 2019-07-29 | 2019-11-05 | 北京易捷思达科技发展有限公司 | A kind of QoS control method based on Ceph distributed storage Osd end data Recovery |
| CN110417677B (en) * | 2019-07-29 | 2021-03-09 | 北京易捷思达科技发展有限公司 | QoS control method based on Ceph distributed storage Osd end data Recovery |
| CN110891089A (en) * | 2019-11-29 | 2020-03-17 | 中经社控股有限公司 | Token bucket algorithm-based Ceph block memory mirror image synchronization QoS optimization method |
| CN110891089B (en) * | 2019-11-29 | 2022-04-29 | 中国经济信息社有限公司 | Token bucket algorithm-based Ceph block memory mirror image synchronization QoS optimization method |
| CN111314238A (en) * | 2020-02-03 | 2020-06-19 | 网银在线（北京）科技有限公司 | Token management method and device, storage medium and electronic device |
| CN111343102A (en) * | 2020-02-03 | 2020-06-26 | ***股份有限公司 | Flow rate control method, server-side server, client-side server and system |
| CN111314238B (en) * | 2020-02-03 | 2023-12-05 | 网银在线（北京）科技有限公司 | Token management method and device, storage medium and electronic device |
| CN111343102B (en) * | 2020-02-03 | 2023-09-26 | ***股份有限公司 | Flow rate control method, server, client server and system |
| CN113472681A (en) * | 2020-03-30 | 2021-10-01 | 阿里巴巴集团控股有限公司 | Flow rate limiting method and device |
| CN111290858A (en) * | 2020-05-11 | 2020-06-16 | 腾讯科技（深圳）有限公司 | Input/output resource management method, device, computer equipment and storage medium |
| CN112363673A (en) * | 2020-11-12 | 2021-02-12 | 深圳大普微电子科技有限公司 | Flow management method and system, solid state disk controller and solid state disk |
| CN112751878A (en) * | 2020-12-30 | 2021-05-04 | 北京天融信网络安全技术有限公司 | Page request processing method and device |
| CN112751878B (en) * | 2020-12-30 | 2023-03-24 | 北京天融信网络安全技术有限公司 | Page request processing method and device |
| CN112698790A (en) * | 2020-12-31 | 2021-04-23 | 浪潮云信息技术股份公司 | QoS control method and system for processing burst I/O in storage system |
| CN113645150A (en) * | 2021-06-11 | 2021-11-12 | 中国电信股份有限公司 | Transmission rate control method, device, electronic equipment and readable storage medium |
| WO2022257425A1 (en) * | 2021-06-11 | 2022-12-15 | 天翼云科技有限公司 | Transmission rate control method and apparatus, electronic device, and readable storage medium |
| CN113296717A (en) * | 2021-07-26 | 2021-08-24 | 紫光恒越技术有限公司 | Optimization method of object storage distributed service quality, server and storage equipment |
| CN113709063A (en) * | 2021-08-05 | 2021-11-26 | 浪潮云信息技术股份公司 | Fine-grained QoS control method of Rados Gateway |
| US11979341B2 (en) | 2021-09-28 | 2024-05-07 | Red Hat, Inc. | Adaptive message queue timeouts for message queues related to storage systems |
| CN114500406A (en) * | 2022-02-11 | 2022-05-13 | 浪潮云信息技术股份公司 | Cloud file storage fine-grained QoS control method and system |
| CN114500406B (en) * | 2022-02-11 | 2024-05-14 | 浪潮云信息技术股份公司 | Cloud file storage fine granularity QoS control method and system |
| CN114760327A (en) * | 2022-03-30 | 2022-07-15 | 阿里巴巴（中国）有限公司 | Method and device for adjusting cloud disk resource allocation |
| CN114760327B (en) * | 2022-03-30 | 2024-03-26 | 阿里巴巴（中国）有限公司 | Cloud disk resource allocation adjusting method and device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107579926B (en) | 2021-02-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107579926A (en) | The QoS methods to set up of Ceph cloud storage systems based on token bucket algorithm | |
| US20230039191A1 (en) | Throttling queue for a request scheduling and processing system | |
| KR101086514B1 (en) | Continuous media priority aware storage scheduler | |
| CN104580396B (en) | A kind of method for scheduling task, node and system | |
| CN105718479B (en) | Execution strategy generation method and device under cross-IDC big data processing architecture | |
| CN106775493B (en) | A kind of storage control and I/O Request processing method | |
| US8612990B1 (en) | Prioritized rate scheduler for a storage system | |
| CN103299271A (en) | Concurrent request scheduling | |
| CN103491024B (en) | Job scheduling method and device for streaming data | |
| US20130061233A1 (en) | Efficient method for the scheduling of work loads in a multi-core computing environment | |
| CN109408229A (en) | A kind of dispatching method and device | |
| CN105022668B (en) | Job scheduling method and system | |
| CN109857535A (en) | The implementation method and device of task priority control towards Spark JDBC | |
| Glazebrook et al. | On the optimal allocation of service to impatient tasks | |
| CN101374109B (en) | Method and apparatus for scheduling packets | |
| CN114721818A (en) | Kubernetes cluster-based GPU time-sharing method and system | |
| CN111193802A (en) | Dynamic resource allocation method, system, terminal and storage medium based on user group | |
| Shu-Jun et al. | Optimization and research of hadoop platform based on fifo scheduler | |
| CN114327894A (en) | Resource allocation method, device, electronic equipment and storage medium | |
| CN117251275A (en) | Multi-application asynchronous I/O request scheduling method, system, equipment and medium | |
| CN111190541B (en) | Flow control method of storage system and computer readable storage medium | |
| CN116880968A (en) | Job scheduling method and scheduling system | |
| EP3550421B1 (en) | Storage controller and io request processing method | |
| CN115858169A (en) | Operation resource allocation method and device, electronic equipment and storage medium | |
| CN117056064A (en) | Resource allocation method, device, server, storage medium and program product |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20221012 Address after: 100094 107-2, 1st Floor, Building 1, East Yard, No. 10, Xibeiwang East Road, Haidian District, Beijing Patentee after: Beijing easy Star Technology Development Co.,Ltd. Address before: 210012 room 109, building 4, No. 168, software Avenue, Yuhuatai District, Nanjing City, Jiangsu Province Patentee before: NANJING EASYSTACK SOFTWARE TECHNOLOGY CO.,LTD. |