CN1519736A - Magnetic disk storage system - Google Patents
Magnetic disk storage system Download PDFInfo
- Publication number
- CN1519736A CN1519736A CNA031024637A CN03102463A CN1519736A CN 1519736 A CN1519736 A CN 1519736A CN A031024637 A CNA031024637 A CN A031024637A CN 03102463 A CN03102463 A CN 03102463A CN 1519736 A CN1519736 A CN 1519736A
- Authority
- CN
- China
- Prior art keywords
- passage
- high speed
- processing unit
- interface
- protocol processing
- 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
Images
Abstract
The system is at least composed of host interface module, cache, magnetic disk interface and disk array. The said host interface module includes host interface channel possessing exchanging function and more than one protocol process unit. Each external host through host interface channel is connected to each protocol process unit. Further, internal high-speed channel possessing exchanging function for information transmission between protocol process unit and magnetic disk interface is also included. Each protocol process unit through internal high-speed channel is connected to each magnetic disk interface, and each magnetic disk interface contains cache. Magnetic disk interface is connected to disk array. Working mode: distributed structure, stage treatment, sharing traffic load is adopted in the invention so as to eliminate bottleneck of bandwidth, raise speed of data transmission, protection mechanism, and reliability.
Description
Technical field
The present invention relates to the computer memory device field, particularly relate to a kind of disk storage system efficiently.
Background technology
Along with the continuous expansion of computer application field, the user is also improving day by day for the performance requirement of disk storage system.Wish that it has bigger capacity, faster speed and higher reliabilty and availability.At this challenge, people have found with a plurality of disc drivers and have formed the way that raid-arrays (RAID) obtain high transfer rate and big memory capacity.And now along with storage data capacity growth and more and more higher to the requirement of reliability, from the attached peripheral hardware of main frame, the centralized stores mode of sharing to many main frames develops storage system.
The structure of existing high capacity RAID disk array storage system mainly comprises the host interface 101, high-speed cache 102 and redundant array of inexpensive disks 1 03 3 parts that link to each other with external host referring to shown in Figure 1.The disk interface 104 that host interface 101, high-speed cache 102 and being used to connects redundant array of inexpensive disks 1 03 connects by same shared bus 105.
The concrete inner structure of this system can be referring to shown in Figure 2.Several host interface 101, high-speed cache 102 and disk interface 104 are articulated on same the inner shared bus 105.Host interface 101 is connected with main frame by the storage switch 201 of system outside.And each host interface 101 all contains independently protocol processor, is used for the request of access of processing host to disk.Its detailed structure is referring to shown in Figure 3, mainly comprise protocol processor 301 and object machine 302 in the host interface 101, contain starter 304 in the host adaptor 303 on the external host, protocol processor 301 is by the information interaction between object machine 302 and starter 304 realizations and the external host.
Yet the shortcoming of this structure storage system is: because the information interaction of all internal system all is to be undertaken by same shared bus, therefore be easy to form bandwidth bottleneck.Solve cache access bandwidth bottleneck problem, very big bus bandwidth and powerful processing power need be provided, then cause the system architecture complexity, cost is very high.
The host interface of system need be connected with the main frame of system outside by switch, otherwise each host interface can only be handled the host disk request of access that is attached thereto on the link, the use cost of system is increased, and the protocol processor utilization factor is low on the other hand, failure rate is high, when a protocol processor breaks down, the business of the host port that is attached thereto will be interrupted.
High-speed cache is concentrated and is placed; be unfavorable for management and data protection to high-speed cache; and All hosts interface and disk interface are all wanted and can be carried out high speed access to high-speed cache; form bandwidth bottleneck easily; solve cache access bandwidth bottleneck problem; need provide very big bus bandwidth and powerful processing power, and this can cause the complicated and cost increase of system architecture.
Summary of the invention
In view of this, the object of the present invention is to provide a kind of storage system efficiently, make it can improve plant factor, eliminate bandwidth bottleneck, improve data rate, and increase protection mechanism, improve system reliability.
A kind of disk storage system, at least comprise host interface module, high-speed cache, disk interface and disk array, wherein said host interface module comprises host interface passage and the more than one protocol processing unit with function of exchange, and each external host is connected with each protocol processing unit by the host interface passage; And further comprise and be used for the inner high speed passage that information transmission has function of exchange between protocol processing unit and disk interface; Each protocol processing unit is connected with each disk interface by the inner high speed passage, and is mounted with high-speed cache on each disk interface; Disk interface links to each other with disk array.
Be connected by dual path between the described host interface passage of this storage system and the protocol processing unit.
The described host interface module of this storage system comprises two separate host interface channels each other, and each host interface passage all is connected with all protocol processing unit.
Be connected by dual path between the described protocol processing unit of this storage system and the inner high speed passage, be connected by dual path between inner high speed passage and the disk interface.
The described inner high speed passage of this storage system is two inner high speed passages independently each other, and each inner high speed passage is connected with all protocol processing unit, and each inner high speed passage is connected with all disk interfaces.
The described disk interface of this storage system disposes in pairs, and the high-speed cache on the every pair of disk interface is mirror image each other, and the high-speed cache of mirror image is preserved a mirror image data mutually each other.
The logarithm of the described disk interface of this storage system is 4～8 pairs.
The described host interface passage of this storage system is the high speed switching network equipment.
The described inner high speed passage of this storage system is the high speed switching network equipment.
The described high speed switching network equipment of this storage system is a fibre channel media.
The described protocol processing unit of this storage system comprises the protocol processor that is used for the processing host request of access, carry out the object machine of information interaction with external host, reach the starter that carries out information interaction with disk interface, object machine and starter are connected on the protocol processor, object machine is connected with the host interface passage, and starter is connected with the inner high speed passage.
The described protocol processing unit of this storage system contains two object machines and starter, two object machines are connected on two paths between host interface passage and the protocol processing unit, and two protocol processors are connected on two paths between protocol processing unit and the inner high speed passage.
The described disk interface of this storage system contains two and carries out the object machine of information interaction with protocol processing unit, and two object machines are connected on two paths between inner high speed passage and the disk interface.
By such scheme as can be seen, a kind of disk storage system provided by the present invention adopts distributed frame, classification processing, traffic load sharing pattern, and each protocol processing unit is shared by the All hosts access port, improved the efficient of protocol processes, expansion host port number.Distributed cache has reduced the complicacy of the management and the data protection of high-speed cache, by disk interface unit internal bus high-speed cache is conducted interviews; do not take system bandwidth; reduce the pressure of system bus bandwidth, simplified system architecture, improved performance.
Description of drawings
Fig. 1 is the structural representation of storage system in the prior art;
Fig. 2 is the detailed structure view of storage system in the prior art;
Fig. 3 is the cut-away view of host interface in the prior art;
Fig. 4 is a disk storage system structural representation of the present invention;
Fig. 5 is a protocol processing unit structural representation of the present invention;
Fig. 6 is a high-speed cache distributed frame synoptic diagram of the present invention.
Embodiment
The present invention is further described in more detail below in conjunction with drawings and the specific embodiments.
The structure of high efficiency storage system of the present invention is referring to shown in Figure 4.Comprise: host interface module 401, inner high speed passage 402, disk interface module 403 and redundant array of inexpensive disks 1 03.
Wherein, comprise host interface passage 404 and protocol processing unit 405 in the host interface module 401.Host interface passage 404 is a high speed exchange network equipment, can link to each other with the interface of a plurality of dissimilar main frames.Host interface passage 404 adopts the dual path Redundancy Design simultaneously, as shown in Figure 4, physically adopt two separate host interface channels 404, each host interface passage 404 all connects all internal agreement processing units 405 separately, be host interface passage 404 with each protocol processing unit 405 between all pass through two separate paths and be connected, there is not public part between two paths, adopt the load sharing mode to work, simultaneously for guaranteeing the Redundancy Design of channel path, the bandwidth of every paths can satisfy whole business of this business unit, when a paths interrupted, business can all be tried a different way directly.
Main frame is handled to each protocol processing unit 405 through the host interface channel allocation the request of access of disk, can dispose a plurality of protocol processing unit 405 in the host interface module 401, do not concern one to one between the interface card of protocol processing unit 405 and external host that all protocol processing unit 405 all can be shared by All hosts; Adopt full redundancy load sharing mode to work, the request of access of each main frame is shared processing by each protocol processing unit 405, any one business unit fault, and the business of itself can be born by other business unit; And can improve the host access performance by increasing the configuration quantity of protocol processing unit 405, realize the dilatation of RAID system performance.
The inner structure of protocol processing unit 405 mainly comprises referring to shown in Figure 5: object machine 302, starter 304 and protocol processor 301.Owing to adopt the structure of dual path between business unit, so protocol processing unit 405 needs two groups of object machines 302 and starter 304 among the present invention, and the object machine 302 on the disk interface 406 that will mention hereinafter also should be two.Host adaptor 303 is the interface cards that are inserted in system's external host, when host access redundant array of inexpensive disks 1 03, starter 304 on the host adaptor 303 sends visit order, after object machine 302 in protocol processing unit 405 receives orders, visit order is sent to protocol processor 301, protocol processor 301 is high performance CPU, after by CPU the Host Command agreement being carried out Translation Processing, deliver to the starter 304 of protocol processing unit 405, at the object machine 302 that sends to disk interface 406 through inner high speed passage 402, carry out read-write operation by 406 pairs of disks of disk interface.
In addition, also can adopt dual path to be connected between main frame and the host interface passage 404 among the present invention, its annexation and working method also with above-mentioned host interface passage 404 and protocol processing unit 405 between the dual path structure identical, just in each main frame, need to be equipped with accordingly two host adaptors 303.
Referring to Fig. 4, inner high speed passage 402 is to realize clog-free redundant channel of communicating by letter between host interface module 401 and the disk interface module 403.Connected mode and inner structure between it and each business unit are all similar with host interface passage 404.Adopt two independently inner high speed passages 402 equally; every inner high speed passage 402 all separately with All hosts interface module 401 in protocol processing unit 405 link to each other with disk interface 406 in the disk interface module 403; separate between the path, load sharing, and can replace, but protection mutually between realizing route.Inner high speed passage 402 is also basic identical with the inner structure of host interface passage 404, all adopt high speed exchange network equipment, as fibre channel media etc., inner high speed passage 402 and host interface passage 404 adopt the high speed network board that designs and produces voluntarily in the present invention.
The quantity of disk interface 406 can according to how many configurations arbitrarily of number of disks, do not concern one to one between the protocol processing unit 405 in disk interface 406 and the host interface module 401, scale and the requirement of host disk access performance according to redundant array of inexpensive disks 1 03, can increase the quantity of disk interface unit, to realize the dilatation of redundant array of inexpensive disks 1 03.The disk interface that only shows a pair of disk interface 406 compositions among Fig. 4 is right, contains the right disk interface module 403 of a plurality of disk interfaces referring to shown in Figure 6.Three disk interfaces drawn among Fig. 6 to 601, and 4～8 disk interfaces are proper to 601 in concrete enforcement.High-speed cache 407 adopts distributed frames, is distributed in each disk interface unit, is used for the buffering between the mechanically actuated of main frame high speed access performance requirement and disk low speed, to realize the hyperdisk visit.The visit of 406 pairs of high-speed caches of disk interface does not take system bandwidth by its internal bus.
When main frame sends redundant array of inexpensive disks 1 03 visit order, give protocol processing unit 405 by the host interface channel allocation, after protocol processing unit 405 processing, through inner high speed passage 402 instruction is sent to disk interface module 403, carry out read-write operation by 403 pairs of redundant array of inexpensive disks 1 of disk interface module 03 again.For the write command operation, the data that disk interface 406 is received are kept at the high-speed cache 407 that this disk interface 406 articulates earlier, by disk interface 406 data are write redundant array of inexpensive disks 1 03 again; For the operation that reads instruction, disk interface 406 is read data earlier from redundant array of inexpensive disks 1 03, be kept at the high-speed cache 407 that is articulated, and sends data to external host again.
Host interface passage and inner high speed passage adopt fiber channel protocol work in embodiments of the present invention, and protocol processing unit and disk interface adopt the SCSI agreement.When certain business unit of system or access path break down, associated business unit will be received alarm, according to warning information, automatically business processing is transferred to the business module or the access path of operate as normal simultaneously, to realize fault isolation and professional automatic recovery.
Technical scheme of the present invention adopts distributed frame, classification processing, traffic load sharing pattern.System architecture comprises host interface, inner high speed passage, disk interface.Can realize that architecture do not have Single Point of Faliure, the system failure is isolated automatically and is recovered automatically.
Claims (13)
1, a kind of disk storage system comprises host interface module, high-speed cache, disk interface and disk array at least, it is characterized in that:
Described host interface module comprises host interface passage and the more than one protocol processing unit with function of exchange, and each external host is connected with each protocol processing unit by the host interface passage;
Further comprise and be used for the inner high speed passage that information transmission has function of exchange between protocol processing unit and disk interface;
Each protocol processing unit is connected with each disk interface by the inner high speed passage, and is mounted with high-speed cache on each disk interface;
Disk interface links to each other with disk array.
2, storage system according to claim 1 is characterized in that being connected by dual path between described host interface passage and the protocol processing unit.
3, storage system according to claim 2 is characterized in that described host interface module comprises two separate host interface channels each other, and each host interface passage all is connected with all protocol processing unit.
4, storage system according to claim 1 is characterized in that being connected by dual path between described protocol processing unit and the inner high speed passage, is connected by dual path between inner high speed passage and the disk interface.
5, storage system according to claim 4, it is characterized in that described inner high speed passage is two inner high speed passages independently each other, each inner high speed passage is connected with all protocol processing unit, and each inner high speed passage is connected with all disk interfaces.
6, according to claim 1 or 4 described storage systems, it is characterized in that described disk interface disposes in pairs, the high-speed cache on the every pair of disk interface is mirror image each other, and the high-speed cache of mirror image is preserved a mirror image data mutually each other.
7, storage system according to claim 6, the logarithm that it is characterized in that described disk interface is 4～8 pairs.
8, storage system according to claim 1 is characterized in that described host interface passage is the high speed switching network equipment.
9, storage system according to claim 1 is characterized in that described inner high speed passage is the high speed switching network equipment.
10, according to Claim 8 or 9 described storage systems, it is characterized in that described high speed switching network equipment is a fibre channel media.
11, storage system according to claim 1, it is characterized in that described protocol processing unit comprises the protocol processor that is used for the processing host request of access, carry out the object machine of information interaction with external host, reach the starter that carries out information interaction with disk interface, object machine and starter are connected on the protocol processor, object machine is connected with the host interface passage, and starter is connected with the inner high speed passage.
12, storage system according to claim 2, it is characterized in that being connected by dual path between protocol processing unit and the inner high speed passage, protocol processing unit contains two object machines and starter, two object machines are connected on two paths between host interface passage and the protocol processing unit, and two protocol processors are connected on two paths between protocol processing unit and the inner high speed passage.
13, storage system according to claim 12 is characterized in that described disk interface contains two and carries out the object machine of information interaction with protocol processing unit, and two object machines are connected on two paths between inner high speed passage and the disk interface.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB031024637A CN1296845C (en) | 2003-01-24 | 2003-01-24 | Magnetic disk storage system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB031024637A CN1296845C (en) | 2003-01-24 | 2003-01-24 | Magnetic disk storage system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1519736A true CN1519736A (en) | 2004-08-11 |
| CN1296845C CN1296845C (en) | 2007-01-24 |
Family
ID=34281736
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB031024637A Expired - Lifetime CN1296845C (en) | 2003-01-24 | 2003-01-24 | Magnetic disk storage system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN1296845C (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1306718C (en) * | 2004-12-23 | 2007-03-21 | 烽火通信科技股份有限公司 | A novel disk protection method |
| CN100437457C (en) * | 2005-03-03 | 2008-11-26 | 富士通株式会社 | Data storage system and data storage control apparatus |
| US7552249B2 (en) | 2004-12-24 | 2009-06-23 | Fujitsu Limited | Direct memory access circuit and disk array device using same |
| CN102484603A (en) * | 2009-08-28 | 2012-05-30 | 惠普发展公司，有限责任合伙企业 | Automatic redundant logical connections |
| CN102929801A (en) * | 2012-10-25 | 2013-02-13 | 华为技术有限公司 | Disk addressing method and device |
| WO2018119742A1 (en) * | 2016-12-28 | 2018-07-05 | 华为技术有限公司 | Method, device and system for forwarding packet in nvme over fabric |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5634033A (en) * | 1994-12-16 | 1997-05-27 | At&T Global Information Solutions Company | Disk array storage system architecture for parity operations simultaneous with other data operations |
| AU7714700A (en) * | 1999-09-22 | 2001-04-24 | Netcell Corp. | Raid controller system and method with ata emulation host interface |
| TW484723U (en) * | 2000-12-20 | 2002-04-21 | Areca Technology Corp | Disk array controller |
| JP2003015826A (en) * | 2001-07-04 | 2003-01-17 | Hitachi Ltd | Shared memory copy function in disk array controller |
-
2003
- 2003-01-24 CN CNB031024637A patent/CN1296845C/en not_active Expired - Lifetime
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1306718C (en) * | 2004-12-23 | 2007-03-21 | 烽火通信科技股份有限公司 | A novel disk protection method |
| US7552249B2 (en) | 2004-12-24 | 2009-06-23 | Fujitsu Limited | Direct memory access circuit and disk array device using same |
| CN100437457C (en) * | 2005-03-03 | 2008-11-26 | 富士通株式会社 | Data storage system and data storage control apparatus |
| CN102484603A (en) * | 2009-08-28 | 2012-05-30 | 惠普发展公司，有限责任合伙企业 | Automatic redundant logical connections |
| CN102484603B (en) * | 2009-08-28 | 2015-09-09 | 惠普发展公司，有限责任合伙企业 | Create the method and apparatus of redundancy logic connection and store automated system equipment |
| US9774675B2 (en) | 2009-08-28 | 2017-09-26 | Entit Software Llc | Automatic redundant logical connections |
| CN102929801A (en) * | 2012-10-25 | 2013-02-13 | 华为技术有限公司 | Disk addressing method and device |
| CN102929801B (en) * | 2012-10-25 | 2016-06-22 | 华为技术有限公司 | A kind of method and apparatus for disk addressing |
| WO2018119742A1 (en) * | 2016-12-28 | 2018-07-05 | 华为技术有限公司 | Method, device and system for forwarding packet in nvme over fabric |
| US10585628B2 (en) | 2016-12-28 | 2020-03-10 | Huawei Technologies Co., Ltd. | Packet forwarding method, device, and system in NVME over fabric |
| US10983731B2 (en) | 2016-12-28 | 2021-04-20 | Huawei Technologies Co., Ltd. | Packet forwarding method, device, and system in non-volatile memory express over fabric |
Also Published As
| Publication number | Publication date |
|---|---|
| CN1296845C (en) | 2007-01-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP1018686B1 (en) | Disk array controller | |
| US6684295B2 (en) | Disk array control device with two different internal connection systems | |
| US6985989B2 (en) | Storage apparatus having plural controller including interconnected memories | |
| EP1595363B1 (en) | Scsi-to-ip cache storage device and method | |
| CN100458677C (en) | Apparatus and method to configure one or more storage arrays | |
| US6009481A (en) | Mass storage system using internal system-level mirroring | |
| KR100740080B1 (en) | Data storage system and data storage control apparatus | |
| US8032610B2 (en) | Scalable high-speed cache system in a storage network | |
| US9250687B1 (en) | High performance flexible storage system architecture | |
| US20090177805A1 (en) | Dual port serial advanced technology attachment (sata ) disk drive | |
| CN101739220A (en) | Method for designing multi-controller memory array | |
| US20050177681A1 (en) | Storage system | |
| JPH08328760A (en) | Disk array device | |
| US20140223097A1 (en) | Data storage system and data storage control device | |
| CN1296845C (en) | Magnetic disk storage system | |
| US6868438B2 (en) | Storage system, disk control cluster and a method of increasing of disk control cluster | |
| EP1609071A2 (en) | Data storage system | |
| CN100351766C (en) | Disk array system | |
| US7426658B2 (en) | Data storage system and log data equalization control method for storage control apparatus |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CX01 | Expiry of patent term | ||
| CX01 | Expiry of patent term |
Granted publication date: 20070124 |