Electrical computers and digital processing systems: memory – Storage accessing and control – Specific memory composition
Reexamination Certificate
2002-10-03
2004-10-19
Sparks, Donald (Department: 2187)
Electrical computers and digital processing systems: memory
Storage accessing and control
Specific memory composition
C711S154000, C711S165000
Reexamination Certificate
active
06807605
ABSTRACT:
FIELD OF THE INVENTION
The invention relates to data storage systems and methods. More particularly, the invention relates to a system and method of managing data within a computer accessible storage array.
BACKGROUND OF THE INVENTION
The use of an array of disks for computer-based data storage is known. One category of disk arrays is referred to as Redundant Array of Inexpensive Drives (RAID). Within a RAID system, varying levels of data storage redundancy are utilized to enable reconstruction of stored data in the event of data corruption or disk failure. These various types of redundant storage strategies are referred to as RAID levels.
For example, RAID level 1, also referred to as “the mirror method”, defines data which is stored with complete redundancy, typically permitting independent, simultaneous access to all copies of the data set. RAID level 6, also referred to as “the parity method”, defines data storage which utilizes bit-parity information, generated by way of an Exclusive-OR operation, to create a parity data set. The parity data set may be used to reconstruct original data (i.e., now corrupt) using another Exclusive-OR operation. In comparison, RAID level 1 provides relatively fast, simultaneous access to multiple copies of the same data set; while RAID level 6 provides for greater storage media efficiency than that of RAID level 1. Accordingly, RAID level 1 is considered a “high performance” or “high” RAID level as compared to level 6.
Due to the higher performance of RAID level 1 storage, it is desirable to keep the most frequently accessed data within RAID level 1, to the extent that physical storage resources permit. Toward this goal, some RAID management systems employ techniques in which more frequently accessed data is moved to RAID level 1 storage, while less frequently accessed data is shifted to RAID level 6 (or other RAID levels). RAID data storage is discussed in detail in U.S. Pat. Nos. 5,392,244 and 5,664,187, which are incorporated herein by reference.
The vast amounts of data that are stored in many computer systems, as well as the ever-growing demand to access that data, has pushed system developers to seek ways of providing fast, multiple-user access to the data. One technique utilizes a system of pointers (i.e, mapping) to point at the data within the storage media. Software applications make use of the pointers to access relevant data within the storage system. A system of pointers may be used in combination with various RAID levels of storage.
The RAID data storage can keep track of the most recently written data blocks. This set of most recently written blocks is called the “Write Working Set”. The pointers to these data blocks are kept in memory. The data blocks that make up the Write Working Set are kept in the physical storage. The Write Working Set does the mapping of the most recently written blocks to the physical storage. The data blocks that comprise this set can be in both RAID levels, this is, some of the blocks that are part of the Write Working Set can be in RAID level Six and the other part can be in RAID level One. Of course, it could be the case that all the data blocks are either in RAID level Six or in RAID level One. Whatever the case, the purpose of keeping the Write Working Set is to realize a list of pointers to the most recently written data blocks, so as to facilitate future migrations of those data blocks to the RAID level 1 storage. This gives the user better response time when accessing these data blocks again. The use of the Write Working Set is already in use and it is not the innovation being described here in this document.
One of the features of the RAID data storage is the possibility to preserve data as it was at some point in time. This feature is called “snapshot”, from now on, in this document, the snapshot term will be used to denote this feature. The data to be preserved is said to be “snapped”. After the data is snapped, it can be considered protected against future changes. This allows the user to take a “snapshot” of some data to be preserved. If the data is updated (written to) in the future then the snapped data is preserved and a new copy with the updated data is also stored in the system. This way, the user can have the snapped (i.e. the original) data and the updated data both in the RAID data storage. The snapping of the data is independent of the RAID level the data is in. When some data is snapped the data will be preserved regardless of the RAID level the data is in. The system of pointers to the snapped data blocks is known as the “Snapshot Maps”. This idea as well as the Write Working Set described above is not an innovative idea and it is not the idea described here in this patent disclosure.
However, when system user operations require that the snapped data be updated (written to), it is necessary to follow a sequence of steps to ensure the snapped data will be preserved and the updated data will be written to. The procedure used to preserve the original data and store the updated data separately in the physical storage is called a “divergence”. Both, the snapped and updated data will be kept in the RAID data storage. The steps performed in divergence are: 1) The data to be written to is stored in memory. 2) The snapped data is read, usually as a data block bigger than the data to be written to. 3) The data block with the snapped data is merged with the new data in memory and the combined data is then considered the updated data. 4) A new pointer to the updated data is created in the general tables that hold the maps for all the RAID level One and RAID level Six data. The pointer to the snapped (i.e. original) data is kept in the snapshot maps. 5) The updated data is written to in the physical storage. At the end of these first four steps the snapped data is still in the physical RAID data storage as well as the new, updated data. This procedure (the divergence), to ensure that the snapped data and the new data are kept in the physical storage, is performed in the foreground when the users writes to the data and results in undesirable delays from the perspective of the system users.
Therefore, it is desired to provide a data management system in which the update of snapped data is performed with a reduced impact on user access to the data stored within the system.
SUMMARY OF THE INVENTION
The invention provides an improved system and method for managing the divergences of snapped data within a RAID storage system, in coordination with the data access requests of system users. The preferred embodiment of the invention provides divergence in the background of the data blocks that are snapped AND are in a write working set (i.e., a map to those data spaces most recently written to). Therefore, the preferred embodiment of the invention involves predicting the divergences of the snapped data blocks by using the write working set.
One embodiment of the invention provides a system for managing a data storage array, comprising a plurality of data storage disks configured as an array, and a RAID controller coupled to each of the data storage disks within the array and configured to access data within the array and to measure the rate at which data within the array is being accessed by a user application, the RAID controller further including a memory and a user interface coupled to the RAID controller, the RAID controller being further configured to store a set of pointers within the memory, the pointers respectively corresponding to blocks of data within the array, each pointer being accessible by the user application, the RAID controller being further configured to anticipate a data modification operation to a particular block of data in the array responsive to the measuring and to selectively copy the particular block of data to a different block of data within a different RAID level storage location within the array responsive to the anticipating.
Another embodiment of the invention provides a method of managing a data storage system, comprising providing an array of data disks configured
Daniels Rodger
Navarro Guillermo
Umberger David
Hewlett--Packard Development Company, L.P.
Peugh Brian R.
Sparks Donald
LandOfFree
Managing a data storage array, a data storage system, and a... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Managing a data storage array, a data storage system, and a..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Managing a data storage array, a data storage system, and a... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3310720