RAID and Indexing: Understanding Data Storage and Access with RAID and Index Structures, Study Guides, Projects, Research of Introduction to Database Management Systems

Download RAID and Indexing: Understanding Data Storage and Access with RAID and Index Structures and more Study Guides, Projects, Research Introduction to Database Management Systems in PDF only on Docsity!

Announcements

•^

Today– RAID– Begin Indexes

-^

Program 1 due Friday– Office Hours today 2-3 pm– I’ll have limited email contact over the weekend– later today I’ll give info for turning in the program

RAID

Redundant Arrays of Inexpensive Disks

•^

Goal of RAID is to even out rates of diskimprovements (small) w/ those in RAM and CPU

-^

RAID use multiple physical disks to behave as asingle logical disk

Naïve Striping Reduces Reliability

•^

Likelihood of failure increases w/ # of disks– Mirroring, error correcting codes are used to increase

reliability at the expense of speed

•^

But is this statement correct?– (from Section 13.10.1)“For an array of

n disks

, the likelihood of failure is

n

times as much as that for one disk. Hence, if theMTTF of a disk drive is 200,000 hours (22.8years), that of a bank of 100 disk drives becomesonly 2000 hours (83 days)”

RAIDOrganizationsbalance speedand reliability

“If you don’t find it in the index, look verycarefully through the whole catalog”

• Sears, Roebuck and Co. consumers’ Guide, 1897

Indexes provide alternative access paths

Query: Find record for student “Troy Allen”

Index on“name”

Step 1: query the index for the RID for the record (hopefully a few IOs)Step 2: query the buffer manager for the appropriate block (1 IO)

RID = (3438, 9)

“Troy Allen”

Some Considerations

•^

What is the organization of the underlying file– Eg, is it ordered on the search key?

-^

Are the values of the indexing field unique (ie, isthe indexing field a key field)?

-^

How are the data entries of the index organized?– Example: make index a hashed file on index field

where each record contains (value, RID) pairs

Some Definitions

-^

primary index: an index on the ordering key field of aordered file

-^

secondary index: an index on any non-ordering field ofthe file

-^

clustered index: an index whose data entries are orderedin the same way as the underlying file

-^

dense index: has an index entry for every search keyvalue (and hence every record) in the data file.

-^

sparse index: has index entries for only some of thesearch values

Cost of Maintaining a Clustered

Primary Index

•^

Inserting of record in the ordered file (alreadyexpensive) may require significant updates tothe index– Why is this?

one way to handlethe “insert” problemof ordered files

More Secondary Indexes

•^

What if the indexing field is not a key field?– Option 1: Keep index entry for each record, so we will

have multiple index entries for each value

• Option 2: Have one record / value and store a

“RID list” for each value. Thus the index records arevariable length records• <‘Jim’, { (389, 3), (3239,30), (193, 78) } >

• Option 3: Mixed type of index records (next slide)