26. RAID

Question 1

Describe workshop papers.

Answer 1

Short (5-6 pages), usually containing only a provocative argument, system design, or very preliminary results

Ex: “Hierarchical Filesystems Are Dead”

Question 2

Describe conference papers

Answer 2

Long (12-14 pages), enough space to describe and evaluate a complete novel system

Example: The Scalable Commutativity Rule

Question 3

Describe journal papers

Answer 3

Longer (>14 pages), usually a published conference paper with extra material (frequently all of the unnecessary results that they removed to make the conference page limit)

Advice: read the conference paper

Question 4

What are the 5 kinds of papers discussed in class?

Answer 4

1. (Big) idea papers
2. Problem papers
3. Data papers
4. New technology papers
5. Wrong way papers

Question 5

What do (big) idea papers cover?

Answer 5

Presents a new approach to an existing problem or a new idea about how to build systems.

Should convince you that the solution is (1) new, (2) works, and (3) is useful

Question 6

What do problem papers cover?

Answer 6

Presents a new problem and, usually, some ideas about how to solve it.

Should convince you that the problem is (1) new, (2) matters, and (3) that there are some ways to solve it

Question 7

What do data papers cover?

Answer 7

Presents novel analysis or analysis of a novel data set that produces interesting insights.

Should convince you that the results are useful to the design of future systems

Question 8

What do new technology papers cover?

Answer 8

Describe some new hardware capability or device feature and why it’s interesting.

Should convince you that the hardware can be used to build better systems

Question 9

What do wrong way papers cover?

Answer 9

Argue that the community is solving an existing problem incorrectly. Frequently these are workshop-level papers and eventually lead to idea papers.

Should be able to convince you that everyone else is confused and misguided (Good luck!)

Question 10

What does RAID stand for?

Answer 10

Redundant Arrays of Inexpensive Disks

Question 11

What kind of paper is the RAID paper?

Answer 11

A big idea paper.

It spawned a commonly-used technology, an entire industry, and lots of similar approaches

Question 12

What is the big idea of the RAID paper?

Answer 12

Several cheap things can be better than one expensive thing

Question 13

Where else do we see this idea applied today?

Answer 13

Multicore processors, Google using thousands of Dell computers in network, and crowdsourcing

Question 14

What is the problem that the RAID paper identifies?

Answer 14

Computer CPUs are getting faster, computer memory is getting faster, and hard drives are not keeping up

While we can imagine improvements in software file systems via buffering for near term I/O demands, we need innovation to avoid an I/O crisis

Question 15

What is the problem with the RAID solution to the cost of SLEDs?

Answer 15

Many cheap things fail much more frequently than one expensive thing.

This means there needs to be a plan in place to handle failures

Question 16

How does RAID 1 work?

Answer 16

RAID 1 is mirroring.

There are two duplicate disks.

Writes must go to both disks, reads can come from either.

Performance: better for reads.

Capacity: Unchanged

Question 17

How does RAID 2 work?

Answer 17

Byte-level striping, single error disk. (meaning data written to disk is spread out byte by byte across multiple disks)

Hamming codes are used to detect failures and correct errors

Most reads and writes require all disks

Capacity: “improved”

Question 18

How does RAID 3 work?

Answer 18

RAID 3 only corrects errors, since disks can detect when they fail.

Byte-level striping, single parity disk. (for each byte, there’s an extra bit of information stored somewhere on another disk)

Most reads and writes require all disks.

Capacity: “improved”

Question 19

How does RAID 4 work?

Answer 19

Block-level striping, single parity disk.

Better distribution of reads between disks due to larger stripe size, but all writes must access the parity disk.

Performance: improved for reads

Question 20

How does RAID 5 work?

Answer 20

Block-level striping.

Multiple parity disks.

Better distribution of writes between disks.

Performance: improved for writes

Question 21

What is RAID 0?

Answer 21

Each disk stores half of the data

No error correction or redundancy

Performance: fantastic!

Capacity: fantastic!

Redundancy: ZERO!

Question 22

What is RAID 0?

Answer 22

RAID 0 is not RAID.

Each disk stores half of the data (2 disks total)

No error correction or redundancy

Performance: fantastic!

Capacity: fantastic!

Redundancy: ZERO!

Question 23

What is block-level striping?

Answer 23

Writes to disk writes blocks to the many disks in RAID one block at a time, so that a multi-block write happens across multiple disks.

Question 24

What is block-level striping?

Answer 24

Writes to disk writes blocks to the many disks in RAID one block at a time, so that a multi-block write happens across multiple disks.

Question 25

How many failures can RAID arrays tolerate?

Answer 25

One (or more) disks. The more failures you want your RAID arrays to handle, the more disk space you need to set aside for keeping track of failures (thus reducing your disk’s capacity).

The challenge is identifying the optimal balance between failure tolerance and disk capacity.

Question 26

What two things happen when a disk fails in a RAID system?

Answer 26

1. Once one (or several) disks fail, the array is vulnerable to data loss.
2. An administrator must replace the disk(s) and then rebuild the array