1. Why LOOK Exists
Before understanding LOOK, let's revisit a limitation of SCAN.
In SCAN, the disk head continues moving all the way to the physical end of the disk before reversing direction, even if there are no pending requests near that end.
Consider a disk with tracks numbered:
0 ------------------------------------ 199
Suppose the farthest request on the right side is:
183
In SCAN, the head still performs:
183 → 199
before reversing.
The Problem
This movement does not service any request.
It is simply extra travel.
As a result:
Additional seek time is incurred
Overall head movement increases
Disk efficiency decreases
What OS Designers Wanted
A scheduling policy that:
Retains the fairness of SCAN
Eliminates starvation
Avoids unnecessary movement to disk ends
Improves seek performance
This led to the development of LOOK.
Key Insight
Instead of moving to the physical end of the disk, LOOK reverses direction as soon as it reaches the last pending request in the current direction.
2. What is LOOK?
LOOK is a disk scheduling algorithm in which the disk head moves in one direction, services all requests along the way, and reverses direction at the last request rather than continuing to the end of the disk.
Definition
The disk head "looks ahead" to determine whether any requests remain in the current direction.
If no requests remain, it immediately reverses direction.
Core Principle
Move in one direction
↓
Service requests
↓
Reach last request
↓
Reverse direction
↓
Continue servicing
Why the Name "LOOK"?
Because the algorithm checks ("looks") ahead before deciding whether to continue moving.
Key Insight
LOOK is essentially an optimized version of SCAN that avoids unnecessary travel.
3. How LOOK Works Internally
Unlike SCAN:
Move Right
↓
Reach Disk End
↓
Reverse
LOOK performs:
Move Right
↓
Reach Last Request
↓
Reverse
Internal Strategy
At any point:
Identify all pending requests.
Continue moving in the current direction.
Service requests in sorted order.
When no further request exists in that direction:
Reverse immediately.
Continue servicing requests in the opposite direction.
Key Insight
LOOK uses request locations—not disk boundaries—to determine movement.
4. Numerical Example
Given
Initial Head Position:
53
Initial Direction:
Right
Request Queue:
98, 183, 37, 122, 14, 124, 65, 67
Disk Range:
0 – 199
Step 1: Divide Requests
Requests to the right of 53:
65, 67, 98, 122, 124, 183
Requests to the left of 53:
37, 14
Step 2: Sort Requests
Right Side (Ascending):
65 → 67 → 98 → 122 → 124 → 183
Left Side (Descending after reversal):
37 → 14
Step 3: Determine Service Order
Since the head initially moves right:
53 → 65 → 67 → 98 → 122 → 124 → 183
At this point:
No more requests exist on the right.
Therefore:
Reverse Direction
Continue servicing:
183 → 37 → 14
Final Service Sequence
53 → 65 → 67 → 98 → 122 → 124 → 183 → 37 → 14
Important Observation
Unlike SCAN:
53 → ... → 183 → 199 → ...
LOOK performs:
53 → ... → 183
Reverse Immediately
No unnecessary trip to track 199 occurs.
5. Visualization of LOOK Movement
0 --------------------------------------------------- 199
14 37 53 65 67 98 122 124 183
Head
Move Right →
53 → 65 → 67 → 98 → 122 → 124 → 183
Reverse ←
183 → 37 → 14
Key Observation
The head only travels where requests exist.
6. Total Head Movement Calculation
Step-by-Step
53 → 65 = 12
65 → 67 = 2
67 → 98 = 31
98 → 122 = 24
122 → 124 = 2
124 → 183 = 59
183 → 37 = 146
37 → 14 = 23
Total Head Movement
12 + 2 + 31 + 24 + 2 + 59 + 146 + 23
= 299 Cylinders
Final Answer
Total Head Movement = 299 Cylinders
7. Why LOOK Performs Better Than SCAN
The primary improvement comes from eliminating unnecessary movement.
In SCAN
After servicing the last request:
183 → 199
The head continues to the disk boundary.
In LOOK
The head stops at:
183
and reverses immediately.
Saved Movement
199 - 183 = 16 cylinders
Since SCAN must travel this extra distance, LOOK achieves lower total movement.
Comparison
SCAN = 331 Cylinders
LOOK = 299 Cylinders
Improvement
331 - 299 = 32 Cylinders Saved
Key Insight
LOOK retains SCAN's behavior while removing waste.
8. Fairness and Starvation Analysis
Does LOOK Cause Starvation?
No
Why?
The head continues sweeping in both directions.
Every pending request eventually lies in its path.
Result
Guaranteed Service
for all requests.
Key Insight
LOOK preserves the fairness of SCAN while improving efficiency.
9. Performance Characteristics
9.1 Seek Time
Lower than SCAN because:
No unnecessary movement to disk ends
9.2 Throughput
Higher than SCAN.
Less movement means:
More requests serviced
Less wasted time
9.3 Response Time
Generally good and predictable.
9.4 Fairness
High.
Requests are serviced in both directions.
Key Insight
LOOK provides a strong balance between efficiency and fairness.
10. Advantages of LOOK
10.1 Reduced Head Movement
Avoids unnecessary travel to disk boundaries.
10.2 Better Performance
Lower seek time compared to SCAN.
10.3 No Starvation
Every request is eventually serviced.
10.4 Fair Scheduling
Requests on both sides receive service.
10.5 Simple Improvement Over SCAN
Only a small modification is required.
Key Insight
LOOK achieves most of the benefits of SCAN with better efficiency.
11. Disadvantages of LOOK
11.1 Direction-Based Waiting
Requests on the opposite side may still wait until reversal.
11.2 Slight Positional Bias
Requests near the current direction are serviced sooner.
11.3 Not Globally Optimal
Total movement may still exceed SSTF in some cases.
Key Insight
LOOK improves SCAN but does not eliminate all scheduling inefficiencies.
12. Real-World Analogy
Imagine an elevator in a building.
SCAN Behavior
Go to top floor
Even if nobody requested it
Then come down
LOOK Behavior
Go only as high as the highest requested floor
Then reverse
Result
Less travel.
Faster service.
Key Insight
LOOK behaves like an intelligent elevator that avoids unnecessary trips.
13. LOOK vs SCAN
| Feature | SCAN | LOOK |
|---|---|---|
| Reaches Disk End | Yes | No |
| Reversal Point | Disk Boundary | Last Request |
| Head Movement | Higher | Lower |
| Seek Time | Higher | Lower |
| Fairness | High | High |
| Starvation | No | No |
| Efficiency | Good | Better |
14. LOOK vs SSTF vs SCAN
| Feature | SSTF | SCAN | LOOK |
|---|---|---|---|
| Strategy | Nearest Request | Full Sweep | Optimized Sweep |
| Seek Time | Lowest (Usually) | Moderate | Lower than SCAN |
| Fairness | Poor | Good | Good |
| Starvation | Possible | No | No |
| Predictability | Low | Good | Good |
| Complexity | Moderate | Moderate | Moderate |
15. LOOK at a Glance
| Property | LOOK |
|---|---|
| Full Form | Look Disk Scheduling |
| Direction | Both Directions |
| Reversal Point | Last Pending Request |
| Starvation | No |
| Fairness | High |
| Seek Time | Lower than SCAN |
| Efficiency | Better than SCAN |
| Complexity | Moderate |
| Practical Usage | Modern Disk Scheduling Variants |
Final Insight
LOOK is an optimized version of SCAN that eliminates unnecessary movement to the physical ends of the disk. Instead of sweeping all the way to track 0 or the maximum track number, the head reverses as soon as the last pending request in the current direction is serviced. This reduces seek time, improves throughput, preserves fairness, prevents starvation, and makes LOOK one of the most practical and efficient disk scheduling algorithms.