Class Note for CMPSCI 377 at UMass(21)
Class Note for CMPSCI 377 at UMass(21)
Popular in Course
Popular in Department
This 3 page Class Notes was uploaded by an elite notetaker on Friday February 6, 2015. The Class Notes belongs to a course at University of Massachusetts taught by a professor in Fall. Since its upload, it has received 20 views.
Reviews for Class Note for CMPSCI 377 at UMass(21)
Report this Material
What is Karma?
Karma is the currency of StudySoup.
You can buy or earn more Karma at anytime and redeem it for class notes, study guides, flashcards, and more!
Date Created: 02/06/15
CMPSCI 377 Operating Systems Fall 2005 Lecture 16 November 29 Lecturer Emery Berger Scribe Sam Farringtzm Today In this class we learned about scheduling algorithms and their many advantages and disadvantages 161 Reducing IO Time IO Time Seek Time Rotational Delay Transfer Time Rotational delay is already very low Transfer time can be reduced Latency can be reduced Ways to reduce latency Defragment a fragmented drive Build smaller disks Build faster disks lncrease sector size change the disk layout lmprove disk scheduling 162 Why Fragmentation Occurs Fragmentation occurs because the le system tries to layout data contiguously As the disk lls up there Will be scattered empty chunks When the disk reaches 50 80 capacity these chunks become the only space left to Write data to Some le systems can avoid fragmentation by using a process Which is equivalent to garbage collection on the disk 163 Disk scheduling algorithms The goal of a disk scheduling algorithm is to order disk accesses to reduce the distance the disk head must travel and the total number of disk seeks 161 162 Lecture 16 November 29 1631 FCFS This is similar to the other rst come rst serve algorithms we have used so far It accesses blocks on the disk in the order that they are requested As with other FCFS algorithms it is not very ef cient 1632 Shortest Seek Time First Shortest Seek Time First is a greedy algorithm similar to Shortest Job First It always moves the disk head to the next closest disk block Unlike SJF7 SSTF is not an optimal scheduling algorithm 1633 Scan Scan is a real disk scheduling algorithm It uses an an algorithm simililar to that of an elevator The disk head starts in the middle and moves up or down picking up all the requests in that direction as it goes Scan travels all the way from the center to either edge of the disk on every sweep This can make the algorithm very slow7 however7 the slowness tends to not be signi cant on a disk with uniformly distributed random data 1634 Scan Look Scan Look is just like scan but it only goes to the highest or lowest requested block This cuts down the overhead of using scan 1635 Circular Scan Circular Scan sweeps the head across disk from one edge to the other and then resets the head back to the beginning each time CScan can be expensive but usually it is more fair than Scan or Scan Look 1636 C Scan Look CScan Look works the same as CScan but the disk heads only go between the highest and lowest requested blocks 164 Read Ahead Read ahead is a great way to increase throughput However7 it is not needed in an software because it is done in hardware by the disk controller 165 RAID RAID Redundant Array of Inexpensive Drives Lecture 16 November 29 163 Raid can increase throughput and reliability by spreading data across multiple disks and adding error check ing or by duplicating data 1651 RAID 0 Is not real RAID it just makes multiple disks look like one big disk 1652 RAID 1 Each disk is mirrored on another disk This has a very large overhead but is very reliable Write times donlt change but read times can be cut in half 1653 RAID 2 RAID 2 adds error checking Every 10 disks requires 4 parity disks The access times are the same as RAID 1 1654 RAID 4 RAID 4 uses data striping Data is spread across multiple disks This Improves read performance but because the parity disks are not striped it hurts read performance 1655 RAID 5 RAID 5 striped data and parity information across multiple disks This removes the bottleneck created by RAID 4
Are you sure you want to buy this material for
You're already Subscribed!
Looks like you've already subscribed to StudySoup, you won't need to purchase another subscription to get this material. To access this material simply click 'View Full Document'