Coursehero >> New Jersey >> Rutgers >> 206 352

Course Hero has millions of student submitted documents similar to the one below including study guides, homework solutions, papers, exam answer keys and textbook solutions.

File References system a survey of backup techniques, Ann Chervenak, et.al. Physical vs logical backup, Hutchinson, OSDI 99 File System Design for an NFS File Server Appliance TR3002 by Dave Hitz, James Lau, & Michael Malcolm, Network Appliance, Inc. backup services Badri Nath Rutgers University badri@cs.rutgers.edu Protecting file syste ms: a survey of backup techniques. WAFL Physical vs logical backup Data backup Size of files increasing Devices on which files are stored becoming heterogeneous Contact data, images audio Data backup is a huge problem Legal and regulations force data to be stored, multiple versions Backup device is tape or disk Disk as backup is device is gaining ground Backup techniques Full vs incremental backup Full backup Copy entire FS to a backup device Easy restore Costly in terms of space and time Incremental backup Copy only changes since last backup Perform occasional full backups Reduce size of backups Restore may have to follow a chain on incremental backups 1 Backup techniques File-based vs device-based File-based Understand file-structure Read entire files and write to backup device (contiguously) Backup operations use random seeks Restore is easy Any changes to file require entire file to be backed up Unix tar program Unix backup Utilities Tar, cpio, dump and restore Manual, scripts Files, types of files Failures during backup Consistency Device-based Backup blocks, faster writes Need file to block mapping structure Not easily portable Restore will need random seeks Unix dump program Backup techniques Off-line vs on-line backup Off-line backup; system in quiescent state No consistency problems System unavailable during backup snapshot Snapshot : create a atomic read-only copy Any modified blocks accessed via new version Use copy-on-write technique Contents of the snapshot can be copied to a back up device An incremental or full backups of the snapshots can be implemented Popularized by netapp filers On-line backup Backup during system operation Consistency problems Movement of directories may not back up files Need to scan the directory structure (atomic operation) and then backup Snapshot techniques 2 Snapshots Make a copy of root-inode Current FS accessible via new rooti-node Snapshot pointed to by old root i-node Use copy-on-write technique Modified blocks accessible only through the new i-node WAFL WAFL (write-anywhere file system) Has snapshot facility Similar to unix file system File metadata in i-nodes i-nodes point to blocks that are files or directories Inodes also considered as files Can write anywhere WAFL Snapshots WAFL creates and deletes snapshots automatically at preset times Up to 255 snapshots stored at once Uses Copy-on-write to avoid duplicating blocks Snapshot allows Users to recover accidentally deleted files admins Sys to create backups from running system System can restart quickly after unclean shutdown Roll back to previous snapshot WAFL: Metadata as files Fixed location Everything else: write anywhere. i-nodes of the file system User data Bit maps 3 WAFL: detailed view When disk block modified, must modify indirect pointers as well Snapshot details Batch, to improve I/O performance Batched writes Every modified block (and its parents) cannot be written immediately Batched write operations What happens during a crash in between snapshots? Define consistency points and use NVRAM WAFL Consistency Points WAFL uses Snapshots internally so that it can restart quickly even after an unclean system shutdown. A consistency point is a completely self-consistent image of the entire file system. When WAFL restarts, it simply reverts to the most recent consistency point. Buffer dirty data in memory (delayed writes) and write new consistency points as an atomic batch (force). A consistency point transitions the FS from one selfconsistent state to another. Combine with FS operation log in NVRAM 4 What Has Changed? Given a snapshot X, WAFL can ask: is block B allocated in snapshot X? Given a snapshot X and a later snapshot Y, WAFL can ask: what blocks of Y should be sent to the mirror? X 0 0 1 1 Y 0 1 0 Free/unused New block send to mirror Deleted block Un modified block The Block-Map File Typical FS uses bit for each free block, 1 is allocated and 0 is free Ineffective for WAFL since may be other snapshots that point to block WAFL uses 32 bits for each block 1 User Access to Snapshots Example, suppose accidentally removed file named 553.hw1 : Snapshot directories lists the snapshot taken Use that snapshot to restore the lost file No more my disk ate my homework excuse Ls .snapshot/553.hw1 .snapshot/today/.. .snapshot/yesterday/553.hw1 .snpashot/feb26/553.hw1 Performance recover most recent version: cp .snapshot/yesterday/553.hw1 553.hw1 Note, snapshot directories (.snapshot) are hidden in that they don t show up with ls 5 SnapMirror Can use this mechanism to mirror data across WAN Can reduce data storage requirements by not backing up deleted/updated data Identifying dirty blocks are easier than logical, file system aware mechanisms WAFS WAFL Conclusions WAFL WAFL s consistency points allow it to buffer writes and push them out in a batch Copy-on-write technique allows on-line backup Write performance optimized Deferred, clustered allocation Batch writes Localize writes Challenges Snap mirror over WAN Expensive (network bandwidth requirements) Performance slow because transaction cannot complete till WAN update finishes Multiple device backups 2 B cell phones Each has 1 M of data (pictures + contact data) How to design backup service? 6

Find millions of documents here - Study Guides, Homework Solutions, Papers, Exam Answer Keys and more. Course Hero has millions of course related materials that will enable you to learn better, faster and get an A in all your courses.
Below is a small sample set of documents: