- Community Home
- >
- Servers and Operating Systems
- >
- Operating Systems
- >
- Operating System - HP-UX
- >
- write back caching and data loss
Categories
Company
Local Language
Forums
Discussions
Forums
- Data Protection and Retention
- Entry Storage Systems
- Legacy
- Midrange and Enterprise Storage
- Storage Networking
- HPE Nimble Storage
Discussions
Discussions
Discussions
Forums
Forums
Discussions
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
- BladeSystem Infrastructure and Application Solutions
- Appliance Servers
- Alpha Servers
- BackOffice Products
- Internet Products
- HPE 9000 and HPE e3000 Servers
- Networking
- Netservers
- Secure OS Software for Linux
- Server Management (Insight Manager 7)
- Windows Server 2003
- Operating System - Tru64 Unix
- ProLiant Deployment and Provisioning
- Linux-Based Community / Regional
- Microsoft System Center Integration
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Community
Resources
Forums
Blogs
- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
тАО06-12-2010 08:46 AM
тАО06-12-2010 08:46 AM
is HP-UX write back data caching ?
when data in cache dirty, if system goes down. is it data loss?
is there some way to get that data back. like some utility that
read the system core dump and rebuild the cache or something like that.
Scotty
Solved! Go to Solution.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
тАО06-12-2010 09:54 AM
тАО06-12-2010 09:54 AM
SolutionSo if the system goes down, memory is never saved unless the system performs a TOC or has a system crash. Then a memory dump is taken with the contents sent to the dump area. However, today's RAM can be enormously large (32 GB to 256 GB) so not every part of RAM is saved. With crashconf, the selected memory sections can be included or excluded.
Although there is no utility to "rebuild the cache" since there would be no way to start all the programs in a way that they could synchronize and use the data.
The first step towards reliability is power and air conditioning. A simple UPS for a server is not enough -- the data center needs a big UPS with a backup generator. Or the big UPS comes with software to force an orderly shutdown of every system *and* disk arrays. Similarly, the air conditioning must also provide an orderly of all servers when the temperature exceed 95F or higher.
Programs and databases must be designed for crash recovery -- it will happen. Crash prevention includes patching the OS at least twice a year and the applications and database engine as required.
Bill Hassell, sysadmin
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
тАО06-12-2010 02:48 PM
тАО06-12-2010 02:48 PM
Re: write back caching and data loss
Are you talking about the cache in disk controllers or what?
There are CPU caches. The file cache in RAM and various caches in array controllers.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
тАО06-12-2010 03:07 PM
тАО06-12-2010 03:07 PM
Re: write back caching and data loss
When using the JFS (VxFS) filesystem, various mount options control data integrity across a crash. You can ramp up performance at the expense of some data loss (e.g. mounting with 'tmplog') or chose to record data and metadata changes before returning to the calling program with the 'log' mount option.
See the manpages for 'mount_vxfs' for more information.
Regards!
...JRF...
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
тАО06-12-2010 10:09 PM
тАО06-12-2010 10:09 PM
Re: write back caching and data loss
i was referring to data cache which termed as buffer cache in hp-ux.
Bill Hassell>>
did you mean,
if system crashed due to some bug in some component and lets say there is
dirty data in cache. then that data will be lost.
the program/application should be written such that it would deal with this
scenario ?
Scotty
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
тАО06-12-2010 11:56 PM
тАО06-12-2010 11:56 PM
Re: write back caching and data loss
On 11.31, it is now called the Unified File Cache.
>the program/application should be written such that it would deal with this scenario?
Are you using a database? They may have their own transaction managers to handle this.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
тАО06-13-2010 03:41 AM
тАО06-13-2010 03:41 AM
Re: write back caching and data loss
dirty data in cache. then that data will be lost.
Correct.
> the program/application should be written such that it would deal with this
scenario ?
Yes.
When any computer crashes, everything you know about the OS is gone. In HP-UX, if there is an OS crash, selected portions of memory are written to the crash dump area on disk (assuming that the disk is working). Then the system boots up and you start your application again. Assuming that you knew enough about the HP-UX kernel and memory mapping (and adb and other tools), if you had a copy of the data, what would you do with it? Your program has no way to know where it was when the crash occurred.
Your programs should be written to recover from a crash, that is, keep transaction and rollback logs. Given that a well-patched HP-UX system very seldom crashes, I would be more concerned about mirroring your vg00 disks, taking regular Ignite backups and rehearsing a hard disk recovery.
Bill Hassell, sysadmin
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
тАО06-14-2010 05:07 AM
тАО06-14-2010 05:07 AM
Re: write back caching and data loss
well written reply. easy to understand.
if talk about writeback advantage.
does anybody have statistic as to how much percent faster is writeback
compared to say writethrough data caching.
Scotty
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
тАО06-14-2010 06:56 AM
тАО06-14-2010 06:56 AM
Re: write back caching and data loss
I think you'll have to explain your terminology. These terms are normally applied to the electrical design of a CPU chip or to the design of a data cache such as the HP-UX buffer cache (Unified File Cache for 11.31). You can learn all about it here:
http://h20338.www2.hp.com/enterprise/w1/en/os/hpux11i-kod-unified-file-cache.html
Bill Hassell, sysadmin
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
тАО06-17-2010 06:12 AM
тАО06-17-2010 06:12 AM
Re: write back caching and data loss
by cache i refer to buffer cache (data cache)
write through
Application does write IO
- data written to cache
- data written to disk
- IO completes
Write back
Application does write IO
- data written to cache
- IO completes
- data written to disk based on write back policy
write back give better performance compared to write through.
i wanted to know if anybody has some percentage numbers.
Scotty