Proposed ADSM/TSM configurations
Hello All, Please provide me your suggestions and comments regarding my proposed configurations. Please help me. I am planning to upgrade Adsm server both hardware and software due to tremedous increase of a heterogeneous clients and the to upgrade ADSM software into a latest version. The estimated clients are about 90 nodes for Windows platforms (NT/2000) with estimated capacity 1350GB and 30 nodes for Unix platforms with estimated disk capacity 2500 GB. We facing a lot of problems from the existing environment, so, we decided to upgrade the system. Please help me if my proposed configurations is fit enough to meet the objectives. But here first the details of a current configurations: Hardware: * Compact Proliant 5500 (2 CPU, 256 RAM, 80 HD, 4 SCSI controllers) * Tape library ibm 3494/L10 with 4 drives (3590/B1A) and 2 drives (3490/C2A) Note: The 4 drives (3590) were currently being used by the system and the other 2 drives (3490) is setting idle in the ATLS. The ATLS with capacity of 2000 tapes Software: * OS - NT4, SP5 * Adsm server (3.1.0.2.40) Adsm server configurations: * Db size = 15GB * Log size = 1GB * Backup disk storage = 20GB * Archive disk storage= 20GB * Two primary tape storage pools for backup and archive with 500 tapes each. * The tapes size in ATLS is 25GB Note: We performed archive for selected server only to avoid running outspace of the database. The offsite copy (copy storage pool) and HSM are not implemented. = My proposed configurations is to provide two servers, one for Windows and one for UNIX platforms. The ideas are to improve the performance and productivity, to separate the adsm processes (windows and unix), to implement the offsite backup, HSM for unix platforms, etc. Here are the details of proposed configurations: For Windows Platforms (with 90 clients with estimated disk space 1350GB): Hardware: * Compact Proliant 5500 (2 CPU, 512 RAM, 70HD with Raid Support, 2 SCSI controllers, 30 size DLT tape) * Utilized the two drives (3490/C2A) that are currently idle in the ATLS Software: * OS = Windows NT/2000 * *SM (latest version compliant to Windows 2000 platform) Adsm/TSM Server Configurations: * Db size = 20GB * Log size = 2GB * Backup disk storage pool = 20GB * Archive disk storage pool = 20GB * Two primary storage pools for backup and archive with 200 tapes each (25GB size each tape) * Two copy storage pools for offsite backup * Backup version (7 versions and 1 version for the deleted files) and we keep 60 days. * Archive (weekly (1 month retention), monthly (1 year retention) and yearly with 60% of estimated disk space) For UNIX Platforms (with 30 clients with estimated disk space 2500GB): Hardware: * Compact Proliant 5500 (2 CPU, 512 RAM, 70HD with Raid Support, 4 SCSI controllers, 30 size DLT tape) * Utilized the four drives (3590/B1A) in ATLS 3494 that are currently used by ADSM server. Software: * OS = Windows NT/2000 * *SM (latest version) *SM Server Configurations: * Db size = 30GB * Log size = 4GB * Backup disk storage pool = 20GB * Archive disk storage pool = 20GB * HSM (space migration) disk storage pool = 30GB * Three primary tape storage pools for backup, archive, and space migration with 200 tapes each (25GB size each tape) * Three copy storage pools for offsite backup * Backup version (7 versions and 1 version for the deleted files) and we keep 60 days. * Archive (weekly (1 month retention), monthly (1 year retention) and yearly with 60% of estimated disk space) Thanks a lot in advance. Zosi
Re: Admin schedule to Backup DB does not start
You can simulate having an EVENT driven scheduler by changing the database backup event from a straight command to a script. See, for example, the Tivoli-supplied script BKUP_STG_DB that is shipped with the product. In the supplied sample, they show doing a storage pool backup, followed by a database backup, with logic to defer the whole event if there is a node in session with the server. A few deft modifications should suit your purposes. For example, their script starts out with select * from sessions where - upper(session_type)='NODE' if (rc_ok) goto reschedule The reschedule label then deletes a "temporary" schedule (if one exists) and creates a onetime schedule that runs the script again, specifying a start time of NOW+0:20. If you were to use this for "inspiration", you could schedule the storage pool backup commands to run in the background, schedule the database backup command / script to run at a reasonable time after they normally complete and start the script out with something along the lines of the following select * from processes where - upper(process)='BACKUP STGPOOL' if (rc_ok) goto reschedule Then, at the reschedule label, play the same game with rescheduling the database backup in a temporary schedule at "some point" in the future. (Reader beware! The syntax may not be 100% accurate. I am assuming that the process name that is returned will indeed match the command name executed to back up the storage pool.) I believe this should work, but, as always, I reserve the right to be wrong. Jack Coyle > -- > From: Talafous, John G.[SMTP:[EMAIL PROTECTED]] > Reply To: ADSM: Dist Stor Manager > Sent: Saturday, March 24, 2001 4:39 PM > To: [EMAIL PROTECTED] > Subject: Re: Admin schedule to Backup DB does not start > > Yes, the compelling reason is to have the backups of storage pools > complete, > hopefully, by 9:00AM. Then the scheduled launch of a database backup at > 10:00AM would coincide nicely with DRM processing and the resultant > ejection > of tapes from an ATL. > > But, if clients send more data than anticipated. Oh my! I run into > the > situation that the earlier ADMIN schedule does not complete soon enough. > (I > have changed the BACKUP_DB schedule to have an eight (8) hour startup > window). This should help. > > Another reason is an attempt to keep the number of tape drives utilized by > administrative schedules and migration tasks at an > INSTALLATION_MAX_DRIVES-1 > to permit random client restores to have at least one drive available. > (Note: INSTALLATION_MAX_DRIVES is a variable applicable to any situation.) > > Is the problem not having an EVENT driven scheduler available to TSM? (ie: > When storage pool X drains to zero, why can't I launch a backup of storage > pool x-tape to x-copy?) Or, can I? > > Thanks in advance, > John G. Talafous IS Technical Principal > The Timken CompanyGlobal Software Support > P.O. Box 6927 Data Management > 1835 Dueber Ave. S.W. Phone: (330)-471-3390 > Canton, Ohio USA 44706-0927 Fax : (330)-471-4034 > [EMAIL PROTECTED] http://www.timken.com > > > > > > > > -Original Message- > From: Coyle, Jack [mailto:[EMAIL PROTECTED]] > Sent: Saturday, March 24, 2001 12:37 PM > To: [EMAIL PROTECTED] > Subject: Re: Admin schedule to Backup DB does not start > > > The issue here would seem to be the "wait=yes" option you have specified > for > the backup command. Doing so causes the server to process the command in > the > foreground. This, in turn, requires that you must wait for the command to > complete before performing any other tasks. Indeed, in such an > environment, > you can only run one scheduled command at a time. Is there a compelling > reason why "wait=yes" is specified (other than the obvious one of not > wanting the database backup to occur before the backups of the storage > pools > complete)? > > Jack Coyle > > > -- > > From: Talafous, John G.[SMTP:[EMAIL PROTECTED]] > > Reply To: ADSM: Dist Stor Manager > > Sent: Saturday, March 24, 2001 12:07 PM > > To: [EMAIL PROTECTED] > > Subject: Re: Admin schedule to Backup DB does not start > > > > The ADMIN schedule that processing storage pools contained a "backup stg > > blahpool blahpool wait=yes". When the wait completed, another backup > stg > > command was issued and the ADMIN schedule ended. Then, the ADMIN > schedule > > to > > do the database backup kicked off. > > > > Does this mean only one ADMIN schedule at a time? > > jt > > > > -Original Message- > > From: Cook, Dwight E [mailto:[EMAIL PROTECTED]] > > Sent: Saturday, March 24, 2001 11:33 AM > > To: [EMAIL PROTECTED] > > Subject: Re: Admin schedule to Backup DB does not start > > > > > > did they have the same start time ? ? ? > > That might cause that... but the actual scheduled event should initiate > > and > > be gone in second
Re: Admin schedule to Backup DB does not start
Yes, the compelling reason is to have the backups of storage pools complete, hopefully, by 9:00AM. Then the scheduled launch of a database backup at 10:00AM would coincide nicely with DRM processing and the resultant ejection of tapes from an ATL. But, if clients send more data than anticipated. Oh my! I run into the situation that the earlier ADMIN schedule does not complete soon enough. (I have changed the BACKUP_DB schedule to have an eight (8) hour startup window). This should help. Another reason is an attempt to keep the number of tape drives utilized by administrative schedules and migration tasks at an INSTALLATION_MAX_DRIVES-1 to permit random client restores to have at least one drive available. (Note: INSTALLATION_MAX_DRIVES is a variable applicable to any situation.) Is the problem not having an EVENT driven scheduler available to TSM? (ie: When storage pool X drains to zero, why can't I launch a backup of storage pool x-tape to x-copy?) Or, can I? Thanks in advance, John G. Talafous IS Technical Principal The Timken CompanyGlobal Software Support P.O. Box 6927 Data Management 1835 Dueber Ave. S.W. Phone: (330)-471-3390 Canton, Ohio USA 44706-0927 Fax : (330)-471-4034 [EMAIL PROTECTED] http://www.timken.com -Original Message- From: Coyle, Jack [mailto:[EMAIL PROTECTED]] Sent: Saturday, March 24, 2001 12:37 PM To: [EMAIL PROTECTED] Subject: Re: Admin schedule to Backup DB does not start The issue here would seem to be the "wait=yes" option you have specified for the backup command. Doing so causes the server to process the command in the foreground. This, in turn, requires that you must wait for the command to complete before performing any other tasks. Indeed, in such an environment, you can only run one scheduled command at a time. Is there a compelling reason why "wait=yes" is specified (other than the obvious one of not wanting the database backup to occur before the backups of the storage pools complete)? Jack Coyle > -- > From: Talafous, John G.[SMTP:[EMAIL PROTECTED]] > Reply To: ADSM: Dist Stor Manager > Sent: Saturday, March 24, 2001 12:07 PM > To: [EMAIL PROTECTED] > Subject: Re: Admin schedule to Backup DB does not start > > The ADMIN schedule that processing storage pools contained a "backup stg > blahpool blahpool wait=yes". When the wait completed, another backup stg > command was issued and the ADMIN schedule ended. Then, the ADMIN schedule > to > do the database backup kicked off. > > Does this mean only one ADMIN schedule at a time? > jt > > -Original Message- > From: Cook, Dwight E [mailto:[EMAIL PROTECTED]] > Sent: Saturday, March 24, 2001 11:33 AM > To: [EMAIL PROTECTED] > Subject: Re: Admin schedule to Backup DB does not start > > > did they have the same start time ? ? ? > That might cause that... but the actual scheduled event should initiate > and > be gone in seconds... > For all admin activity that I can think of, tsm simply spawns off a > process > and the scheduled event is complete, unless it is something really short > like an "update stg diskpool high=1 low=0" type thing... > What was your other schedule that actually tied things up for a while... > Like with the dbbackup schedule, it will start a process and then the > schedule/event is complete... > > Dwight > > -Original Message- > From: Talafous, John G. [mailto:[EMAIL PROTECTED]] > Sent: Saturday, March 24, 2001 9:38 AM > To: [EMAIL PROTECTED] > Subject: Re: Admin schedule to Backup DB does not start > > > Well what do you know?!? Another ADMIN schedule to manage storage pools > was > running. When it finished, the BACKUP_DB admin schedule started. But, does > this mean that only one (1) admin schedule can be active at any given > time? > > TIA, > John G. Talafous IS Technical Principal > The Timken CompanyGlobal Software Support > P.O. Box 6927 Data Management > 1835 Dueber Ave. S.W. Phone: (330)-471-3390 > Canton, Ohio USA 44706-0927 Fax : (330)-471-4034 > [EMAIL PROTECTED] http://www.timken.com >
Re: Admin schedule to Backup DB does not start
The issue here would seem to be the "wait=yes" option you have specified for the backup command. Doing so causes the server to process the command in the foreground. This, in turn, requires that you must wait for the command to complete before performing any other tasks. Indeed, in such an environment, you can only run one scheduled command at a time. Is there a compelling reason why "wait=yes" is specified (other than the obvious one of not wanting the database backup to occur before the backups of the storage pools complete)? Jack Coyle > -- > From: Talafous, John G.[SMTP:[EMAIL PROTECTED]] > Reply To: ADSM: Dist Stor Manager > Sent: Saturday, March 24, 2001 12:07 PM > To: [EMAIL PROTECTED] > Subject: Re: Admin schedule to Backup DB does not start > > The ADMIN schedule that processing storage pools contained a "backup stg > blahpool blahpool wait=yes". When the wait completed, another backup stg > command was issued and the ADMIN schedule ended. Then, the ADMIN schedule > to > do the database backup kicked off. > > Does this mean only one ADMIN schedule at a time? > jt > > -Original Message- > From: Cook, Dwight E [mailto:[EMAIL PROTECTED]] > Sent: Saturday, March 24, 2001 11:33 AM > To: [EMAIL PROTECTED] > Subject: Re: Admin schedule to Backup DB does not start > > > did they have the same start time ? ? ? > That might cause that... but the actual scheduled event should initiate > and > be gone in seconds... > For all admin activity that I can think of, tsm simply spawns off a > process > and the scheduled event is complete, unless it is something really short > like an "update stg diskpool high=1 low=0" type thing... > What was your other schedule that actually tied things up for a while... > Like with the dbbackup schedule, it will start a process and then the > schedule/event is complete... > > Dwight > > -Original Message- > From: Talafous, John G. [mailto:[EMAIL PROTECTED]] > Sent: Saturday, March 24, 2001 9:38 AM > To: [EMAIL PROTECTED] > Subject: Re: Admin schedule to Backup DB does not start > > > Well what do you know?!? Another ADMIN schedule to manage storage pools > was > running. When it finished, the BACKUP_DB admin schedule started. But, does > this mean that only one (1) admin schedule can be active at any given > time? > > TIA, > John G. Talafous IS Technical Principal > The Timken CompanyGlobal Software Support > P.O. Box 6927 Data Management > 1835 Dueber Ave. S.W. Phone: (330)-471-3390 > Canton, Ohio USA 44706-0927 Fax : (330)-471-4034 > [EMAIL PROTECTED] http://www.timken.com >
Re: Admin schedule to Backup DB does not start
The ADMIN schedule that processing storage pools contained a "backup stg blahpool blahpool wait=yes". When the wait completed, another backup stg command was issued and the ADMIN schedule ended. Then, the ADMIN schedule to do the database backup kicked off. Does this mean only one ADMIN schedule at a time? jt -Original Message- From: Cook, Dwight E [mailto:[EMAIL PROTECTED]] Sent: Saturday, March 24, 2001 11:33 AM To: [EMAIL PROTECTED] Subject: Re: Admin schedule to Backup DB does not start did they have the same start time ? ? ? That might cause that... but the actual scheduled event should initiate and be gone in seconds... For all admin activity that I can think of, tsm simply spawns off a process and the scheduled event is complete, unless it is something really short like an "update stg diskpool high=1 low=0" type thing... What was your other schedule that actually tied things up for a while... Like with the dbbackup schedule, it will start a process and then the schedule/event is complete... Dwight -Original Message- From: Talafous, John G. [mailto:[EMAIL PROTECTED]] Sent: Saturday, March 24, 2001 9:38 AM To: [EMAIL PROTECTED] Subject: Re: Admin schedule to Backup DB does not start Well what do you know?!? Another ADMIN schedule to manage storage pools was running. When it finished, the BACKUP_DB admin schedule started. But, does this mean that only one (1) admin schedule can be active at any given time? TIA, John G. Talafous IS Technical Principal The Timken CompanyGlobal Software Support P.O. Box 6927 Data Management 1835 Dueber Ave. S.W. Phone: (330)-471-3390 Canton, Ohio USA 44706-0927 Fax : (330)-471-4034 [EMAIL PROTECTED] http://www.timken.com
Re: Admin schedule to Backup DB does not start
did they have the same start time ? ? ? That might cause that... but the actual scheduled event should initiate and be gone in seconds... For all admin activity that I can think of, tsm simply spawns off a process and the scheduled event is complete, unless it is something really short like an "update stg diskpool high=1 low=0" type thing... What was your other schedule that actually tied things up for a while... Like with the dbbackup schedule, it will start a process and then the schedule/event is complete... Dwight -Original Message- From: Talafous, John G. [mailto:[EMAIL PROTECTED]] Sent: Saturday, March 24, 2001 9:38 AM To: [EMAIL PROTECTED] Subject: Re: Admin schedule to Backup DB does not start Well what do you know?!? Another ADMIN schedule to manage storage pools was running. When it finished, the BACKUP_DB admin schedule started. But, does this mean that only one (1) admin schedule can be active at any given time? TIA, John G. Talafous IS Technical Principal The Timken CompanyGlobal Software Support P.O. Box 6927 Data Management 1835 Dueber Ave. S.W. Phone: (330)-471-3390 Canton, Ohio USA 44706-0927 Fax : (330)-471-4034 [EMAIL PROTECTED] http://www.timken.com
Re: Admin schedule to Backup DB does not start
Well what do you know?!? Another ADMIN schedule to manage storage pools was running. When it finished, the BACKUP_DB admin schedule started. But, does this mean that only one (1) admin schedule can be active at any given time? TIA, John G. Talafous IS Technical Principal The Timken CompanyGlobal Software Support P.O. Box 6927 Data Management 1835 Dueber Ave. S.W. Phone: (330)-471-3390 Canton, Ohio USA 44706-0927 Fax : (330)-471-4034 [EMAIL PROTECTED] http://www.timken.com
Admin schedule to Backup DB does not start
I have an Admin schedule to backup the TSM database that has missed it's schedule start window twice this week. Environment is 3466-C00, AIX 4.3.2, TSM 3.7.2 and a q schedule shows this: tsm: FSPHNSM1>q sched backup_db t=a f=d Schedule Name: BACKUP_DB Description: Backup DataBase Command: backup db type=full devclass=3590-E1A Priority: 1 Start Date/Time: 02/27/2001 10:00:00 Duration: 1 Hour(s) Period: 1 Day(s) Day of Week: Any Expiration: Active?: Yes Last Update by (administrator): TALAFOUS Last Update Date/Time: 03/19/2001 13:08:01 Managing profile: This schedule missed on Wednesday and I did a manual backup at about 5:00PM. It looks like it's going to miss again today (10:26AM and not started yet!). Does anyone have any ideas? TIA John G. Talafous IS Technical Principal The Timken CompanyGlobal Software Support P.O. Box 6927 Data Management 1835 Dueber Ave. S.W. Phone: (330)-471-3390 Canton, Ohio USA 44706-0927 Fax : (330)-471-4034 [EMAIL PROTECTED] http://www.timken.com