June 9, 2011 at 10:57 am
So I have been researching for a couple weeks, read through this forum (most informative place minus a couple white papers), and am looking for final input while starting forward.
I just got into a server system with roughly 1.5 TB data, split between 1 Raid 10, 2 Raid 5s and a Raid 1 OS and Raid 1 other. The file autogrowth on many databases is set to very small numbers and needs to be changed. I have seen I/O bottlecks almost right away and CPU bottlenecks that could be stemming from poor fragmentation. I cannot even run a check on internal fragmentation (index/table/other) because the I/O it takes to do a full index report will take up too many resources. SAS Raid systems.
I have been talking to the Network staff and the drives are most likely very fragmented. Can't even risk the frag analysis at normal operation times. There had been log files that had auto growth and not backed up or maintained and took up disk.... I'll sum it up. Drives fragmented, looking at fix. Going to push forward on testing with PerfectDisk, looking to see if anyone would have any better info.
I have seen a lot of posts with DisKeeper (55) only only few with PerfectDisk on this forum. I have looked over free utilities as well but these two always hit on topic for server level, SQL Server cannot be taken down, manage everything well... etc.
Anyone ever do a direct comparison or have good stats on either individually? DisKeeper better? PerfectDisk better? Advantages from one over the other based on server specs?
Raxco got my attention with this (http://www.raxco.com/user_data/white_papers/unique_differences_pd11_dk2010.pdf) paper that pushed me to discuss with Net and agreed to put the trial version on our test server. Also, the S.M.A.R.T. capabilities are a plus for Net. DisKeeper has more input from actual users saying success and good throttling control while SQL server runs continuosly.
If you put all that together, thanks, and for any feedback.
-Me
June 9, 2011 at 11:12 am
I have used both of them and there is no real comparison. PerfectDisk by far outperforms DisKeeper.
Unfortunately, if you have heavy fragmentation at the OS level - you would need to stop SQL Server for proper disk defrag using any util.
You may want to look into getting approved outages to defrag the drives.
Jason...AKA CirqueDeSQLeil
_______________________________________________
I have given a name to my pain...MCM SQL Server, MVP
SQL RNNR
Posting Performance Based Questions - Gail Shaw[/url]
Learn Extended Events
June 9, 2011 at 11:19 am
If you get Diskeeper, patch it to the latest version immediately. Some of the older versions have nasty corruption- or error-causing bugs
Gail Shaw
Microsoft Certified Master: SQL Server, MVP, M.Sc (Comp Sci)
SQL In The Wild: Discussions on DB performance with occasional diversions into recoverability
June 9, 2011 at 12:26 pm
Haven't tested them together, but both have worked for me. I have been hesitant to use them without stopping SQL Server. Even then, make sure I have 2 backups, just in case.
I'd schedule an outage, like Sat night/sun am, try to defragment as much as you can. Or at least analyze and see if you have lots of log/data file growths. Maybe think about a shrink of the log and then a controlled growth of files after defragmentation.
June 10, 2011 at 3:21 am
Don't forget to "fix" SQL Server after your OS defrag. Check your indexes to see if they need to be defragged (because a hard drive defrag will move your SQL files around once you've stopped the services) and update your statistics after fixing any index problems.
I'm probably just being anal, but IMHO, it's never a good idea to defrag hard drives without doing SQL Server housekeeping afterward.
June 10, 2011 at 8:12 am
Thanks all. 1 PerfectDisk, 1 not DisKeeper and 1 Neutral leans me in that I was going the correct path for the current needs.
I have been trying to avoid the maintenance window, but due to the drives being online for about 3 years with no real fragmentation statistics I will be looking to go for the maintenance window over slow time. The biggest problem I had is that there are no analytical tools going for WHAT is the peak usage time on that server. The Net guys are working to get Ignite going so I can get that data. It was preferred that I get use that software as opposed to perfmon and my own SQL server level monitoring tools I developed as it will give the usage store in an independant server. Thanks for the posts on that, it gives good reassurance when I take the window to Net and Dir of IT.
We have the perfect opportunity here to test out the S.M.A.R.T. capabilities on the new test server for drive monitoring. I have not seen anything like this around the DisKeeper application.
Just a few other Qs in case the answers are known:
Does DisKeeper have any analysis tools as part of it other than fragmentation that would otherwise not be listed on their site (drive health monitoring, disk usage reporting, other reporting)?
Does PerfectDisk have the same throttle control capability (and as good of one) that I have heard DisKeeper is good at, as to not pool all resources or allow other processes resources first?
Does anyone leave either application running for preventative fragmentation ability they both have ON, 24/7 and if so are there any statistic that would lead to believe that it would put extra wear on spidles and heads?
Side note is if anyone knows of any defrag utility for server environment supporting Raid drives on large (over 1 TB) systems that also has S.M.A.R.T. capabilities, I would love to hear the name and do some research into. I am concerned that the head movement over the years could weaken the drive stability so the analytics on that and any advanced notice I could get would be of help.
Thanks again!
-Matt
June 14, 2011 at 6:14 am
Regarding your PerfectDisk throttling question, yes, PerfectDisk has throttle control for their guests and hosts – they call it Virtual Awareness. And they also have CPU throttling for non-virtual as well.
June 14, 2011 at 6:51 am
GilaMonster (6/9/2011)
If you get Diskeeper, patch it to the latest version immediately. Some of the older versions have nasty corruption- or error-causing bugs
+1
---------------------------------------------------------------------
June 14, 2011 at 8:03 am
Thanks George and Dragonette.
Dragonette, I am curious if the throttling control is as good as the DisKeeper mechanism. I have heard very positive remarks to DisKeeper's ability but have not heard anything about PerfectDisk's throttle capabilities. Do you have any comparison notes between the two?
Thanks!
-Matt
June 14, 2011 at 8:17 am
Unfortunately I do not have any comparison notes on throttling. I can tell you PerfectDisk’s works from my experience but I have not used Diskeeper’s.
June 14, 2011 at 8:20 am
Good to know! Thanks much!
June 14, 2011 at 11:55 am
I have not used Perfect Disk, but I know that Diskeeper can do what you need.
It does provide throttling control (called I/O Smart, I think) but with the latest edition using InvisiTasking, it is not really needed.
Here is some information on InvisiTasking:
http://files.diskeeper.com/pdf/InvisiTasking_Sheet.pdf
http://www.diskeeper.com/defrag/invisitasking/[/url]
Talking with over 100 IT Managers and DB Admins, the vast majority of them prefer Diskeeper over any other solution as they point out that InvisiTasking made it possible for Diskeeper 2011 to defrag their servers without having to take them offline and they saw no resource hit while it was running with the system's performance continually improving as the disk fragmentation was continually resolving.
Once the defrag was complete, many pointed out that Diskeeper actually prevented most of the the fragmentation (over 90% of it according to some), enabling their system to maintain high performance levels with the residual fragmentation handled pretty much instantly and again no resource drain.
I know that it does have a lot of reports (some home users say too much info, but Sys Admins and IT Managers love that) and a wide range of control as well.
I suggest getting a trial at www.diskeeper.com (I am told you can request 45 days or more). Best for you to evaluate it and see what it does for your site.
Based on my experience and that of many Sys Admins and IT Managers I have spoken to, I think you will be very pleased with the results.
..
June 14, 2011 at 12:15 pm
Wow. You should be a sales person for Diskeeper.
June 17, 2011 at 9:42 am
I'm the Product Manager for Diskeeper Corporation, & want to speak to some of the questions posed directly.
Users SQLRNNR & Steve Jones point out a best practice that has been historically necessary: in the event that you've struck a point where your server environment is at a triage level of both fragmentation & free space, taking the server offline during a dedicated scheduled defrag job that isn't resource throttled. We have arrived in an era where disk optimization needs to be tailored to constant production demands and definitively preventative, and there was a time when automatic defragmentation scheduled around peak usage was the ideal approach. However, Diskeeper’s patented InvisiTasking technology obviates the need to utilize a maintenance window for defragmentation. This white paper including testing by Percept Technologies speaks to the efficiency of InvisiTasking in ensuring zero resource overhead while Diskeeper optimizes disks: http://downloads.diskeeper.com/pdf/invisitasking-exchange-sql-server.pdf
Steve Jones also points out, "Maybe think about a shrink of the log and then a controlled growth of files after defragmentation." This can be an excellent strategy so long as a solution is in place to prevent the same outcome a second time. Consolidation of free space coupled with ensuring contiguous file placement at write accomplishes the desired effect (zero performance impact resulting from fragmentation) without the necessity to establish a future maintanance schedule. We do talk about this process a bit more in a blog post here: http://www.diskeeper.com/blog/post/2009/11/20/Inside-IntelliWrite-technology.aspx. IntelliWrite is a file system driver which assists the file write process by shoring up the (somewhat feeble) OS logic revolving around file placement.
With regards to your query about SMART: This is factually an industry standard for drive failure detection/alerting-- a consolidation and display of data provided by the drives. There are a variety of free tools that perform SMART monitoring. Diskeeper's 'Disk Alert' (circa 2003) did this as well. I suppose as somewhat of a spoiler, this will be reintroduced in the next version of Diskeeper as an added benefit for workstations, however it will not be present on the Server edition. The primary reasoning is that, as SMART is a drive-specific solution, individual RAID controllers may or may not pass the information along from the drive to the OS.
I encourage you to really investigate the product comparison, not just reading this. As a historical standard that we have no intention of departing from, we don’t get into publicized product comparisons (whether that’s through white paper dissemination or other avenues of “marketing”) as a matter of company competitive ethics/morals that we abide by.
June 17, 2011 at 10:35 am
damiang (6/17/2011)
I'm the Product Manager for Diskeeper Corporation, & want to speak to some of the questions posed directly.
That is a very well written post. Thank you for putting up those white papers and answering the questions in the way that you did.
Viewing 15 posts - 1 through 15 (of 29 total)
You must be logged in to reply to this topic. Login to reply